When I was CEOing at Twitch one of the thing I’d do every batch of interns was a very short presentation on the origins of the company and then a Q&A. One of the questions was always, “Where should I work and what job should I get, or should I start a company?”
It’s an interesting question to try to answer for an intern I didn’t really know, because of course the actual answer is dependent on that person and their life. So I had to figure out how to articulate the framework I used.
First there’s money. Obviously you want money. But money is well-known for diminishing returns, after you have enough for rent and food and so on. So you don’t want to optimize for cash, it’s more of a constraint.
Then there’s prestige. How much will getting this job impress people? Prestige is mostly a trap, for the same reason designer clothing brands are bad deals: brand might make some less discerning ppl think better of you, but it won’t actually make you better or better off.
Power is also on offer. Usually called “impact” or “the chance to make a difference” or “mission”. We like power bc it feels good to wield, the pure joy of throwing a stone in a lake and watching the splash. And also bc we want to contribute to the future, we want purpose.
The problem with the opportunities for power on offer at the start of your career is that they are all borrowed power. You are being offered the ability to wield an instrument you don’t control in a direction you don’t choose. And in truth all jobs offer this!
So if they’re pitching you on power it’s usually a trick of some kind. They’re trying to convince you to accept less compensation in other ways by offering a mirage. Don’t be fooled!
The exception here is starting your own thing, or joining a very early stage project. That can offer real power, real impact. But it’s very risky and it won’t *feel* like joining the powerful and strong, it will feel like joining the small and weak (but hopefully nimble).
Another thing you can paid in is work. Fun work, interesting work, challenging work, unusual work. Work is valuable bc you can learn from it, and bc certain kinds of work can be their own reward.
Usually these two factors come together. Work is intrinsically rewarding when you learn and grow from it.
Getting paid in growth and learning is great bc it’s good for you, it makes you more valuable in the future and also improves your character.
Finally there’s opportunities for advancement. A career track. Certain jobs pay off in the right to take another more desirable job later; “partner track” at law firms is notoriously structured this way.
For a certain kind of person who likes competition with their peers more than anything else, who loves grinding to get ahead, I think this can be great. For most ppl I think tracks are also traps.
So I mean, of course it comes down to “know thyself”. But in general it seems clear ppl should satisfice for cash, optimize for learning/growth, and ignore everything else.
If you love knife fight competition, get a tracked job.
If you love economy-sized pain and feel like you have no other option, consider starting something. But don’t say I didn’t warn you about the suck. Here’s my startup advice if you decide to do that:
Epistemic status: wild speculation but also I’m clearly right
There is a single general factor — we could call it maybe somatic integrity — which determines a large fraction of the total variance in attributes between people.
It’s appears to be mostly inherited, bc it appears to be driven by things like low mutation load, lack of environmental insults, healthy womb environment, etc. It’s mostly baked by the time you’re born and can only go down from there.
That’s because somatic integrity is basically successful execution of the healthy human body plan as learned by evolution. When it all goes right, all the hard work pays off and the biological system hums.
All heuristics are the same when they give you the right answer, because the right answer doesn’t change depending on how you arrive at it.
So when you’re picking a heuristic, you’re really choosing what kinds of mistakes you want to make. And that depends on what kind of failures you can tolerate.
Sometimes you want a specific type of mistake: if when you’re wrong you want to be wrong the same way as everyone else, then the heuristic “go with the popular option” will work great.
0) Causal resonance is an important idea that I have quite seen defined anywhere so I’m going to take a whack at it.
1) you can view any system as an agent, or as a channel
2) as an agent, a system can be seen as an OODA loop with a locally accurate free-energy minimizing model of the world driving it (or loss minimizing whatever it’s driven by avoiding surprisal)
I got prescribed a drug you took once a day for ~120 days, preferably at the same hour. So I set an alarm on my phone, carried my pills with me everywhere, and took them. When my MD asked about compliance and I said I hadn’t missed any pills, they expressed significant surprise.
Reflecting on why I had such high consistency was interesting. There’s no way I’d have spontaneously remembered to take the pill at the same time without a system. I’m not particularly disciplined! I’d even say the opposite. I was surprised every evening when my alarm went off.
But I know that about myself and because I was carrying a cell phone I could have an alarm have the discipline for me. And because I love process design making myself a little process was fun and natural.
Catalysis lowers the free energy requirements to access some part of the state space. Constraint does the opposite by increasing instead.
So if you understand the world as a space subject to thermodynamics, catalysis and constraint are the two fundamental actions any agent can take to change the world.
Everything you do either constrains outcomes away from parts of the space, or catalyzes a pathway into a new part of the space. No other moves possible!
The idea of a sufficient statistic is one of the most powerful ideas that I somehow missed in my academic education even though I took a fair amount of statistics.
Like ~everything else in statistics, it was invented by Sir Ronald Fisher but had fallen out of favor because of the rise of descriptive statistics. With inferential statistics coming back in fashion sufficient statistics have come back too.
The formal definition for a sufficient statistic is basically a function over a whole set which suffices to infer any prediction you want of a given parameter. That's really abstract though and I don't find very useful.