Advice for young coders

Sources, University, Programming Languages, AI

After the article "How I got wealthy without working too hard", I received many messages from young people asking about their future careers. I will try to answer some of those questions in this article.

As I have more time now, feel free to write me about anything, either on the Substack Chat, or via Twitter @AmacaCapital.

I probably won't manage to reply to everyone, but I promise I'll never ask for money, I really hate internet gurus. I have enough, and all I really want is thinly sliced artichokes and white wine, which don't come that expensive - unless you are in London.

TLDR

• Seek free advice, possibly from boring people.

• Conformism works.

• If it’s very entertaining, it’s entertainment, not education.

• You don’t need to go to University, but it’s really fun.

• More Algebra, less C++.

• Don’t worry about AI: build it or learn to use it.

• Some things get interesting as soon as you discover you can succeed at them, so try things.

Choose your sources

I've met so many people that cannot distinguish good sources from bad ones, so this is likely the most important piece of advice I’m giving.

As a rule, if someone asks you for money, they probably don't have much money to begin with. So, this is advice number one: seek free advice. Information can be duplicated for free, so there is no need to pay for it. Some people say “you get what you pay for” because they are about to sell you something.

Advice number Two: get advice from boring people. There are no get-rich-quick schemes. If the advice you are getting sounds entertaining, it's probably entertainment. The same goes for education: valuable education is a bit boring, if it’s too entertaining it’s probably not education. For example, University professors are usually good sources: they get paid by the state, so they reply for free, and they are usually pretty boring, so they count as good advice. Boring is the edge.

There are way too many books printed today, and most are bad. If you enjoy cheesy self-help books, read them. Reading is not an exclusive club, there are no gateways, and that is a good thing. But you should always have an inner intuition if what you are reading is education or entertainment.

For example, on the YouTube homepage, Stanford University videos take up the same amount of pixels as terrible clickbait videos.

Same goes for Twitter: the crap tweet that explains how to make money with digital marketing in 7 steps, takes the same amount of pixels as the researcher explaining a scientific paper. Be aware of what you are consuming.

Good sources look a bit boring, they never ask for money, and they come from institutions that have a reputation to defend.

I don’t want to sound too conformist, but conformism works: if you feel like you don’t know much, get your education from traditional institutions: textbooks, public schools and Universities, professors, and teachers. That’s the way. If you feel like you know a lot already, you are 100% wrong, so go back to textbooks and University courses anyhow.

There is some epistemology hidden in there: if something is entertaining, people will do it for free. If something is boring, people are more willing to pay to not do it. So, as a fuzzy guideline, “boring” has a bit more value than “entertaining”.

Should I Worry about AI?

No. AI will probably change everything, but I see it as an opportunity. I see two scenarios:

• A) If AI is a “regular” enabling technology, it will change everything and gradually disrupt various parts of the economy, creating opportunities for technical people that know how to program AI, scale AI, configure AI, use AI.

• B) If AI ends up being an exponential self-reinforcing compounding technology that goes “fooom” and doesn’t even stop by the train station of “Human intelligence” (which I find unlikely), then you don’t have anything to worry about: we’d have bigger problems than job security.

So basically, you can only plan for scenario A, which is also more likely than a Yudkowskian/Bostromian explosion. I’m preparing a long form about this topic, but my conclusion is that you shouldn’t worry about becoming obsolete.

Furthermore, this is one thing that traders know very well: Knowledge has value in some timeframe. In the long run, everything happens, you just don’t know when. There is no economic value in knowing that Google Stock will go down if you don’t know precisely when. The same goes for technical skills: we cannot predict when AI will replace all programming. Everything happens in the long run, but if it happens in 20 years, and you do 20 years of well-paid contracting, it’s irrelevant to you. Furthermore, of all things that we know for sure will happen, it’s really dumb to worry about AI: not because the world never changes, but because it’s not clear that “infinite intelligence” is physically feasible. I’d worry a million times more about Climate change (can we keep living like this?) and Western demographics (can we keep not having babies?).

If anything, AI is the coolest thing ever, and one of the reasons why I started coding young. I remember trying to code an assistant that could talk (Festival TTS anyone?) when I was twelve and being able to hack it in a weekend is mindblowing.

AI theory is very interesting because there isn’t much theory. We have some theory but is not really helpful, to the point Machine Learning has become an experimental Art. I love seeing my mathematician friends lost in trying to grasp it. AI is fundamentally experimental engineering, and that doesn’t appeal to the typical aesthetic of quantitative academics. AI is getting developed more like how we built Aeroplanes, than how we built Physics.

More importantly: Computing and Math are very interesting subjects. As AI unfolds and improves, it will be nice to be part of the field, or even just be able to understand what is going on. That’s another reason to study more theory, versus just learning the practice. Today, you can code a small Transformer network in a few hundred lines of code, but understanding what is going on, is no small feat, although very rewarding. I rewrote my own neural network framework from scratch and the satisfaction that gave me was way bigger than any monetary reward.

Ultimately, pursue programming if you’re fascinated by it. If AI ends up replacing us all, you’ll just be more fascinated than ever.

Should I go to University?

Yes, I recommend it, but you don't need to. In some countries, University is free or really cheap (eg. Europe), and studying is very hard. So, most people are better off taking up a Bachelor’s degree in a technical subject, which typically lasts 3/4 years, rather than taking 15 years to learn it all alone online.

Some of you can probably learn to code on your own in a few months, but you'd miss all the fun: meeting like-minded people, partying, and studying way too many theoretical. Three-to-Four years, I promise, will feel really short in hindsight. And if anything, by the time you are 40, you’d wish you had more abstract subjects (more on that later).

At Uni you'll meet friends, and good professors that guide you through arcane subjects you think you’ll never need, and most importantly: you will study things that you'd never approach alone. You can spend years trying to learn Python from scattered YouTube videos, or go to University and learn it all in a few years, with guidance, while meeting new friends. If you can go, the best spot in the world (maybe, of life) is small University towns: some examples in Europe are Bologna, Wageningen, Heidelberg, Coimbra. Big cities are a bit too sparse.

What if University is not free for you? It depends on how expensive it gets. You don't need to go to a top university, any public college will do. The University ranking is mostly irrelevant. You are going to study exactly the same technical books anyhow. For some Americans is probably cheaper to take an aeroplane and get an Engineering degree at some University in Germany rather than taking a loan in the US, but I digress.

The real reason you’d want to go to Harvard or Stanford is the network and the badge. The actual courses are all on YouTube already (Stanford and MIT do a great job at that). But elite universities are really exclusive clubs whose advantage is mainly the fact that they are exclusive. They also manage to attract the best researchers from all over the world. But if you can’t get into one of those, it doesn’t matter. Just study the basics in a technical field you like, in a minor University, and you’ll be fine. If you then are lucky enough to find inspiring professors, all the better. But those are randomly distributed in academia.

Should I learn Python or Rust?

Probably Python, but it doesn’t matter. Technology moves so fast, it moves faster than you can catch it. Get used to it. By the time you have completed a degree in Computer Science, you will have built abstractions in your mind that are up high enough to rapidly pick up any programming language.

In general, it’s useful to try different technologies. Like different programming paradigms (procedural, functional, logic), but not many of the same kind. For example: if you try one reactive programming framework (state-view binding) like React, any other similar framework has minor differences.

If anything: study more theory. Algebra, Calculus and Algorithmics haven’t changed much in the last decades, and I can’t recall using any of the hip Java toolkits I picked up during my undergraduate.

I remember complaining about my Computer Science course, I thought it was too theoretical. Actually, university in Italy is slightly more theoretical (than in the UK for example), but now I regret not doing more abstract math. If you don’t study hard subjects when you are 20 and have a lot of free time, it’s unlikely you’re going do it later. So study the hardest stuff you can, earlier. Hard Theory also makes us better thinkers, it teaches how to comfortably juggle abstract objects and I’m pretty sure the world’s appetite for abstract objects jugglers will not decrease.

Should I join a Startup?

It ultimately depends too much on the situation to give specific advice. In Startups, you can learn a lot, especially if you join early. My suggestion is to not join too many startups: don’t buy the promises they attract you with. Startups usually pay less for the same job, and you are not always learning more. If you just want to code for money and make as much as possible, just reread How I Got Wealthy.

It makes more sense to fund a startup: if you can start a company with good people and attract some capital, even if the company doesn’t do great, you can later sell it. And the company now has a lot more value than the sum of its parts because it’s a package of great people. But being an entrepreneur is a whole different life, it’s not for everybody.

Should I do a PhD?

Many of my friends did a PhD and I remember very vividly that none of them was very happy with it in the process. I’m not sure what it is, but Doctoral students seem to be perpetually in doubt about everything they are doing. Maybe it’s something to do with the competitiveness or the fact that you have to start working very autonomously.

I you really like the academic environment, writing papers and getting them rejected at random, please go for a PhD. It can be can be great. But know that in the last decade, there has been an explosion of PhD positions, and there won’t be a place for everybody in Academia. So most of those PhDs will end up later in the job market. Also, technology will make teachers less in demand, as we can now spread videos of the best professors teaching subjects for free. But surely there will be an increasing need for researchers and it’s possibly one of the collect things that can be done on this planet.

— -

I’m sure I will come up with more advice, but I have to stop somewhere today. For more remote-work/freelancing advice, read my previous article on the matter if you haven’t already.

Consider subscribing to this mailing list, it makes me happy to see the numbers go up and keeps me going: