Transcription
Welcome back, everyone. How do you solve difficult problems? One way to solve them is to use first principles thinking, also called reasoning from first principles.
First principles thinking originates from the domains of philosophy and physics, which are both dealing with difficult problems and are looking for the fundamental "truth". In this video of first principles thinking, we will look at two remarkable practitioners. The first is Richard Feynman, who was a brilliant physicist and educator. The second is Elon Musk, who is the mind behind ventures like SpaceX, Tesla, and Neuralink.
First principles thinking is breaking down a complex problem into its fundamental truths, of course, as far as we know the fundamental truths. Or let's call it highly probable assumptions, what we call here in general the first principles. And then, based on the first principles, we can reconstruct the solution with a more fundamental understanding of the given problem. And part of the process of deconstructing and also reconstructing is that we often have to challenge our current assumptions about a given problem.
In other words, consider here the definition by Elon Musk: "First principles is kinda a physics way of looking at the world. And what that really means is you kinda boil things down to the most fundamental truths and say 'ok, what are we sure is true, or sure as possible is true:' And then reason up from there."
There are some ways to approach first principles thinking. Here we will look at one intuitive way, which consists of asking the question "why?". Many children are inherently good at it. "Why do I have to eat vegetables?" for example. And then we have basically two options to answer: the first option is that we explain them that vegetables contain micronutrients and vitamins, for example, and they are essential for our health. And the second option is that we can tell them "because I tell you so". The second option is exactly the option which kills first principles thinking in the beginning. Because it is not based on finding the first principles, on finding the truth. It is based on authority. Repeatedly asking the question "why?" can lead us to the most fundamental first principles, which is exactly our goal.
"It is hard to think that way. You can't think that way about everything, it takes a lot of effort. But if you are trying to do something new it is the best way to think."
When you are trying first principles thinking out by yourself, you will notice pretty fast, at least it was for me the case, that it takes a lot of energy and time. More often than not, we will hit the boundaries of our own knowledge. And that can feel pretty uncomfortable, but I would say this is a really good way to identify our gaps in our own knowledge, which then we can fill by additional learning.
Elon Musk recommends using first principles thinking when we try to do something new. In general, when we try to solve complex and difficult problems. On the other side, he says that we can't think in terms of first principles all the time. It is way too demanding and not practical for every little decision or problem we face.
Now, this leads us to the question of how do we face normally, or how do we think normally about those little everyday decisions and problems? Reasoning by analogy means when we have two things, A and B, and they are similar in some way, but not identical, then they are probably similar in another way too.
"So, the normal way we conduct our lives is through reason by analogy. We are doing this because it is like something else that was done, or it is like what other people are doing."
When we face a problem, we either solve it the same way as we always solved it (when we faced a similar or identical problem in the past), or we solve it how other people solve it (when other people face a similar problem). This saves us a lot of time and energy, and I would say this is in many scenarios super useful.
Especially in domains like science, where scientists are faced with complex and new problems, it is often not working to "do like it's done before". Even in our everyday lives, when we face big decisions or big problems, it can prove fatal to reason by analogy. Maybe a problem I have now looks similar to a problem another person had or has, but who tells me that the solution the other person had will also work for my problem? Because the circumstances can be totally different, and the problem itself can have also little nuances which makes it totally different.
To further visualize the distinction between first principles thinking and reasoning by analogy, we take here an example from Tim Urban from the famous blog "Wait But Why". In one of his articles, he made the distinction between the chef and the cook:
"The chef reasons from first principles and for the chef, the first principles are raw edible ingredients. Those are her puzzle pieces, her building blocks, and she works her way upwards from there, using her experience, her instinct, and her taste buds. The cook works off of some version of what's already out there - a recipe of some kind, a meal she tried and liked, a dish she watched someone else make."
Many of us are cooks in nearly all kinds of areas, and it makes sense because reasoning by analogy, as said before, saves time and energy. It makes things easier. For new and difficult problems, however, it can be really beneficial to step up and be a chef and use first principles thinking. The easier part is to understand the concept. The more difficult one is to apply first principles thinking. We have to put a lot of effortful thinking into the process, and more often than not, we will approach the boundaries of our knowledge.
In the first example, Richard Feynman will reason up from the first principles. In the second example, I will try to break down a complex problem into its most basic parts and then I will try to reconstruct them again.
In this example, we will reason up from first principles. And I will go through the process from Richard Feynman, what he described in his book "Six Easy Pieces", but I will let out a lot of details because he explains it much better in his book. And it's really fascinating to see in his book how he reasons up from first principles. And not only will you see first principles thinking in action, but you will also learn some basic parts of physics.
Feynman starts with a basic atomic hypothesis: "all things are made of atoms - little particles that move around in perpetual motion, attracting each other when they are a little distance apart, but repelling upon being squeezed into one another." This symbol of the atom will be now the illustration for our atomic hypothesis. From the atomic hypothesis, he reasons up and explains atoms in motion, that the jiggling of atoms is heat. As an example, he takes water and explains water in its three forms: solid in form of ice, liquid (water), and gas. Then he explains atomic processes in his example, what happens at the surface of water. All the while, everything is based on the atomic hypotheses. From there, he explains chemical reactions as he calls it, "a process in which the rearrangement of the atomic partners occurs". As an example, he uses the process of burning carbon. When carbon atoms interact with the air's oxygen molecules, they can create carbon monoxide or carbon dioxide. Read the book, he explains it much better than I can. But it's so fascinating to see that we can reason up a whole process, a whole world, so to speak, from just first principles.
The problem I want to solve in this example is how to get well-placed by the YouTube algorithm. And I'm aware that many of you don't face a similar problem, and it's maybe not so practical for you, but mainly I want to show here the process of first principles thinking, while the topic itself is secondary. Our problem: effectively placing videos on YouTube. Distributing them to the right people, etc. Responsible for that is the algorithm. Why? The algorithm distributes. As far as I know, many are concerned with the so-called keywords. Another way is to look at metrics that are important for the algorithm. And those are first click rate, which means how many people click on a video (basically if people are interested in the title or the topic and the thumbnail). And the second metric is watch time, how long people watch the video (basically if people are interested in the video itself). About the deep technical function of the algorithm, I have no idea. I probably need to acquire here more information or knowledge in general, and I'm hitting the boundaries of my knowledge.
What I researched and heard from some YouTube employees' interviews is that the algorithm mainly serves the viewer. So it should distribute the most fitting videos to a specific viewer and his or her interest. Instead of the technical aspects, we can break it further down to the human aspects. The one way concerns human nature. Most humans are self-serving; they are interested in what serves themselves. Why are they self-serving? Probably because of survival? My knowledge ends here. Humans like stories and the narrated. They don't like too much the abstract. Why? Probably because coherent information is easier to memorize, and our brain likes to not use much energy. Why? I'm not so sure, my knowledge ends here. What are humans interested in? In intellectual or emotional stimulation. This both we can probably break further down to the neuroscience level, but here also my knowledge ends. Takeaway: the fundamental part of YouTube is human behavior. Not focusing on some algorithm directly, but focusing on serving the viewer best.
Even if I wasn't able to arrive completely at the first principles, I hope I could give you a feeling of how this process looks like. First, I broke my main assumptions apart, and then I tried to reconstruct with a new gained fundamental or more highly probable assumptions to build a solution for this problem. When we now look, for example, I reason by analogy, and I would do what (all) other people are doing. It's mostly that they focus on keywords. But keywords are probably good for the algorithm, but not necessarily for the viewer. And from my thinking, it is now really effective to just focus on the viewer. Because focusing on the viewer will probably also indirectly be good for the algorithm itself, because the algorithm should a little bit imitate the interest from the viewer.
I'm really interested now what you think of all of that. I would be really appreciating that when you give me some comments in the comment section about this problem and let me really know what you think about it.
First principles thinking can be a useful way to think. It is certainly not a way of thinking to use for every problem or decision we face, but it can be really useful when we face difficult or complex problems or new problems. In addition, when we break things down to its most basic parts, often we gain a deeper understanding of something, so we also get in addition some learning.
Thank you so much for taking the time to watch this video! As always, I wish you a fantastic day.