Career, Physics and Goals (was: Artificial General Intelligence Speculations)

I originally had really long answers to these questions, but I tried to trim my answers to their fundamentals.

I originally planned to be a professor, which is the only profession that can use string theory. More recently I’ve decided that I almost certainly don’t want to do this, because I don’t think I could be a professor and also keep my integrity: Professor salaries come from student tuition and government grants, both of which are scam-like.

In approximate order of importance

  1. I am ambitious and I want to do something groundbreaking, but I no longer think it’s possible to do anything truly important in my field.
  2. It’s a lot of work, which conflicts with other interests that I have. Much of it is soul-draining busywork too, and no matter how much work you’re doing, you aren’t doing nearly enough work to get the ~1 desirable tenure-track professorship that opens up in the world every year.
  3. The salary (especially for PhD students) is extremely low.

If I had a clear answer to that question, I would just do it.

One thing to note is that the term “superstring theory” is kind of vague, and it encompasses a very large set of ideas. Deutsch might have a specific thing in mind, but his statement is false if interpreted to apply to all the things that people mean when they talk about string theory.

Here are some problems in theoretical physics that have motivated string theorists:

  • They were looking for a microscopic description of what we now call quarks. It turns out that the theory describing quarks as open strings was false, but it’s how string theory got started.
  • Classically, black holes can’t have microstates, but we know that they nonetheless must have entropy, which is a problem. String theory offers consistent descriptions of black hole -like structures that have a clear notion of microstates (basically you can model a black hole as a string in a highly excited state).
  • Classical gravitational field theories are notoriously badly behaved when you try to apply the methods of QFT to them. One problem is that they blow up at low length-scales. String theory solves the problem of wanting to have a gravitational theory that is well-behaved at low length-scales, and which reproduces a classical or quasi-classical theory at large length-scales.

There are problems with all the above solutions, but at least they are motivated by problems in theoretical physics.

1 Like

Are these thoughts and problems related to CF in some way, or pre-existing?

Writing more quickly/casually/informally/chatty might be better.

I’m going to split these posts into their own topic since they aren’t about AI. Would you prefer I put it in Unbounded or Other?

unbounded

Well, I didn’t have any of these problems before I started learning more philosophy, and the type of philosophy I think is true happens to be related to CF, so my problems are both related to CF and pre-existing.

One part of the thoughts I’m having that is closely related to CF stuff is that I’ve realized (edit: CF has made me realize) the epistemological importance of goals. Basically, my main reason for thinking this:

is that it seems impossible for me to come up with good goals to pursue in my field.

Basically, none of the things being worked on seem to have known current or future practical applications, so there is no objective standard of value. There’s no way to objectively say that one project matters more than another. It’s like playing a video game.

In practice, since people need some standard of value or else they can’t function, I think that the standard of value I was adopting in practice (before I started learning more philosophy) was some mix of the following bad things:

  1. A social, Peter Keating-like standard of value, thinking stuff like, “it’s worthwhile to try to prove this result because others think it would be valuable, and if I can do it I’ll get lots of citations and accolades” or “I need to prove I’m smarter than that other guy.”

  2. A mystic / Platonic standard of value, believing that what I was doing is Important (writ large) because I was accessing these eternal Platonic objects from another realm—things which “practical” people *scoffs* can’t even begin to conceive of.

  3. I had a partially pro-life standard of value, but deluded myself into thinking that the things I was working on were much closer to being applicable in the real world than they actually are.

  4. I became cynical about it all and didn’t work as hard as I otherwise would have worked.

I think a lot of people in string theory or pure math (my field is also closely related to pure math by the way; for a period of time I was strongly considering transitioning into that) are the same, but it’s really hard to know. All I know is that my peers can’t ever give me clear answers for why they are doing what they are doing.

Okay, I’ll try. I tried with this post, but I had to think hard about what I wanted to say.

1 Like

Do you know what my answer to that question is?

1 Like

Having read Dialog: Non-Consumption of Philosophy · Elliot Temple, I think I can make an educated guess that you think doing philosophy full time would be better.

Do you have a disagreement with my reasoning? Regarding the abstract issue. Not regarding what you should do with your life. Never mind that. Switching careers is revolutionary and problematic. Let’s just focus on the conceptual issues for now.

Also, here’s a short viewpoint for discussion which should be easier to engage with than the dialog, but you can also quote from the dialog if you find it useful.


People in all fields make systematic errors due to bad philosophy (which is common/widespread), because they’re all using some philosophy in important roles in their activities. So there’s significant risk of doing things that won’t work or are bad.

(I really mean “all fields”, not all other fields. Philosophers screw up due to bad philosophy!)

Philosophy is currently unsafe to outsource to experts. You can’t just assume the mainstream views are right, build on them, and focus on your own field. Lots of mainstream philosophy is wrong.

Philosophy is the most fundamental field. (Specifically epistemology: stuff about critical thinking, how to create, use, judge and deal with ideas, how to learn, how to resolve disagreements between ideas, etc.) When you make errors and check your premises, you will end up having to deal with philosophy (and other things too, but definitely philosophy).

Philosophy is a pretty dead/empty field. There’s almost no good work being done in it currently. This has some advantages and disadvantages.

3 Likes

I think you found CR a few years ago and CF more recently. When did you find Objectivism and what’s your history with it? Is there any other philosophy that’s important to you?

The abstract point I agree with. Like, a lot of people are wasting their lives doing something that would be a complete waste of time if their philosophical premises are wrong, and they evade thinking about it, implicitly adopting someone else’s philosophical ideas in a second-handed way.

Correct.

I found it in ~December of 2020. My history with it is literally just that I’ve read a bunch of Rand/Peikoff and listened to lectures/podcasts. I have also spent a significant amount of time (waayyyyy too much tbh) debating with a friend of mine, who is somewhat sympathetic to Objectivism but is not an Objectivist. We haven’t been doing things in a systematic way or using anything like paths forward, but I do think it has really helped me understand the ideas better.

Short answer: no.

Long answer:

I used to be interested in mainstream philosophy, and the things that they consider problems (is/ought, hard problem of consciousness, how do we know that we aren’t brains in a vat, what are we doing when we do math, what do Godel’s theorems really mean, many-world semantics and the principal principle, AI alignment, anthropic selection, free will, interpretations of quantum mechanics, weird stuff like Newcomb’s paradox, and more). edit: It was only a passing interest though, I never spent huge amounts of time digging into it.

Most of those things that mainstream philosophy considers problems are things that I think CR / Deutsch’s ideas either solved or exposed as non-problems, but admittedly I haven’t thought about some of them in a long time, so I might still value some of that stuff.

I don’t remember being a particularly huge fan of any individual philosophers until I found Deutsch/Popper, probably because none of them actually seemed to solve any problems. I want to read Aristotle at some point. Also some of the philosophers or quasi-philosophers you list on the FI books list look interesting, like Godwin.

What are some downsides to a philosophy career? In general, not for you, but in the current world situation/context.

Maybe wanting to switch careers because you have some new ideas reminds me of a problem I had long ago. It’s a fairly common experience that lots of people have when playing computer RPG games which have a lot of decision making at the beginning during character creation.

This came up for me most when playing the Exile games, which I later speedran. But originally I wasn’t very good at the games (though I think well above average).

So, when the game starts, you make 6 characters and spend 60 skill points for each character. In Exile 2, you also pick some traits: race, advantages and disadvantages for each character.

This is really complicated and heavily frontloads decision making when you’re new to the game and don’t know what you’re doing. Inevitably, you make some bad decisions.

So you play the game, reach level 8, recognize some bad decisions, and come up with some cool ideas about how to design a better party. So you start over. Exile makes starting over pretty convenient: you don’t have to wait through dialog or cutscenes, and you can move around basically as fast as you can press keys (no waiting 10 seconds for your character to move somewhere at a limited movement speed). So if you know what you’re doing, you can replay parts of the game quite quickly and get back to where you were with your new and improved party.

So, great, party 2 reaches level 12. And now you’ve done a few new dungeons, fought some new monsters, gotten more experience with more spells, seen more of the item drops available in the world, run into some new challenges … and now you have a bunch of great ideas about how to make an even better party. And you think “OK, this time I know what I’m doing. I don’t want to keep starting over, so I’ll just make party3 and then play through the game.”

Party 3 reaches level 10 and you find out that something didn’t work how you expected. You were counting on the game handling something one way, but due to a bug or counter-intuitive design decision, it doesn’t work. You were relying on that too much, so you start over.

Party 4 reaches level 20, and it seems fine, but you realize some ways it could be better. You also estimate that you’re only 25% of the way through the game (it’s a very long game), so you’re pretty tempted to start over again.

I forget the exact details of how it went for me, but I remember starting over repeatedly to make better parties. I forget how much I thought it’d be the last time, but I vaguely remember recognizing and understanding this problem to some extent.

One thing people do next is start over and then burn out from playing the early game again and then quit the game entirely without ever playing the later parts of the game with any party.

There are some things that you could do better: if you took the first party to level 25 and tried out more stuff with them, and then started over, you could have dealt with the learning curve more efficiently. You could get multiple errors out of your way with one experimental party. You could try out a bunch of stuff at once. You could even design your party for that (get some of everything on purpose to try it; do not try to make your party actually good). Basically, you learn a reason to restart, but don’t restart, learn a second reason to restart, still don’t restart, and keep that up for a while, and build up to like 8 reasons for starting and then restart and fix those 8 things all at once. This lowers the number of restarts you end up doing. Sometimes people restart over one little thing, and then as soon as they experience a new part of the game they find one more little thing, restart again, and barely make any progress per restart.

Another thing people do is beat the game (which can be done even with a far from optimal party, by design) while thinking “I will play through a second time with a great party”. But then once they win, they quit. They never try out their ideas about how to make a better party. They think they’ve mastered the game now and they move on, but they only played it with a mediocre party and never tested out their optimizations, many of which are probably wrong.

A lot of the times people start over they think it will be the last time. I was a ignorant newbie/fool when I made my last party, but this time I totally know what I’m doing! They learned a few things and are so impressed by their knowledge, but they don’t realize how much they still don’t know. Some people can start over 10 times in a row, every single time thinking it will be their final restart. There is dishonesty involved.

Anyway, if you try to pick a new career now … you may just run into problems with it in a few years. You know some things that you didn’t when you started down this career path, but there is still a ton you don’t know. You aren’t now in a good position to pick a great career, and foresee how it will go. You still don’t really know. There’s a high risk that if you switch careers you will find something dissatisfactory about it and be tempted to switch again.

Also, physicist is a pretty flexible credential. You can pivot to lots of stuff that uses actual physics or doesn’t. People think you’re smart and capable of math. This is true with just an undergrad or masters degree with no PhD, or with a PhD in the wrong specialty. You don’t have to get things perfect at character creation to have some good options. I don’t know how much added marginal value there is for any physics PhD, or a PhD in the right specialty, for which jobs, nor the added cost of getting a PhD (in string theory or in another specialty) from where you currently are (which I’m guessing is already having a masters degree). I think PhDs matter the most in academia. I think you’re saying a PhD specifically in string theory doesn’t add much value for options outside academia (but i don’t know if that’s compared to a different PhD or to no PhD) – you’re saying string theory is an inflexible speciality that isn’t used in industry.

1 Like

I assume you mean a real philosophy career as opposed to an academic philosophy career. :P

I imagine that there would be HUGE downsides given the current world situation.

  1. Very few people “get it,” and this probably comes with a lot of downsides. It would be really difficult for to explain to one’s friends or family or significant other whatever, and society would basically treat the philosopher like a loser or low social-status person. This shouldn’t matter, and wouldn’t matter to someone like John Galt, but it would be very very hard for a normal person to cope with, especially a person who is (presumably) smart enough to get a high social status job if he wanted one.
  2. It would be really difficult to know for sure if one was actually making progress, and so it would be really easy to second-guess oneself into a crisis. Like, with more traditional careers there are simple objective measurements to know if one is making progress (in a local way at least), like did you prove the theorem, or did how many units did you sell.
  3. Since no one gets it, very few people would be consuming one’s philosophical products, and it would suck doing a bunch of hard work and knowing that one is not having an impact, and that maybe one’s philosophical ideas would even be lost to history.
  4. As a corollary of the above, it would be impossible to make a real living doing philosophy, so one would need another profession, which complicates things.
1 Like

IMO, AI Alignment is a good example of people working on really bad, dangerous, harmful stuff due to mistaken philosophical premises. (Also I don’t think it’s very mainstream, though idk that well what’s mainstream. Also in general your list doesn’t match very well what I think you’ll find in a general philosophy overview type book.)

1 Like

Yeah. So what are some upsides?

Makes sense. A similar idea caused me to say this:

I’m still studying for these comprehensive exams in a week, in part as a precaution, and in part so that I still have the option to do my physics PhD in a different subfield (it would be hard, but I still have a few months left to do that).

There’s a big downside you didn’t list. The failure rate for working on (rational) philosophy is high. You might fail to ever learn it well enough to catch up to what’s already been done by e.g. Popper, and be unable to build on his work and contribute anything. People (as seen in many people’s forum posts) often just kinda get upset and start evading, get stuck, won’t discuss, won’t try activities that might help, etc.

  1. I find this ideal of having all of one’s ideas fully explicit and fully integrated with one’s life to be really inspiring. Like how young Francisco could always answer the question of why he was doing what he was doing, or how Ayn Rand could explain every emotion she felt, and the reason behind every single word in Atlas Shrugged.

  2. Similarly, I’m inspired by the related CF/CR ideas of living a life free of self-coercion, and having ideas which are completely unrefuted.

  3. I bet there are lots of low-hanging fruits given how few rational philosophers there are.