I’ve always wanted to learn more about economics.
A metaskill to develop an understanding of where most of our incentives for action flow from as a society, is probably one of the highest ROI possible.
So this week, I tried to understand the basics of economics. Here I summarise what I learnt in my explorations.
Principles, not Unified Theories
The first thing I learnt when learning economics was to not naively construct unified theories. There’s a lot more nuance here, and it is much better to look at it through a lens of practicality.
E.g. Econ 101 concepts like supply and demand can seem as simple, alluring explanations for concepts like the minimum wage, but as this video showed, this often leads to an oversimplified model which doesn’t account for all the necessary interactions (such as elasticity of demand here). This may be compared to a model with high bias in machine learning.
Something this taught me, perhaps even preceding the first lesson is that there is intrinsic linkage of economics to power and politics. Though this is an issue in natural sciences as well, I think it is much less relevant due to the larger agreement on the fundamental principles necessary to obtain a working understanding of the world (but still rears its head in fundamental physics e.g. with the string theory discourse)
I think this fundamentally stems from the easier replicability of experiments in natural sciences, which leads to much faster iteration in the scientific process, whereas with economics its highly contestable whether experiments to test a certain (possibly) falsifiable hypothesis can even be constructed.
So, from the beginning my focus is on searching for fundamental principles that can be applied in a way that an analysis of a situation may be considered economic and to be skeptical of methods that claim to easily spit out conclusive answers.
Political Economy
Naively constructing unified theories doesn’t necessarily have to create an underfit model though. In exploring socialism and political economy, I realised it may also create a model with too high variance which may then perform poorly on test data.
That is my understanding mostly coming from reading about the computation problem in command economies and this video on Socialism, specifically in the parts where he talks about the reformism of Eduardo Bernstein as a response to the failed prediction of the collapse of capitalism.
Anyway, I think the critiques of capitalism and contemporary culture coming from this tradition starting with Marx merit a reading even if you disagree with their economics. At this point in time however, I do not know nearly enough to comment on their economic prescriptions which is something I would like to look into.
A framework and Econophysics
From whatever I researched about economics this week, this video - seemed to be the one I got the most out of.
Correcting for the fact that the creator is extremely wealthy, it provides a basic understanding which you can then build upon. What I like about this video is that the principles of how the current economy works are conveyed, and you could still derive different prescriptions from it based on the politics you look at it with or the nuanced details you disagree on. E.g. the Austrian and Keynesian schools may disagree on whether the cycles should be allowed to go as they are or government intervention is necessary. At the same time, it still gives you a good enough understanding to link this with a historical understanding to start understanding geopolitical interactions or to take other real-world developments and start predicting market moves based on them.
As a fun aside this video about econophysics was an interesting take on transactions as molecular colisions. This sits in well with the idea that spending drives the economy explored in the previous video, and I got curious about how this model may be developed further e.g. the relation to field theory where debt and credit are created as a particle anti-particle pair was very interesting.
The power law distributions for the top percentage of owners vs exponential for most of the rest was also something I had not seen before; not much has changed in real terms for most people.
There’s a lot I still have to learn about, e.g. I want to understand how exactly to look at the concept of value, especially in how much it has changed now which I will probably do next week.
Other Cool Stuff
Started looking at this Stanford course on convolutional Neural Networks for Visual Recognition to use in my research. I think it has been really well explained by the instructors and I’m really grateful for them to have put this world class education out on the internet for free.
Read this paper on evolutionary optimization for model merging as a part of a reading group in my college. Though I didn’t get too much of an in depth understanding as often tends to happen with people explaining papers to you in reading groups where you are not too interested in them (not relevant for my own research, no skin in the game).
Wrote this post on my blog about ‘The Nietzschean Yogi’. I wanted to write in a bit of a different direction than the one I took about frames and epistemology here but that was not what decided to come through me in the moment.