📱

Get Our Mobile App

Take your business learning on the go!

Download on the App StoreGet it on Google Play

OpenAI is collapsing?

Stacked Podcast48:34

Transcription

Nick, we have a lot to talk about. We've got Elon Musk, we've got Anthropic and world domination, we've got some rivals, we have some very controversial things that have been happening. We're covering it all, we're pulling no punches.

We need to start with the Elon Musk and Sam Altman. Dude, for those people that don't know, the messiest thing, Elon's supposed to be building beautiful things instead he's playing with Judge Judy in the courtroom. What the hell is going on there? Do you want to give us a TLDR and we can just go straight into it?

So, essentially what's happening is Musk is suing OpenAI over their conversion from a non-profit to a for-profit. He's alleging that Altman betrayed OpenAI's founding mission to make AI accessible for everybody. And the courtroom's getting pretty spicy. I mean, I don't know exactly how much of this is being reported on modern sort of mainstream media, but I wanted to show you guys a little back and forth that I saw happen a moment ago.

So, basically Musk was asked a question during examination by Wachtell Lipton, which is Sam Altman's leading Supreme Court clerk. His whole job is to break witnesses. So, he asked Musk a question which he thought stupid. And in Musk's typical style, he retorted, "Well, the classic answer to a yes or no question is not so simple. For example, if you ask the question, will you stop beating your wife?" The entire court >> [laughter] >> laughs. And Musk just continuously pushes back on this guy, "You know, I find it funny you're saying this wasn't an unfair question, since you're only asking unfair questions." So, that's his "I'm doing my best." Musk says, "That's not true." So, there are a lot of spicy things that happened basically during this cross-examination. I think we're up to day three or four, but given the fact that as Jack mentioned offscreen a moment ago, this is basically like drama or reality TV, but for the AI world. I thought it was at least worth mentioning.

Musk is 100% dead set on this basically being like a civilizational issue. At least that's his defense. He says, "I'm only doing what I'm doing. The only reason I'm suing you, the only reason why I'm pushing back against this non-profit to for-profit conversion is because I genuinely believe the fate of civilization is at stake." And as we know, I think, you know, a lot of the stuff that he does is sort of in line with that overarching thesis of like, you know, I care about broader civilization. Disagree with the man on his politics what you will, you know, there are few people building Mars missions and whatnot. And while I think some of the ways that he's doing them are kind of silly and unnecessarily controversial, you know, the guy obviously cares a lot about that. So, that's his main defense. Whereas Altman, I think, is you know, sort of taking like the almost Machiavellian play. So, I thought it'd be interesting to bring the subject to hear your take on it. And yeah, I was wondering if maybe you could steel man or straw man one or another of the sides.

That's Yeah, that's a really interesting point of view. I think Musk is a provocateur. I'd actually I would much prefer Elon to be building beautiful technologies and getting us on the moon, which we'd love to go to the moon, than in the courtroom. I think this whole thing could have been avoided actually if they changed their name to ClosedAI. I think that was part of the deal. Um but I think one one broader point that people aren't touching on here is the fact that actually like it has a significance in terms of like the architecture I say Well, basically like how you architect companies in America. Like can you do a for-profit, do incredible amounts of fundraising, and then say, "Actually, we want to create a profit now." Because there's some intricate loopholes they did. They created a subsidiary that was for-profit, for those of you don't know, of which the um the non-profit is like I think like a majority shareholder, and you've got Microsoft that owns this, and you know, they did some funky fresh stuff with the cap table. For example, they had certain clauses in there like you can get a return, but it's capped at 100x or 1,000x, etc. So, Musk is looking for $130 billion in damages. in damages. At least.

>> I imagine he's also looking for just political clout and sort of like a win. You know, obviously he's vested in Grok now, supposedly, right? And so, I think him taking down OpenAI is just like a financial thing, it's almost like a spiritual thing. Do you know what I mean? It's about like the will of the people supposedly against you know, the nature of closed source AI. Although, obviously if I was Musk and I didn't give a [ __ ] about the will of the people, I'd be making the exact same case regardless. You know, Jack, what I find really funny is back in November 2017, Greg Brockman, who you know, is one of the founding members of OpenAI as well, actually wrote in a private diary, "The true answer is that we want Elon Musk out. If three months later we're doing a B corp, then it was a lie." He's basically admitting that they are lying to Elon Musk essentially. Um their end goal is to turn it into a for-profit, and he even like supposes that at some point in the future, which, you know, this is nine years out, it'll turn into a nasty fight. So, I mean like as these as these revelations come out, it really does feel like the desperate housewives of San Francisco. I'm just glad that it's going to be televised, dude. It's kind of be the Amber Heard Johnny Depp thing all over again. And I think like if you go look at charisma multiple, like you would if you win in the court of public opinion, that's like almost the main thing. So, the fact it's coming out is pretty freaking sick. So, I'm I'm glad that they're doing that.

But what's the end play? Why is he doing it? I To me, it does genuinely look like he wanted it to be open. He wanted an open AI, and he was I mean like as the name suggests, right? Like he he wanted this open AI system and not have it in the hands of people who want to make a profit, which is the whole reason why he started Grok. So, I think of all the philosophical weightings that we have with all the models, Grok has the strongest for truth truth seeking. So, I think it's fascinating. So, the the only concern there, bro, is that I would much rather all the attention being in building the models. Like from my point of view, I'm like, it's funny. I will chuckle a little bit, and we'll have a good time. But dude, just get back to the boardroom and start making these epic models as soon as possible. So, that's my only one concern with OpenAI is how how entrenched are they? Because you know, when you sell and you get to your due diligence, it's like it's another thing you've got to do, right? It's like, where's Sam's head at? Like is he in the product or is he kind of being sidetracked by this huge legal battle? Like we don't want them entangled with Judge Judy for six months, Nick.

Yeah, my third-party take, you know, as somebody that has no visibility into their operations whatsoever, is he was probably a pretty core team a core part of the the founding team and the product team initially, but as things have sort of gone more and more advanced, obviously on the model end, and then also crazier and crazier on the political and then PR end, he's just kind of grown more more detached. I don't think he has anywhere near like the same sort of like level of hands-on, you know, day-to-day visibility into the processes as he used to. That's just like me making, you know, some pretty high-level uh you know, inferences, but yeah, I mean like imagine the amount of time and energy this must take. I mean, dude, if I had, you know, probably like the world's most eccentric billionaire sue my company over what he construes to be like a civilizational affair, you best bet I'm not spending the vast majority of my time on like, you know, improving the hyperparameters of XYZ model. So, maybe he's not even needed anymore. Who knows? Maybe the OpenAI operation is just churning kind of without him at this point. This is funny. I mean, if that happens, what's your play? The only way you can do that is make a ironic move, which would be changing the name to ClosedAI or something. Like that's a Either that or you need to do some huge public move on it, 100%. I mean, you've kind of got to pick your enemies wisely, and he's not a great enemy to have. You know, the richest man on the planet who defies the odds in every business he does. You know, let's not piss this guy off if we can. Um So, I mean, there's a bit of a strategy thing there, which is interesting.

Do Do you think it's closed off with the trial, which is going to be four months? And we'll If there's interesting clips, we'll pull them up. But the trial is in front of a nine-person jury, and it's only going to take about three weeks. Yeah. And what's worth noting is most of them do not have any like direct experience with AI. Like they took a >> That is hilarious. >> They took a panel, and two of them said AI was stupid and it wasn't very productive. Another two of them said something like, "Yeah, I've used AI, but it usually outputs wrong results." They're very like you know, that's kind of what a jury's supposed to be, right? Representative sample. But like they do not know what the hell's going on.

Do Do you think we would be biased if we And like would we make good jury members in this case? Uh that's interesting. Uh I would I would hope that my understanding of this technology would would like unbias me, but there is something I think the term is cohort bias, where when you get into like a new company or new industry or something, you start sympathizing with the average kind of kind of consensus beliefs of the people in said industry. Like you and I consistently get people commenting on our YouTube videos saying like AI is [ __ ] and it's slop, and AI doesn't do anything, and AI is just like a bunch of billionaires playing God. >> [gasps and sighs] >> And you know, I think about like the unbounded upside potential of this technology and its ability to like, you know, give every human being a million times the level of personal, you know, abundance and wealth. And then I'm like, how do you square those two opinions? Clearly it's that person that's uneducated. But But I can also understand, obviously, their point of view lacking that. So, it's tough for me to say, man. I'd I'd like to think I would at least be the counter-balance. I'd be the one out of nine jury members that'd be like, "Yeah, I'm finding AI is pretty economically powerful." Um but I can't say for sure. What about yourself?

I I would [clears throat] like to think so. I like cuz I think it I think we would cuz it's an integrity thing. And I think we could put aside our personal opinions for the greater good. Do Do you see, dude? I I watched a video recently and it's like, if you believe what you're doing for the greater good, you can justify anything. So, it's quite dangerous, actually. So, you have to >> Dangerous. Yeah. You've got to get the right You've got to get the right people in that. But dude, maybe we can get into jury duty later cuz it's pretty crazy. But the one I think call out on this as well is that the the verdict matters less than the discovery. So, I think there's going to be a lot of juicy things that pop out, a lot of email chains and funny things that are going to be good to look at and understand.

Um Nick, should we Should we take this Should we take this and go into the next thing, which is about um Anthropic AI in the workspace? Cuz I actually also want to double click on a controversial topic on Claude CoWork and where that's heading. I have some pretty zesty opinions on this topic now. Uh not as zesty as this I might add. These These aren't sponsored in the podcast, but uh oh no. He's been taken by the LaCroix gang. Oh my goodness. LaCroix. Ridiculous. >> [laughter] >> Dude, I think there's actually an evil cabal of like lizard people that just hand out LaCroix to you once you make it past the a freaking net worth multiple or something. They just show up at your door with the big thing and they say, "Drink this on camera or else." The The rule in the group chat actually. Um so what what basically happened was when you were around LaCroix for more than I don't I think it's like 6 or 7 hours you start buying it basically. So they have cuz in in school HQ they have like every flavor. Nick, you you were the beast and he has like you said he's a LaCroix magnate. LaCroix fan. They're everywhere at the Mr. Beast HQ, yeah. Yeah, it's it's I think it's an alternative to Coke basically. It's like above sparky sparkling water, but below Coke Zero. So it just kind of occupies this interesting niche I didn't know existed. Uh I don't What can I say? It's okay.

Should we do you want to take this on to Anthropic or is LaCroix part of Anthropic's chat? Uh if pamplemousse, too. Jesus Christ. Uh yeah, for sure. I mean like a couple of funny things have happened here. Let me share my screen again. Uh it could have been a tweet, should have been a tweet. We're putting it on it's the Stacked podcast anyway. I was just scrolling through this and I found it interesting enough to petition Jack to chat about on the pod uh this week. Basically, I've noticed that the distribution of updates that Claude is making has fundamentally changed over the course of the last month or so. They've gone from the vast majority of these updates being directly programming related to instead being just knowledge work related. And what I mean by that is Anthropic, I think as a company, has solved programming. I think they believe programming is a solved problem. And so what they're doing now is they're starting to go down the chain and look at things that are less technically complicated as programming and instead just focus on knowledge work in general. So let me give you an example. Every single update for like the last I don't know like maybe a week or so has had something to do with like connectors essentially. And on things that directly improve the average non-technical person's ability to do knowledge work. I mean like Adobe Creative Cloud, Ableton, Splice, Canva, Affinity, SketchUp, Resolume, uh you know, Blender, you know, 3D design and whatnot. Uh you know, they're they're they're focused on Tripadvisor, Booking.com, Instacart, Spotify, Audible, Alltrails, Thumbtack, Intuit, TurboTax. All of these are not like programming use cases, right? To be abundantly clear, what these use cases are, okay, are just like knowledge work. It's just like helping out the average accounting firm. It's helping out like the average 3D design company. They've basically taken their Sauron god eye and they've moved from like focusing on code and now they're like, "Okay, how do we just like automate everything else in existence at least in the knowledge front, too." And I think it's just part of their broader play. Whereas OpenAI and all these other companies started off really broad with Sora, right? With like multi-modality, with like, you know, rolling out their health connectors like a year ago. Anthropic just zoned it like locked in on programming, right? And they like saturated all the benchmarks. They made Mythos. They made like the best programming advancements ever. And now that they're done with that, now they're moving over as opposed to the more multi-tasking approach of a lot of other large model companies. So I thought that would just be interesting to riff on and maybe I hear your take on it.

>> Yeah. >> [clears throat] >> Definitely. So I think like the the non-technical audience are where the money's at. It's all enterprise B2B stuff. Get in the code cuz I think actually Anthropic struck a deal with Copilot actually. It's worth it, really. Yeah, yeah. Yeah, so you think of Co-work, right? Um it's basically for those who don't know, you have these three layers of of Claude, right? Claude chat, which is the ChatGPT alternative. Co-work, which is basically Claude code for normies or non-technical people. And then you have Claude code. Um before we double click on the Co-work thing, I do think that Anthropic's making a play to kind of be the everything AI. It's basically Co-work, I think, is where they see the future going, to be honest with you. Code's going to service the development of the builders, but I think they see this almost like, you know, you may as well if you're doing work, you're using you may as well use Co-work. It's their version of Copilot. That's definitely got to be the play. I mean you imagine you're in IBM, you're in Airbus, all these different companies, they may as well be using these tools, these integrated play. So I think there's going to be a a groundswell of competition for what they're doing. Um cuz you know, Claude code's doing billions in ARR. Claude Co-work probably has a much higher market cap, right? Yeah, that's a good point. Yeah, I mean like I think this is the actual play strategically from from a business perspective as opposed to like launch a giant wall of like a thousand different things surfacing a thousand different needs. They honed right in on programming, crushed it with Claude code, had massive ARR, and then said, "Okay, now that we've saturated this, how can we replicate that same approach just for general knowledge work?" And I'm anecdotally finding Co-work is like, you know, I'm consulting with a lot of businesses now. Left Click does workshops and and and and a lot of training and stuff like that across mid-market and enterprise. And I'm finding there's just like there's there's so much demand for Co-work, Co-work training, out of the box skills, uh you know, Apple applying a Co-work to like very simple deliverables across, you know, like low-paid staff, VAs, and and accounting and and bookkeeping people all the way up to like actual managerial decisions. So I mean I I find it really neat. Um you talk about it being the everything AI. Well, you know it's funny, public consensus is now changing a little bit. So everybody's using Codex for programming, but I think that's just because Anthropic as a company has just sort of said, "Oh, we've already solved this with Mythos. We don't actually need to worry worry about this anymore. We haven't dropped our best model. We're like keeping it in the tank." So now they're focusing on the knowledge work and actually OpenAI's swooping in being like, "Oh yeah, our models are marginally better for programming use cases." And and they're attacking the programming audience. It's almost like they're doing what Anthropic was doing like 3 months ago.

The So [clears throat] that that is a powder keg right there. I The thing is the problem with Co-work is the official I love Co-work. I think it's a really good user-friendly entry point. I think it's the stabilizers on the bike. But honestly, if you can use Co-work, you probably should be in code because you will very quickly find its limitations. The fact that it's in a sandbox, it really struggles with API calls. Like the Claude code's biggest problem is its name. Code is too intimidating. Like if they'd called it Claude Plus, I guarantee you. Yeah, I I guarantee that people would be using it. >> [clears throat] >> For example, I saw a video recently that said like, "You know Claude exists. Code seems too complicated. Let's forget code. Cut them out of code. No one's a coder. Get coding out of here. Claude Co-work, however, interesting." It's honestly just a naming thing. Claude Co-work was only created as a user-friendly positioned version of code. And actually if you're in the interface, it's one button click, but a million more opportunities. So my advice to people is if you're using Co-work, you may as well be in code now. Because it's less limited. You can just do more in it. But the problem is it's it's we're actually running into, Nick, a branding issue. Like about what is a developer. Like for example, do you classify yourself as a developer? Yeah, I think I I classify myself as a developer at this point. I actually classify myself less so as a developer today than I used to because I used to like actual actually did the development myself. But uh I would say so. It's interesting cuz like when I think of developer, I think of four-year computer science degree. I think of knows everything about Python. But it's it's it's not true. I mean a developer by their nature is somebody who develops. Okay. Develops what? Software, apps? Well, I mean you're building full software, you are a developer. Using back end and front end, technically you are. So if you built a website in Claude code, you are a developer. Now there'll be an army of people with four-year computer science degrees in the comments section saying, "I disagree with that." And it's a good conversation to have, but technically it is. So my point would be this. Like I think Co-work is a user-friendly version of it. I think it's basically ChatGPT with extra buttons. They simplify down that user interface, they're all going to use it 100%. Now in terms of Claude's performance and Codex, I agree. I've seen that shift um from Codex. I I mean I was in I was in rooms with people uh and I see it all the time. I'm like, "Who are you What are you using right now?" And most of them said I'm using Codex. I'm like, "Okay, interesting." Um so I'm I'm playing around a lot with Codex as well uh today which and we can get into that if you want to. But what Anthropic is doing, I think, is a little dangerous, to be honest with you, because yes, people Okay, the Codex thing shows that AI users are fickle. The second the Opus 4.7, which by the way I think by all accounts is worse than 4.6, they probably shouldn't have released it uh to be honest with you because the the way that it works architecturally as well, the way you have to use it is so different. It required a completely different use case to it. I think 4.6 was I mean there there are even threads, Nick, on Reddit. Not what I'm saying Reddit is a tier one source, of people saying, "Please go back to 4.6." So, you know, Anthropic is allowed to make mistakes. They're shipping like pirates. It's fine. They're crushing it. Um but it's just so it's just so interesting to be like the second your model doesn't perform, they're gone. They're they're over to a completely different coding system. And people are not shy about doing that.

Yeah, yeah, it's true. I mean like I I do wonder, man. I I just wonder what exactly they're sitting on. I've done, you know, model card breakdowns of of Mythos and whatnot and you know, I've I've talked a fair amount about it. Everything that they're suggesting makes it seem like the best model ever. But if you think about it, this is like an interruption of the natural hierarchy of things. Anthropic's been ahead for a while now, you know, at least half a year if not more. And basically it goes Anthropic launches the best model ever. Uh you know, then ChatGPT launches a slightly, you know, a model that's a little bit better. And then Google will launch something that's a little bit better. And then Anthropic launches one that's even better. And so that is what should have happened right here. That fourth step, right? And it should have just continued all the way on until we reached the best. But because they're they're basically pulling an old school OpenAI and they're saying, "No, we can't release this. This is way too powerful." That like fourth step is sort of like been it's sort of down here now. And so they're losing a lot of that mind share. But I do wonder if at the same time that all this is happening in public and you have a massive loss in like consumer trust and enjoyment of the models, if on the enterprise side, you know, when the with with them rolling out mythos across all of these like large enterprises and forming like these very strong partnerships, keeping in mind that like public consensus may be fickle and it may shift, you know, like a moment's notice to win. But these enterprise deals, these are like pretty big, right? They take months and years. You [clears throat] know, when you move a big enterprise, it's like pushing pushing pushing pushing and then finally it starts moving. If you want to push it in another direction, you got to slow all that momentum down and then do the exact same thing in the other direction. You can't just roll out open AI to, you know, a 50,000 person organization overnight. Like it takes some time, right? So, I wonder if that's actually what their play is. And then they're like, "Hey, we're actually going to like just let mythos do its thing on the enterprise development side and cybersecurity, and then we're going to come into the consumers with co-work, try and establish a really big market share, and then crush it there."

I I think I think they must be separate. I think that the model and I it's my experience when when I've sold an enterprise, it's like what people don't appreciate is it's slow to seal the deal, but it's also extremely sticky because everybody gets a you penetrate several layers of organizations, so you can sit there for longer. So, I think the strategy of how do we build this model is going to be different than the models that they're using. It must be different than mythos. There must be a separate box, so to speak, for mythos. There's a mythos box and there's a 4.7 box. I'm sure it's like loads of like CPUs and stuff. Um But I mean I I've just found my honest experience with it recently is it just makes I've just found its blunder rate is higher. It doesn't mean that everyone is bad, but it's blunder rate per 100 interactions is higher. Like it just says stuff that's not true. I'm like it's just app quantifiably inaccurate. So, it's not that it's the the vibes aren't vibing with 100%, so I've become slightly more skeptical of what it's telling me and its output. So, I'm I'm using other models to review it a lot more now. Uh whereas before my which is interesting because it's somewhat of a stop using it, but my confidence has gone down. So, I I double-check more what it's giving me um because of my experience with it. Which is like kind of undermines the whole purpose, right? I'm actually using 4.6 on fast mode still just because there's no fast mode for 4.7. So, I'm just spinning up these sub agents. I do all my [ __ ] Like maybe I'll give it to like a 4.7 at the end of the day to review. I did, what's really interesting, just put a big query that I've been like continuously re-running over the uh course of the last few months on YouTube growth and strategy and and stuff like that. I did just take the query that I've been feeding into 4.6, then feed it to 4.7 and 5.5, GPT 5.5 at the same time, and just like compared results. And anecdotally, I found the results given to me by 5.5 were like much much more in-depth. Uh but they were like almost like like kind of autistic on the spectrum in a way, which I find very interesting cuz I'm also seeing that consensus in the public discourse. How like, you know, GPT 5 is like a one-shot super extraordinarily like talented, almost like savant level intelligence, but it doesn't know how to present it to you. And like I was reading through this big analysis of YouTube and I was like, "Oh yeah, I kind of get that vibe." Like, you know, you're super honed in on all these like extraordinarily tiny details, but then, you know, on the broader broader spectrum, Opus is still better. But I'll be honest, man, at this point, I feel like it's so subjective. Like where's the where's the quantifiable metrics here? I try running evals on my end, and I think a lot of people do. But aside from like the benchmarks, some days it's good, some days it's less good. To me, it it feels like a human Yeah, you know, it's like, "Oh, the vibes are off." But it's like what exactly does that mean? It's like, "Uh shit." These things are so freaking intelligent at this point, Jack. I don't have any other way to benchmark them myself for my own use cases. It's like what is what is the juice flowing to it? Because it's it is so weird. It's more akin to ordering a dinner at a restaurant than it is like a consistent thick finite thing. Like your experience day-to-day can change. So, I think they've got like the the joke is like the people who like gamble online, like they like they take away your like super like they can randomly remove it. I don't want to like Opus like it's just turned down the juice, knock him down to 50%, knock Nick's connection down to 40%. He's getting he's getting too much done and his things they get back down. So, yeah, mileage mileage varies day-to-day, which is weird.

Yeah, it does. But again, it but these could be just random biases. I mean this could just be I think we're very biased about the outputs that we get. We're probably not accurate at assessing how effective it is. Um and it's like you say, like running an eval, how do you run an eval on a website? How do you run an eval on a dashboard? There's stuff to do cuz it's so subjective. And our prompts are so much more complex nowadays, too, if you think about it. I uh I rolled back the clock a little while back and I saw uh some Google Docs that I put together back in 2021 uh where I was prompting GPT-3 on just some very routine queries. And I was just like, "Holy shit." Like the level of depth here. I I could have it do one thing. I could put in one piece of information. I could give it one, you know, introduction and have it like generate a completion and like 40% of the time. And then I look at some of my prompts nowadays and I'm like I'm feeding in like like 10 to 15 different separate things that I'm asking it to do with like mountains and mountains of context. I mean it's it's not even comparable, right? Like, you know, if I fed that old 2021 prompt into one of the models today, obviously it would saturate every possible benchmark on planet Earth. >> [gasps] >> And so it's like, yeah, I think we've just gone, frankly, I think we've just gone beyond our abilities really measure how good these things really are. And so we're sort of defaulting to biased human takes like vibes. When you Yeah. You go on a date with somebody and you're like, "Wow, I just felt awesome about that." And it was like, "What did you feel awesome about? What was it specifically that they said or did?" It's like, "Are you going to point to a specific joke that they made that made you feel awesome?" It's like, "Yeah, maybe it made you feel awesome for a few minutes, but probably not the whole thing." Are you going to point to their appearance? Well, it's like, "Yeah, I'm sure it's a portion of why you felt awesome, but it's not the whole thing." And so, just like we evaluate, you know, dates and relationships and stuff, I feel like that's how we're evaluating models now, man. It's got to be vibes. It's like design, like you can have good design, bad design, but there are a million segments within that. Uh and it's like where does it fit? Where does it go?

Nick, the other thing we should probably touch on is going to be Deep Seek V4 um the rise of China. Um Chinese AIs. [laughter] Have you Have you played around with V4 Deep Seek, bro? Uh no, I haven't. But can I just say the rise of China I feel is too big of a concept to cover in this podcast. Holy [ __ ] >> [laughter] >> It depends how long this podcast can go for, dude. I mean I've got my coffees lined up, so a generation run. So, so Deep Deep Seek V4 for those who don't know is the latest model. Um its performance is comparable with the best Claude models and its price is significantly cheaper. So, you're looking at in some cases 100x cheaper than comparable models in Claude. They're running a little bit of a promo right now, a little Black Friday sale, if you will, um which expires in 7 or 8 days. But even after that point, the the cost per token is significantly less. So, but the thing is, Nick, we don't want to benchmark a queen, right, Nick? We want something that performs out in the wild. So, uh I put it to the test and uh here's my honest experience and also given the benchmarks is that it's not good at design. So, the the way I label lines I currently see it is this, which is Claude is still the best at UI and design. However, if you actually build designs with Claude and then tag in Deep Seek, it is really good at understanding existing design structures and replicating in that. So, I built this really cool website and I asked to Deep Seek to do certain bits and pieces for it, and it will just use the same things that which is great. Um So, its performance is good and the cost is down. So, I I always say these things, you're trading off value, basically performance, cost, and speed. And if you're happy, say, you know, "Well, I want 95% of the performance for 5% of the cost," that might be a trade that you go for. Um what do you think is interesting? But the only watch out with Deep Seek, obviously, is you cannot put API keys in the chat. That's it. >> Really? And also, if you want No, well, because um they appear in the dark web sometimes. Wow, really? Damn. I think so, dude. Yeah, I was doing some leaks and stuff. So, I mean that's the issue. So, you cannot use it if it's like Well, I wouldn't recommend people use it if it's like company data or private stuff. But if you're vibe coding projects, like it's and you'll and you need to do that cost balance stuff, it's it's out there. Yeah, I mean it's cheap for a reason. Um but that's just that's just one to watch out on it.

Yeah, that's fair. You know, you raise an interesting point there. 95% of the quality for 5% of the cost. I mean if you look at it on a quantifiable basis, you're like, "Well, that sounds awesome." I mean just mathematically, what is that? Like 19 times the leverage per dollar? Yeah, that's [ __ ] yeah. Why the hell not? But what I find interesting is that at the margins, even a single percentage point of quality matters so much more. Basically, there's like an exponential relationship between every ensuing successive percentage of quality. And 95% the gap between 95 and 96% might look like X, but the gap between 96 and 97% might look like 100x. And so, these orders of magnitude change, and then the end result is like a user is if it's at 95%, maybe it's in that weird shitty uncanny valley where it's like kind of good, but it's not good enough for you to really want to buy it. It's almost like imagine if you're a business and you're sending out invoices. And your invoices are 95% correct. That's probably going to have a bigger than 5% impact on your business, right? It's like it only takes one of those invoices being mildly wrong for a client to never want to work with you again, and boom, there goes 50% of your whole revenue. So, that's what I >> your payment details. Yeah. Like give it a you know, the extra cost that I'm spending on using the frontier models. I'm going to steelman the frontier model closed source thing if you want to steelman the other side. But the extra cost that you spent is um is entirely worth it because we're dealing with orders of magnitude here. There's some routine tasks, data collection, research, and gathering of stuff. Obviously, you should hand off just like you hand off to Haiku and Sonnet. Yeah, for the most part, like, man, I'm going to want the best intelligence the world applied to everything.

Okay, I Okay, I I agree with that. I think with an unlimited budget, use the best models for everything. But I I think that the 95% quality hides some of the performance. So, if you think about you think about performance or like IQ for humans, for example, it's assessed on eight different axes. And I think with DeepSeek, for example, there are some axes where it outperforms other models, like mathematical capabilities, reasoning in numbers and science, it does exceptionally well. So, and we have this with every model, right? Gemini is the best at video analysis, multimodalities. ChatGPT is exceptional at reasoning code. You have Claude is fantastic at UI. So, all these models have different strengths. I kind of think the play is to tag in the best model for the job. And if if you're running any kind of Ralph Wigam loop type activity where the models need to do have For those who don't know, very, very strict um instructions. Um that they run on loops. I'd say the price can price can matter for some people for that sort of stuff. So, um and [snorts] don't forget, you can always have more senior, just like with any organization ever, more senior developers that oversee the work of their junior devs. So, we can let DeepSeek run a mock at 5%, 1% of the price, build something over here, and have Anthropic look over and decide what it likes and what it doesn't like, and bring it over. So, your net output is actually higher. Um so, I I think the the 95% disguises some of the quality stuff. Personally, what am I doing? I'm not moving all my stuff over per se, but I do think for large development stuff, I think there for this non-client facing for our clients, I think there is an interesting for those that want the that are really pushed by tokenomics, I think there's an interesting application for DeepSeek.

>> [laughter] >> No, I think you're right. I mean, yeah. One more. Go ahead. No, I was I was going to say, one of the thing people say is, "What about Tiananmen Square neck? What if you want to talk to it about Tiananmen Square?" Well, if you want to talk about Tiananmen Square, just use a different model. >> [laughter] >> I'm not going to China anytime soon. Thanks for that, Jack. Um that's it. Cool. >> [laughter] >> You God damn there goes my trip to Guangzhou. God, man. I'm sorry. I'm sorry, Cass. >> it's all right. I guess we'll just have to hang out somewhere else instead. Um >> [gasps] >> Yeah, man. I mean, the truth is you know, always somewhere in the middle, right? So, I I like your idea of orchestration. I also like that idea of like potentially using, you know, a smarter model as a router and then delegating different types of tasks different types of models. That's really interesting. I mean, like if you think about it, most models are trained architecturally with this mixture of experts approach sort of under the hood already, which is kind of like that to me, honest. Um and then we aggregate and and do consensus across like dozens, if not hundreds, of possible out- outputs now. Very different from like the old school, you know, next token predictors um of like, you know, 2019, 2020. So, maybe there's something to be said. Maybe that's sort of architecture is the best way to do things. Basically, just treat AIs like a a router, and then you benchmark on different types of intelligence, and then you just like always give to the specific model that does that best sort of intelligence.

I I find that, dude. As a because like I I know you did the mixture of experts uh so, basically, when something produces a step change in quality, it very quickly becomes a standard which is across all the models. I will say that I found that, you know, you can tag in Gemini and ChatGPT into what you'd call it, you know, all have a different viewpoint and disagree and argue, you know, debate it. Uh which we spoke about before, didn't we, in the previous podcast episodes. I will say that I've noticed the output when I ask it those sorts of questions is always of a significantly high quality. The problem with having one model debate itself is that its own biases are amplified. So, if you if it has a blind spot or a a particular inclination for different way of thinking, maybe it's a woke model versus a based model, it just gets amplified out massively. So, I think it's good to bring in a different flavor of LeCroix to the table. There's a different perspective in the debate. Or pamplemousse. Pamplemousse. What the I didn't even know pamplemousse. Okay, the flavor of this is pamplemousse. I couldn't even tell I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I "

So, basically Musk was asked a question during examination by Wachtell Lipton, which is Sam Altman's leading Supreme Court clerk. His whole job is to break witnesses. So, he asked Musk a question which he thought stupid. And in Musk's typical style, he retorted, "Well, the classic answer to a yes or no question is not so simple. For example, if you ask the question, will you stop beating your wife?" The entire court >> [laughter] >> laughs. And Musk just continuously pushes back on this guy, "You know, I find it funny you're saying this wasn't an unfair question, since you're only asking unfair questions." So, that's his "I'm doing my best." Musk says, "That's not true." So, there are a lot of spicy things that happened basically during this cross-examination. I think we're up to day three or four, but given the fact that as Jack mentioned offscreen a moment ago, this is basically like drama or reality TV, but for the AI world. I thought it was at least worth mentioning.

Musk is 100% dead set on this basically being like a civilizational issue. At least that's his defense. He says, "I'm only doing what I'm doing. The only reason I'm suing you, the only reason why I'm pushing back against this non-profit to for-profit conversion is because I genuinely believe the fate of civilization is at stake." And as we know, I think, you know, a lot of the stuff that he does is sort of in line with that overarching thesis of like, you know, I care about broader civilization. Disagree with the man on his politics what you will, you know, there are few people building Mars missions and whatnot. And while I think some of the ways that he's doing them are kind of silly and unnecessarily controversial, you know, the guy obviously cares a lot about that. So, that's his main defense. Whereas Altman, I think, is you know, sort of taking like the almost Machiavellian play. So, I thought it'd be interesting to bring the subject to hear your take on it. And yeah, I was wondering if maybe you could steel man or straw man one or another of the sides.

That's Yeah, that's a really interesting point of view. I think Musk is a provocateur. I'd actually I would much prefer Elon to be building beautiful technologies and getting us on the moon, which we'd love to go to the moon, than in the courtroom. I think this whole thing could have been avoided actually if they changed their name to ClosedAI. I think that was part of the deal. Um but I think one one broader point that people aren't touching on here is the fact that actually like it has a significance in terms of like the architecture I say Well, basically like how you architect companies in America. Like can you do a for-profit, do incredible amounts of fundraising, and then say, "Actually, we want to create a profit now." Because there's some intricate loopholes they did. They created a subsidiary that was for-profit, for those of you don't know, of which the um the non-profit is like I think like a majority shareholder, and you've got Microsoft that owns this, and you know, they did some funky fresh stuff with the cap table. For example, they had certain clauses in there like you can get a return, but it's capped at 100x or 1,000x, etc. So, Musk is looking for $130 billion in damages. in damages. At least.

>> I imagine he's also looking for just political clout and sort of like a win. You know, obviously he's vested in Grok now, supposedly, right? And so, I think him taking down OpenAI is just like a financial thing, it's almost like a spiritual thing. Do you know what I mean? It's about like the will of the people supposedly against you know, the nature of closed source AI. Although, obviously if I was Musk and I didn't give a [ __ ] about the will of the people, I'd be making the exact same case regardless. You know, Jack, what I find really funny is back in November 2017, Greg Brockman, who you know, is one of the founding members of OpenAI as well, actually wrote in a private diary, "The true answer is that we want Elon Musk out. If three months later we're doing a B corp, then it was a lie." He's basically admitting that they are lying to Elon Musk essentially. Um their end goal is to turn it into a for-profit, and he even like supposes that at some point in the future, which, you know, this is nine years out, it'll turn into a nasty fight. So, I mean like as these as these revelations come out, it really does feel like the desperate housewives of San Francisco. I'm just glad that it's going to be televised, dude. It's kind of be the Amber Heard Johnny Depp thing all over again. And I think like if you go look at charisma multiple, like you would if you win in the court of public opinion, that's like almost the main thing. So, the fact it's coming out is pretty freaking sick. So, I'm I'm glad that they're doing that.

But what's the end play? Why is he doing it? I To me, it does genuinely look like he wanted it to be open. He wanted an open AI, and he was I mean like as the name suggests, right? Like he he wanted this open AI system and not have it in the hands of people who want to make a profit, which is the whole reason why he started Grok. So, I think of all the philosophical weightings that we have with all the models, Grok has the strongest for truth truth seeking. So, I think it's fascinating. So, the the only concern there, bro, is that I would much rather all the attention being in building the models. Like from my point of view, I'm like, it's funny. I will chuckle a little bit, and we'll have a good time. But dude, just get back to the boardroom and start making these epic models as soon as possible. So, that's my only one concern with OpenAI is how how entrenched are they? Because you know, when you sell and you get to your due diligence, it's like it's another thing you've got to do, right? It's like, where's Sam's head at? Like is he in the product or is he kind of being sidetracked by this huge legal battle? Like we don't want them entangled with Judge Judy for six months, Nick.

Yeah, my third-party take, you know, as somebody that has no visibility into their operations whatsoever, is he was probably a pretty core team a core part of the the founding team and the product team initially, but as things have sort of gone more and more advanced, obviously on the model end, and then also crazier and crazier on the political and then PR end, he's just kind of grown more more detached. I don't think he has anywhere near like the same sort of like level of hands-on, you know, day-to-day visibility into the processes as he used to. That's just like me making, you know, some pretty high-level uh you know, inferences, but yeah, I mean like imagine the amount of time and energy this must take. I mean, dude, if I had, you know, probably like the world's most eccentric billionaire sue my company over what he construes to be like a civilizational affair, you best bet I'm not spending the vast majority of my time on like, you know, improving the hyperparameters of XYZ model. So, maybe he's not even needed anymore. Who knows? Maybe the OpenAI operation is just churning kind of without him at this point. This is funny. I mean, if that happens, what's your play? The only way you can do that is make a ironic move, which would be changing the name to ClosedAI or something. Like that's a Either that or you need to do some huge public move on it, 100%. I mean, you've kind of got to pick your enemies wisely, and he's not a great enemy to have. You know, the richest man on the planet who defies the odds in every business he does. You know, let's not piss this guy off if we can. Um So, I mean, there's a bit of a strategy thing there, which is interesting.

Do Do you think it's closed off with the trial, which is going to be four months? And we'll If there's interesting clips, we'll pull them up. But the trial is in front of a nine-person jury, and it's only going to take about three weeks. Yeah. And what's worth noting is most of them do not have any like direct experience with AI. Like they took a >> That is hilarious. >> They took a panel, and two of them said AI was stupid and it wasn't very productive. Another two of them said something like, "Yeah, I've used AI, but it usually outputs wrong results." They're very like you know, that's kind of what a jury's supposed to be, right? Representative sample. But like they do not know what the hell's going on.

Do Do you think we would be biased if we And like would we make good jury members in this case? Uh that's interesting. Uh I would I would hope that my understanding of this technology would would like unbias me, but there is something I think the term is cohort bias, where when you get into like a new company or new industry or something, you start sympathizing with the average kind of kind of consensus beliefs of the people in said industry. Like you and I consistently get people commenting on our YouTube videos saying like AI is [ __ ] and it's slop, and AI doesn't do anything, and AI is just like a bunch of billionaires playing God. >> [gasps and sighs] >> And you know, I think about like the unbounded upside potential of this technology and its ability to like, you know, give every human being a million times the level of personal, you know, abundance and wealth. And then I'm like, how do you square those two opinions? Clearly it's that person that's uneducated. But But I can also understand, obviously, their point of view lacking that. So, it's tough for me to say, man. I'd I'd like to think I would at least be the counter-balance. I'd be the one out of nine jury members that'd be like, "Yeah, I'm finding AI is pretty economically powerful." Um but I can't say for sure. What about yourself?

I I would [clears throat] like to think so. I like cuz I think it I think we would cuz it's an integrity thing. And I think we could put aside our personal opinions for the greater good. Do Do you see, dude? I I watched a video recently and it's like, if you believe what you're doing for the greater good, you can justify anything. So, it's quite dangerous, actually. So, you have to >> Dangerous. Yeah. You've got to get the right You've got to get the right people in that. But dude, maybe we can get into jury duty later cuz it's pretty crazy. But the one I think call out on this as well is that the the verdict matters less than the discovery. So, I think there's going to be a lot of juicy things that pop out, a lot of email chains and funny things that are going to be good to look at and understand.

Um Nick, should we Should we take this Should we take this and go into the next thing, which is about um Anthropic AI in the workspace? Cuz I actually also want to double click on a controversial topic on Claude CoWork and where that's heading. I have some pretty zesty opinions on this topic now. Uh not as zesty as this I might add. These These aren't sponsored in the podcast, but uh oh no. He's been taken by the LaCroix gang. Oh my goodness. LaCroix. Ridiculous. >> [laughter] >> Dude, I think there's actually an evil cabal of like lizard people that just hand out LaCroix to you once you make it past the a freaking net worth multiple or something. They just show up at your door with the big thing and they say, "Drink this on camera or else." The The rule in the group chat actually. Um so what what basically happened was when you were around LaCroix for more than I don't I think it's like 6 or 7 hours you start buying it basically. So they have cuz in in school HQ they have like every flavor. Nick, you you were the beast and he has like you said he's a LaCroix magnate. LaCroix fan. They're everywhere at the Mr. Beast HQ, yeah. Yeah, it's it's I think it's an alternative to Coke basically. It's like above sparky sparkling water, but below Coke Zero. So it just kind of occupies this interesting niche I didn't know existed. Uh I don't What can I say? It's okay.

Should we do you want to take this on to Anthropic or is LaCroix part of Anthropic's chat? Uh if pamplemousse, too. Jesus Christ. Uh yeah, for sure. I mean like a couple of funny things have happened here. Let me share my screen again. Uh it could have been a tweet, should have been a tweet. We're putting it on it's the Stacked podcast anyway. I was just scrolling through this and I found it interesting enough to petition Jack to chat about on the pod uh this week. Basically, I've noticed that the distribution of updates that Claude is making has fundamentally changed over the course of the last month or so. They've gone from the vast majority of these updates being directly programming related to instead being just knowledge work related. And what I mean by that is Anthropic, I think as a company, has solved programming. I think they believe programming is a solved problem. And so what they're doing now is they're starting to go down the chain and look at things that are less technically complicated as programming and instead just focus on knowledge work in general. So let me give you an example. Every single update for like the last I don't know like maybe a week or so has had something to do with like connectors essentially. And on things that directly improve the average non-technical person's ability to do knowledge work. I mean like Adobe Creative Cloud, Ableton, Splice, Canva, Affinity, SketchUp, Resolume, uh you know, Blender, you know, 3D design and whatnot. Uh you know, they're they're they're focused on Tripadvisor, Booking.com, Instacart, Spotify, Audible, Alltrails, Thumbtack, Intuit, TurboTax. All of these are not like programming use cases, right? To be abundantly clear, what these use cases are, okay, are just like knowledge work. It's just like helping out the average accounting firm. It's helping out like the average 3D design company. They've basically taken their Sauron god eye and they've moved from like focusing on code and now they're like, "Okay, how do we just like automate everything else in existence at least in the knowledge front, too." And I think it's just part of their broader play. Whereas OpenAI and all these other companies started off really broad with Sora, right? With like multi-modality, with like, you know, rolling out their health connectors like a year ago. Anthropic just zoned it like locked in on programming, right? And they like saturated all the benchmarks. They made Mythos. They made like the best programming advancements ever. And now that they're done with that, now they're moving over as opposed to the more multi-tasking approach of a lot of other large model companies. So I thought that would just be interesting to riff on and maybe I hear your take on it.

>> Yeah. >> [clears throat] >> Definitely. So I think like the the non-technical audience are where the money's at. It's all enterprise B2B stuff. Get in the code cuz I think actually Anthropic struck a deal with Copilot actually. It's worth it, really. Yeah, yeah. Yeah, so you think of Co-work, right? Um it's basically for those who don't know, you have these three layers of of Claude, right? Claude chat, which is the ChatGPT alternative. Co-work, which is basically Claude code for normies or non-technical people. And then you have Claude code. Um before we double click on the Co-work thing, I do think that Anthropic's making a play to kind of be the everything AI. It's basically Co-work, I think, is where they see the future going, to be honest with you. Code's going to service the development of the builders, but I think they see this almost like, you know, you may as well if you're doing work, you're using you may as well use Co-work. It's their version of Copilot. That's definitely got to be the play. I mean you imagine you're in IBM, you're in Airbus, all these different companies, they may as well be using these tools, these integrated play. So I think there's going to be a a groundswell of competition for what they're doing. Um cuz you know, Claude code's doing billions in ARR. Claude Co-work probably has a much higher market cap, right? Yeah, that's a good point. Yeah, I mean like I think this is the actual play strategically from from a business perspective as opposed to like launch a giant wall of like a thousand different things surfacing a thousand different needs. They honed right in on programming, crushed it with Claude code, had massive ARR, and then said, "Okay, now that we've saturated this, how can we replicate that same approach just for general knowledge work?" And I'm anecdotally finding Co-work is like, you know, I'm consulting with a lot of businesses now. Left Click does workshops and and and and a lot of training and stuff like that across mid-market and enterprise. And I'm finding there's just like there's there's so much demand for Co-work, Co-work training, out of the box skills, uh you know, Apple applying a Co-work to like very simple deliverables across, you know, like low-paid staff, VAs, and and accounting and and bookkeeping people all the way up to like actual managerial decisions. So I mean I I find it really neat. Um you talk about it being the everything AI. Well, you know it's funny, public consensus is now changing a little bit. So everybody's using Codex for programming, but I think that's just because Anthropic as a company has just sort of said, "Oh, we've already solved this with Mythos. We don't actually need to worry worry about this anymore. We haven't dropped our best model. We're like keeping it in the tank." So now they're focusing on the knowledge work and actually OpenAI's swooping in being like, "Oh yeah, our models are marginally better for programming use cases." And and they're attacking the programming audience. It's almost like they're doing what Anthropic was doing like 3 months ago.

The So [clears throat] that that is a powder keg right there. I The thing is the problem with Co-work is the official I love Co-work. I think it's a really good user-friendly entry point. I think it's the stabilizers on the bike. But honestly, if you can use Co-work, you probably should be in code because you will very quickly find its limitations. The fact that it's in a sandbox, it really struggles with API calls. Like the Claude code's biggest problem is its name. Code is too intimidating. Like if they'd called it Claude Plus, I guarantee you. Yeah, I I guarantee that people would be using it. >> [clears throat] >> For example, I saw a video recently that said like, "You know Claude exists. Code seems too complicated. Let's forget code. Cut them out of code. No one's a coder. Get coding out of here. Claude Co-work, however, interesting." It's honestly just a naming thing. Claude Co-work was only created as a user-friendly positioned version of code. And actually if you're in the interface, it's one button click, but a million more opportunities. So my advice to people is if you're using Co-work, you may as well be in code now. Because it's less limited. You can just do more in it. But the problem is it's it's we're actually running into, Nick, a branding issue. Like about what is a developer. Like for example, do you classify yourself as a developer? Yeah, I think I I classify myself as a developer at this point. I actually classify myself less so as a developer today than I used to because I used to like actual actually did the development myself. But uh I would say so. It's interesting cuz like when I think of developer, I think of four-year computer science degree. I think of knows everything about Python. But it's it's it's not true. I mean a developer by their nature is somebody who develops. Okay. Develops what? Software, apps? Well, I mean you're building full software, you are a developer. Using back end and front end, technically you are. So if you built a website in Claude code, you are a developer. Now there'll be an army of people with four-year computer science degrees in the comments section saying, "I disagree with that." And it's a good conversation to have, but technically it is. So my point would be this. Like I think Co-work is a user-friendly version of it. I think it's basically ChatGPT with extra buttons. They simplify down that user interface, they're all going to use it 100%. Now in terms of Claude's performance and Codex, I agree. I've seen that shift um from Codex. I I mean I was in I was in rooms with people uh and I see it all the time. I'm like, "Who are you What are you using right now?" And most of them said I'm using Codex. I'm like, "Okay, interesting." Um so I'm I'm playing around a lot with Codex as well uh today which and we can get into that if you want to. But what Anthropic is doing, I think, is a little dangerous, to be honest with you, because yes, people Okay, the Codex thing shows that AI users are fickle. The second the Opus 4.7, which by the way I think by all accounts is worse than 4.6, they probably shouldn't have released it uh to be honest with you because the the way that it works architecturally as well, the way you have to use it is so different. It required a completely different use case to it. I think 4.6 was I mean there there are even threads, Nick, on Reddit. Not what I'm saying Reddit is a tier one source, of people saying, "Please go back to 4.6." So, you know, Anthropic is allowed to make mistakes. They're shipping like pirates. It's fine. They're crushing it. Um but it's just so it's just so interesting to be like the second your model doesn't perform, they're gone. They're they're over to a completely different coding system. And people are not shy about doing that.

Yeah, yeah, it's true. I mean like I I do wonder, man. I I just wonder what exactly they're sitting on. I've done, you know, model card breakdowns of of Mythos and whatnot and you know, I've I've talked a fair amount about it. Everything that they're suggesting makes it seem like the best model ever. But if you think about it, this is like an interruption of the natural hierarchy of things. Anthropic's been ahead for a while now, you know, at least half a year if not more. And basically it goes Anthropic launches the best model ever. Uh you know, then ChatGPT launches a slightly, you know, a model that's a little bit better. And then Google will launch something that's a little bit better. And then Anthropic launches one that's even better. And so that is what should have happened right here. That fourth step, right? And it should have just continued all the way on until we reached the best. But because they're they're basically pulling an old school OpenAI and they're saying, "No, we can't release this. This is way too powerful." That like fourth step is sort of like been it's sort of down here now. And so they're losing a lot of that mind share. But I do wonder if at the same time that all this is happening in public and you have a massive loss in like consumer trust and enjoyment of the models, if on the enterprise side, you know, when the with with them rolling out mythos across all of these like large enterprises and forming like these very strong partnerships, keeping in mind that like public consensus may be fickle and it may shift, you know, like a moment's notice to win. But these enterprise deals, these are like pretty big, right? They take months and years. You [clears throat] know, when you move a big enterprise, it's like pushing pushing pushing pushing and then finally it starts moving. If you want to push it in another direction, you got to slow all that momentum down and then do the exact same thing in the other direction. You can't just roll out open AI to, you know, a 50,000 person organization overnight. Like it takes some time, right? So, I wonder if that's actually what their play is. And then they're like, "Hey, we're actually going to like just let mythos do its thing on the enterprise development side and cybersecurity, and then we're going to come into the consumers with co-work, try and establish a really big market share, and then crush it there."

I I think I think they must be separate. I think that the model and I it's my experience when when I've sold an enterprise, it's like what people don't appreciate is it's slow to seal the deal, but it's also extremely sticky because everybody gets a you penetrate several layers of organizations, so you can sit there for longer. So, I think the strategy of how do we build this model is going to be different than the models that they're using. It must be different than mythos. There must be a separate box, so to speak, for mythos. There's a mythos box and there's a 4.7 box. I'm sure it's like loads of like CPUs and stuff. Um But I mean I I've just found my honest experience with it recently is it just makes I've just found its blunder rate is higher. It doesn't mean that everyone is bad, but it's blunder rate per 100 interactions is higher. Like it just says stuff that's not true. I'm like it's just app quantifiably inaccurate. So, it's not that it's the the vibes aren't vibing with 100%, so I've become slightly more skeptical of what it's telling me and its output. So, I'm I'm using other models to review it a lot more now. Uh whereas before my which is interesting because it's somewhat of a stop using it, but my confidence has gone down. So, I I double-check more what it's giving me um because of my experience with it. Which is like kind of undermines the whole purpose, right? I'm actually using 4.6 on fast mode still just because there's no fast mode for 4.7. So, I'm just spinning up these sub agents. I do all my [ __ ] Like maybe I'll give it to like a 4.7 at the end of the day to review. I did, what's really interesting, just put a big query that I've been like continuously re-running over the uh course of the last few months on YouTube growth and strategy and and stuff like that. I did just take the query that I've been feeding into 4.6, then feed it to 4.7 and 5.5, GPT 5.5 at the same time, and just like compared results. And anecdotally, I found the results given to me by 5.5 were like much much more in-depth. Uh but they were like almost like like kind of autistic on the spectrum in a way, which I find very interesting cuz I'm also seeing that consensus in the public discourse. How like, you know, GPT 5 is like a one-shot super extraordinarily like talented, almost like savant level intelligence, but it doesn't know how to present it to you. And like I was reading through this big analysis of YouTube and I was like, "Oh yeah, I kind of get that vibe." Like, you know, you're super honed in on all these like extraordinarily tiny details, but then, you know, on the broader broader spectrum, Opus is still better. But I'll be honest, man, at this point, I feel like it's so subjective. Like where's the where's the quantifiable metrics here? I try running evals on my end, and I think a lot of people do. But aside from like the benchmarks, some days it's good, some days it's less good. To me, it it feels like a human Yeah, you know, it's like, "Oh, the vibes are off." But it's like what exactly does that mean? It's like, "Uh shit." These things are so freaking intelligent at this point, Jack. I don't have any other way to benchmark them myself for my own use cases. It's like what is what is the juice flowing to it? Because it's it is so weird. It's more akin to ordering a dinner at a restaurant than it is like a consistent thick finite thing. Like your experience day-to-day can change. So, I think they've got like the the joke is like the people who like gamble online, like they like they take away your like super like they can randomly remove it. I don't want to like Opus like it's just turned down the juice, knock him down to 50%, knock Nick's connection down to 40%. He's getting he's getting too much done and his things they get back down. So, yeah, mileage mileage varies day-to-day, which is weird.

Yeah, it does. But again, it but these could be just random biases. I mean this could just be I think we're very biased about the outputs that we get. We're probably not accurate at assessing how effective it is. Um and it's like you say, like running an eval, how do you run an eval on a website? How do you run an eval on a dashboard? There's stuff to do cuz it's so subjective. And our prompts are so much more complex nowadays, too, if you think about it. I uh I rolled back the clock a little while back and I saw uh some Google Docs that I put together back in 2021 uh where I was prompting GPT-3 on just some very routine queries. And I was just like, "Holy shit." Like the level of depth here. I I could have it do one thing. I could put in one piece of information. I could give it one, you know, introduction and have it like generate a completion and like 40% of the time. And then I look at some of my prompts nowadays and I'm like I'm feeding in like like 10 to 15 different separate things that I'm asking it to do with like mountains and mountains of context. I mean it's it's not even comparable, right? Like, you know, if I fed that old 2021 prompt into one of the models today, obviously it would saturate every possible benchmark on planet Earth. >> [gasps] >> And so it's like, yeah, I think we've just gone, frankly, I think we've just gone beyond our abilities really measure how good these things really are. And so we're sort of defaulting to biased human takes like vibes. When you Yeah. You go on a date with somebody and you're like, "Wow, I just felt awesome about that." And it was like, "What did you feel awesome about? What was it specifically that they said or did?" It's like, "Are you going to point to a specific joke that they made that made you feel awesome?" It's like, "Yeah, maybe it made you feel awesome for a few minutes, but probably not the whole thing." Are you going to point to their appearance? Well, it's like, "Yeah, I'm sure it's a portion of why you felt awesome, but it's not the whole thing." And so, just like we evaluate, you know, dates and relationships and stuff, I feel like that's how we're evaluating models now, man. It's got to be vibes. It's like design, like you can have good design, bad design, but there are a million segments within that. Uh and it's like where does it fit? Where does it go?

Nick, the other thing we should probably touch on is going to be Deep Seek V4 um the rise of China. Um Chinese AIs. [laughter] Have you Have you played around with V4 Deep Seek, bro? Uh no, I haven't. But can I just say the rise of China I feel is too big of a concept to cover in this podcast. Holy [ __ ] >> [laughter] >> It depends how long this podcast can go for, dude. I mean I've got my coffees lined up, so a generation run. So, so Deep Deep Seek V4 for those who don't know is the latest model. Um its performance is comparable with the best Claude models and its price is significantly cheaper. So, you're looking at in some cases 100x cheaper than comparable models in Claude. They're running a little bit of a promo right now, a little Black Friday sale, if you will, um which expires in 7 or 8 days. But even after that point, the the cost per token is significantly less. So, but the thing is, Nick, we don't want to benchmark a queen, right, Nick? We want something that performs out in the wild. So, uh I put it to the test and uh here's my honest experience and also given the benchmarks is that it's not good at design. So, the the way I label lines I currently see it is this, which is Claude is still the best at UI and design. However, if you actually build designs with Claude and then tag in Deep Seek, it is really good at understanding existing design structures and replicating in that. So, I built this really cool website and I asked to Deep Seek to do certain bits and pieces for it, and it will just use the same things that which is great. Um So, its performance is good and the cost is down. So, I I always say these things, you're trading off value, basically performance, cost, and speed. And if you're happy, say, you know, "Well, I want 95% of the performance for 5% of the cost," that might be a trade that you go for. Um what do you think is interesting? But the only watch out with Deep Seek, obviously, is you cannot put API keys in the chat. That's it. >> Really? And also, if you want No, well, because um they appear in the dark web sometimes. Wow, really? Damn. I think so, dude. Yeah, I was doing some leaks and stuff. So, I mean that's the issue. So, you cannot use it if it's like Well, I wouldn't recommend people use it if it's like company data or private stuff. But if you're vibe coding projects, like it's and you'll and you need to do that cost balance stuff, it's it's out there. Yeah, I mean it's cheap for a reason. Um but that's just that's just one to watch out on it.

Yeah, that's fair. You know, you raise an interesting point there. 95% of the quality for 5% of the cost. I mean if you look at it on a quantifiable basis, you're like, "Well, that sounds awesome." I mean just mathematically, what is that? Like 19 times the leverage per dollar? Yeah, that's [ __ ] yeah. Why the hell not? But what I find interesting is that at the margins, even a single percentage point of quality matters so much more. Basically, there's like an exponential relationship between every ensuing successive percentage of quality. And 95% the gap between 95 and 96% might look like X, but the gap between 96 and 97% might look like 100x. And so, these orders of magnitude change, and then the end result is like a user is if it's at 95%, maybe it's in that weird shitty uncanny valley where it's like kind of good, but it's not good enough for you to really want to buy it. It's almost like imagine if you're a business and you're sending out invoices. And your invoices are 95% correct. That's probably going to have a bigger than 5% impact on your business, right? It's like it only takes one of those invoices being mildly wrong for a client to never want to work with you again, and boom, there goes 50% of your whole revenue. So, that's what I >> your payment details. Yeah. Like give it a you know, the extra cost that I'm spending on using the frontier models. I'm going to steelman the frontier model closed source thing if you want to steelman the other side. But the extra cost that you spent is um is entirely worth it because we're dealing with orders of magnitude here. There's some routine tasks, data collection, research, and gathering of stuff. Obviously, you should hand off just like you hand off to Haiku and Sonnet. Yeah, for the most part, like, man, I'm going to want the best intelligence the world applied to everything.

Okay, I Okay, I I agree with that. I think with an unlimited budget, use the best models for everything. But I I think that the 95% quality hides some of the performance. So, if you think about you think about performance or like IQ for humans, for example, it's assessed on eight different axes. And I think with DeepSeek, for example, there are some axes where it outperforms other models, like mathematical capabilities, reasoning in numbers and science, it does exceptionally well. So, and we have this with every model, right? Gemini is the best at video analysis, multimodalities. ChatGPT is exceptional at reasoning code. You have Claude is fantastic at UI. So, all these models have different strengths. I kind of think the play is to tag in the best model for the job. And if if you're running any kind of Ralph Wigam loop type activity where the models need to do have For those who don't know, very, very strict um instructions. Um that they run on loops. I'd say the price can price can matter for some people for that sort of stuff. So, um and [snorts] don't forget, you can always have more senior, just like with any organization ever, more senior developers that oversee the work of their junior devs. So, we can let DeepSeek run a mock at 5%, 1% of the price, build something over here, and have Anthropic look over and decide what it likes and what it doesn't like, and bring it over. So, your net output is actually higher. Um so, I I think the the 95% disguises some of the quality stuff. Personally, what am I doing? I'm not moving all my stuff over per se, but I do think for large development stuff, I think there for this non-client facing for our clients, I think there is an interesting for those that want the that are really pushed by tokenomics, I think there's an interesting application for DeepSeek.

>> [laughter] >> No, I think you're right. I mean, yeah. One more. Go ahead. No, I was I was going to say, one of the thing people say is, "What about Tiananmen Square neck? What if you want to talk to it about Tiananmen Square?" Well, if you want to talk about Tiananmen Square, just use a different model. >> [laughter] >> I'm not going to China anytime soon. Thanks for that, Jack. Um that's it. Cool. >> [laughter] >> You God damn there goes my trip to Guangzhou. God, man. I'm sorry. I'm sorry, Cass. >> it's all right. I guess we'll just have to hang out somewhere else instead. Um >> [gasps] >> Yeah, man. I mean, the truth is you know, always somewhere in the middle, right? So, I I like your idea of orchestration. I also like that idea of like potentially using, you know, a smarter model as a router and then delegating different types of tasks different types of models. That's really interesting. I mean, like if you think about it, most models are trained architecturally with this mixture of experts approach sort of under the hood already, which is kind of like that to me, honest. Um and then we aggregate and and do consensus across like dozens, if not hundreds, of possible out- outputs now. Very different from like the old school, you know, next token predictors um of like, you know, 2019, 2020. So, maybe there's something to be said. Maybe that's sort of architecture is the best way to do things. Basically, just treat AIs like a a router, and then you benchmark on different types of intelligence, and then you just like always give to the specific model that does that best sort of intelligence.

I I find that, dude. As a because like I I know you did the mixture of experts uh so, basically, when something produces a step change in quality, it very quickly becomes a standard which is across all the models. I will say that I found that, you know, you can tag in Gemini and ChatGPT into what you'd call it, you know, all have a different viewpoint and disagree and argue, you know, debate it. Uh which we spoke about before, didn't we, in the previous podcast episodes. I will say that I've noticed the output when I ask it those sorts of questions is always of a significantly high quality. The problem with having one model debate itself is that its own biases are amplified. So, if you if it has a blind spot or a a particular inclination for different way of thinking, maybe it's a woke model versus a based model, it just gets amplified out massively. So, I think it's good to bring in a different flavor of LeCroix to the table. There's a different perspective in the debate. Or pamplemousse. Pamplemousse. What the I didn't even know pamplemousse. Okay, the flavor of this is pamplemousse. I couldn't even tell I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I

I think people would take 70% of the value, 70% of the performance for 100% local. Well, let me give you some Nickonomics right back. So, eventually the quality of these models gets so good that they can all do economically valuable work. And if you're using Mythos versus GPT-5.7, it doesn't actually have a day-to-day impact on the knowledge work that you yourself are doing, compiling your books or whatnot. And as we know, local models lag behind these super advanced frontier models with way more compute. So, we reach that point, and it's like, well, local models can do everything that a frontier model would have done anyway, but they cost 1/10 the price, and then I'm not leaking my quote-unquote preferences to, you know, like a big big company. So, it's like, why don't I just use the local model, right?

>> [snorts] >> I could see that actually genuinely happening at some point in the next couple years. I used to be very anti that, very bearish on local models, cuz I was like, why would I use something that's less intelligent? But in a future where maybe the biggest companies in the world actually do gatekeep like the most advanced intelligences, and we're all in a permanent underclass anyway, why the hell not use something that doesn't leak any more data than I have to? You know what I mean? Or maybe none of it's valuable. None of it matters anyway, because we're all hacked and screwed to begin with. Like, who knows, right?

Yeah, dude. And it's it's so true. I agree. Like, you need to open a door. I could use a bazooka or a shotgun. They both do the job, but one. It's like, they pass the threshold requirement to get it done. The only drawback though, I I think it. We saw this with Claude Code. It goes from being a very niche esoterical interest to something that appears to mass market. The biggest drawback though, I think I I see this with all the memory system stuff as well, is that it can't be architecturally complex. It's got to be simple. It's got to be cuz I I think we're moving to a world where everything's connected. Like, Claude Code, like for example, we're going to bring in Gmail. We're going to bring in Canva. We're going to bring in these different things. I think when you run a local, even like I've got two laptops. I've got my laptop, and I've got my desktop. I've been destroying my keyboard as we speak here. Like, even that, it's like, I have separate. I mean, I'm connected now, but you have separate things happening on different devices, and it just adds complication to your flow. So, I think there's got to be some simplification movement where you you're basically you're going to trade off the optimal system for an easier system. So, again, 95% of the value, but dude, you're not going to waste 5 hours a week scrambling around trying to figure out where the hell did I put this document.

The good The good thing is there's a tendency to overcomplicate when we do. I need to sometimes. Okay, imagine a future where, I don't know, it's a year from now, and there are all these new model distillation techniques and sparsity techniques that improve the quality at like a lower computer requirement. And somebody puts out this GitHub repo, and it's like, this is your local assistant. All you have to do is double-click this, download it, boom, you now have a local assistant. And it does all the knowledge work at more or less every level of quality, aside from like super intense math olympiad problems, which nobody here is doing anyway. You didn't have to sign up. Everything was stored 100% local. Everything was like 100%, you know, running on your local compute, so it's as fast as like the compute price, whatever the cost of your your CPU or GPU is. I think probably everybody would want to do it, yeah.

Yeah. Why wouldn't you? Why would you Why would you go to like a big model provider, sign up, put your credit card in, pay a $20 a month fee, whatever, even if you're like a little bit slower? And also, why wouldn't you? People are not as rational as we give them credit for. Yeah, you have a little like bunny thing and it's like, you know, bunny GPT or whatever. Everybody's just double-clicking that.

>> Those GPT, buy that immediately. Buy that immediately, yeah, 100%. Well, dude, on that note, should we bring in some of the those GPTs from last week? Should we check out some of the phenomenal comments we got on my last week? Who is suggesting that our listeners are not real people, Jack? What our our volume of engagement is so high that we're getting astroturfed by AI or something? I think we must be. I think we must be.

>> bots. Actually, Jack and I are the only real things on Earth. Listeners, just letting you all know.

>> [laughter] >> Do you can you imagine the dead internet theory exists and it's just us talking into the void right now?

>> Yapping constantly. John says, "Hey, actually, why don't we Okay, I'll start with one then you do the other and we'll just work our way down, okay? Uh John says, "60 minutes into today's episode, I've learned nothing about AI, but I have learned that Nick's pretty good at accents." Thanks, John. That's what we're going for. Wait for the some Tesla comments over here. Uh Niffecimi Frank says, "I'd love to know your content editing stack. What tools do you use?" I can knock this out of the park easy peasy lemon squeezy. Uh we use Riverside. And then we just give the link to Tyler. Tyler does it all from there. Probably the best editing stack in the world.

>> you want to Do you want to say that again for those that missed it? It was quite complicated, mate.

>> [laughter] >> All right, what about David? What's David saying here? Yeah, we did. I think Tyler was rushing to get the last one out, so we didn't get the cake in the job. Um so, should we apologize about that? That was That will be with us for a long time. I don't know if I have you apologizing on behalf of somebody else here. Well, dude, I mean, Tyler we're on to we're on to Tyler V.2 now after that. We said, "Dude, you mess up. You can mess up many things. You mess up the cake, you're out of here." We told him this. And he knows this.

>> [laughter] >> Oliver hears a Mustang GT owner. He can confirm. You owned the Mustang GT, right? When you ran your first business? No, BMW. But I did play around with the Mustang when I was in LA and Vegas. It was very cool. Um it was very very good times. The people want the cake SVG. You know what I say? Let them eat cake SVG. You're on a digital diet right now. No cake SVGs for you, pal.

>> [gasps] >> No fluff. All right, so we're getting a lot of people that are thanking us and being lovely human beings. We're not seeing many questions, so I just want you to know every we we appreciate every comment. We really do. And so we're just going to heart and shout all you guys out. But uh if if somebody has a question here, that would be really cool. This is interesting. Thanks, Thomas, and thank you uh Rylan. Okay, so this one here seems pretty interesting. And I think we could probably have a conversation about it. You want to You want to kick us off?

Listen, look, okay, what I really don't like about all this is that Anthropic only comes clean when Codex releases something big and it's completely on them. It's one of those eye-rolling moments. You put your entire customer base through fire over the last couple months and the only reason you're being honest now is you have no other choice. That's worse than doubling down on the lie. Just slimy business. So, I think what ultimate Just one thing. Has he got a dog's profile photo? No. What is the What is this? Is eye-rolling moment. What's he pulling in there? I don't know. Did you write that [laughter] one yourself at Ultimate Prompts or did somebody write that for you? Oh, actually, that's true. Dude, that is hot That's chat GPT psyops right there. Yeah. What do you think? What's the sentiment of it? Um It's Dude, honestly, I think probably there's too much theory crafting going on about that they do this. Were they I don't think there's that much. I don't think there's that level of strategy going on in the HQ. I really, really doubt. I don't think they're these 600 IQ masterminds that we think they are. I think, honestly, they're just trying to create good models and get them out. Um I do think there's a little bit of a timing play that goes on. Um especially with Methos, which is mystically mythically epic, allegedly. We haven't We haven't confirmed it yet. Also, by the way, I think if it was mythically epic, um would we have had a leak by now? Cuz people can't keep things under wraps. So, I I feel like I would have heard something. I would have would have been a little birds like Varys's little birds that would have come and tweeted and said some different things. Um but something happened with Claude recently, the performance. Cuz, you know, we keep it 100% real on this podcast. You guys know this. Um we keep it real for the 17. And um but I that's inexplicable to me. That doesn't make any sense. So, something odd has happened to affect the performance. Um but I do think if they could get the best models safely, I I think they would have pushed it, personally. Yeah, never attribute to malice what is adequately explained by negligence or ignorance. And I think that's a a very classic example of it at Ultimate Prompts, you know? It's a classic example of it. They They just screwed up, right?

Okay, @blackforlife09 says, "Tyler ain't doing [ __ ] bro." I did Tyler. This is the roast of Tyler, dude. Tyler's never going to miss a cake SVG again. This is an interesting one, man. What do you think about that?

Uh dude, well, this is this podcast, right? Deep Seek is about to help us get better token value. Though, it's China. And we wonder if the US should be sending personal info and ideas. Um well, dude, I just want to I just add on to that. I don't know your thoughts on it, but like you know, even in America, like is where is your data safe? Do you know what I mean? It's like, why are we so concerned about China getting your data? This is America. It's not like American companies are the you know, the the Mount Rushmore of virtue. Like it's been known for them to steal and use your data for bad reasons. I'm not saying they do, but I'm just saying like just because it's China doesn't mean that it's de facto the worst possible place for your data to go. It's probably already there anyway. It's already gone. You've already lost the data again. They already have it. Let's Let's be real. I mean, they spy at about level of government and all level of company.

I think um you raise an interesting point there. I would imagine the reason why, and this is coming from somebody whose family uh you know, left uh communist country in '89, basically right when the wall fell, '89, '90. I think the reason why is because of how fragile an economy and a belief system is when it's fully controlled by one person as opposed to like the supposed American values of democracy and whatnot that are more bottom-up. Where it's like, you know, that our our our country and our beliefs, they're sort of like an amalgam of a a bunch of things. It's almost like bottom-up laissez-faire economics versus top-down kind of controlled communist so socially democratic uh you know, economics. Do you know what I mean? It's like some fellows in a board room sort of make all the decisions of the company, including what to do with personal info and ideas. Whereas in America, it's more like the the economic apparatus at large. And I think one of the core theoretical pieces there, and I'm veering away into an area where I'm not an expert, so some people probably call me out on this, but I think the core of thesis there is like the intelligence of a system that is like composed of a bunch of nodes all independently operating in their own best interests is going to be so much larger than, you know, a bunch of folks in a board room somewhere trying to make limited decisions off their human minds. So, when it comes to our personal info and ideas, I feel safer with the first. You know what I mean?

Well, I said I think the the other thing to bear in mind is like Anthropic, for example, have had clear examples where they could have put profit over ethics and they didn't do it. And that means it's more likely to be true that your data is safer. So, that's generally true.

>> [snorts] >> But if I think business or government, they're all operating in their best interests. Uh it is not in their best interest to manipulate and use your data to oust you. Anyway, so even if they did have the data, which to be fair, if you've had TikTok, let's be just super honest for a second, they probably already got all your stuff anyway. They just probably do. You're hitting me with some interesting angles. That's for sure. They they they sure have, man. So, um it's just some good food for thought. Um just some really good food for thought on that. We'll We'll do We'll We'll continue to debate the dude, but if the the whale keeps getting bigger, the whale keeps getting stronger, and the tokenomics just make more sense, um I think we'll hit a pressure point where more people move over to Deep Seek at some point.

>> I think as long as the Jackonomics continue trending. Dude, I actually have a Deep Seek video coming out in 60 minutes, ironically, which is hilarious. 60 minutes? [laughter] What do you Sorry, Jack. You've been You've been sitting on this for 60 minutes? You don't just give the people what they want immediately. Is that outrageous? Is that outrageous? Dude, I know. You've got to get out there, man, 100%.

All right, I think we've uh done done some due diligence here. Guys, I had I had a blast on today's pod. That was really fun. Thanks, Jack. It was fun. It was beautiful. Um guys, we'll catch you in the next one and we will bring digital cake. Have a blessed week and we'll catch you in the next one. Take care, everybody.