Neel Nanda


How to teach things well


Anchoring the new set of intuitions with a succinct anchor phrase or image that ideally has conceptual hooks into the relevant problem domains so that the concept automatically gets triggered in the situations in which it is useful

Strongly agreed, I've been very pleasantly surprised by how valuable this approach is. I think having a clear label to important intuitions is one of the really valuable things I've gotten from the rationalist community. When writing blog posts, I try fairly hard to give clear labels to the key ideas and to put them in bold.

Creating a toy example of the core concept that has moving parts the student can then move themselves to see how other parts move (conceptually)

I'd be curious to see any examples of this you have in mind? I'm super excited about this as a form of learning, but struggle to imagine a specific example for anything I've tried teaching. This seems better suited to tutoring 1 on 1 than to larger groups/talks, I think?

Live a life you feel excited about
Perhaps a distillation is that you can build intrinsic motivation in three ways: [...]

That's a good distillation of half of the point I'm making. The other half is that, tasks are more or less intrinsically motivating because of traits they have. You can make guesses at these traits based on your understanding of yourself and past data, and then make future tasks be more like the good traits and less like the bad traits.

I think this is worth emphasising because "create your own intrinsic motivation" feels obviously reasonable to me, but doesn't feel very actionable. So I was trying to also give a concrete starting point, with some prompts based on my personal experience (where I'd expect some to generalise, some to not)

This seems inauthentic, or else it just shifts more burden on the individual to make intolerable situations feel tolerable. Not only do you have hoops to jump through, you have to like it.

I think it's a stable equilibrium to feel like I have a burden to make intolerable situations tolerable, and a stable equilibrium to find it fun to make intolerable situations tolerable. The first equilibrium sucks, the second one is great. But they're both stable equilibria that it's hard to escape from. I find I can sometimes pull off jumping to the good one, but not always, and definitely agree that it sucks to end up in the bad one.

I think being in the good equilibrium often happens because I feel a spark of whimsy to make things better and run with it, and the bad equilibrium is when I don't feel that spark of whimsy and it instead comes from a place of obligation. So I think part of the skill is to notice those whims and nurture them. But I don't have a great model here.

Are you sure that the excitement you feel is as strong and lasting as you claim? Are you sure it’s deriving from the activities you’ve listed? Why do you think these activities would help other people?

I'm pretty confident? I find excitement one of the easier emotions to introspect on. And a good chunk of this comes from looking back on my life and thinking about things where I'm really satisfied that they happened, so it feels obviously tied to the activity.

I don't think it's at all obvious that these activities would help other people - they were intended more as prompts and to give the flavour of what I was talking about, and I trust people to see what sounds like them and what doesn't. The important part is noticing the traits that are common in worthwhile activities and the traits that are not. Eg, I'm extraverted and get a lot of joy from meeting cool new people, but that obviously doesn't generalise.

Is it better to take time to optimize for excitement, or just to slog through as efficiently as possible and carve out more time to do thinks that you find exciting without having to artificially generate that emotion?

I feel a bit surprised at your framing of "artificially generate that emotion". If I can successfully generate excitement, then it doesn't matter that it's artificial. If I can't, then it's not artificial, it's just not there.

My logic is that, if I have to do the task anyway and it'll take a fair amount of time, then it's much nicer to feel excited when doing it. And, in practice, tasks I find fun often get done faster because I procrastinate less on them, even if I'm a bit less efficient. If you procrastinate less, that one doesn't obviously generalise though.

I also think that "make the nonsense things you need to do anyway" into something fun is a skill - which is initially high effort, but can become much closer to a reflex (eg, make a checklist of what you need to get done, do it with a friend and joke about it the whole time, etc). And so you're both investing time for short-term happiness and for long-term happiness, making it a much better trade

Are you sure that these practices aren’t just part of your normal workflow? Are you discovering something that is new and helpful to you, or just noticing an experience you’ve been having all along?

I'm a bit confused by this question - it feels like you're pointing to a dichotomy between "did naturally" and "did artificially", while I'm arguing more for "nudge yourself towards the things that you expect to work". These are things I've tried over time, common trends I've noticed, and decided to double down on. Eg, I know from past experience that I find editing and being a massive perfectionist unpleasant, so for my month of daily blogging I had a rule of "publish a first draft ASAP, no editing". This is a rule I might have come up with anyway without explicitly thinking through this process, but these thoughts nudged me towards it. And, in hindsight, it's definitely made the project far better.

Thanks for all the questions, I found it interesting to think through answers!

On Systems - Living a life of zero willpower

Hmm, my intuition leans strongly towards preserving willpower over practicing, but that's mostly an intuition formed from personal experience, rather than based in anything robust.

One of the reasons I find thinking in systems super useful is that my willpower is highly variable with time (as a function of mental health, general stress levels, sleep, health, etc). So if I don't have systems then at those times a lot of things in my life break, and I lack the willpower to fix them. So systems don't matter too much during high-willpower times when I could mostly do the right thing anyway, but are basically a way to smooth out that curve, and make low-willpower times much better. And I would be very surprised if practicing using willpower removed those low-willpower periods.

I imagine the case is less obvious if you don't have periods of relatively low willpower?

On Systems - Living a life of zero willpower

Awesome, really glad you enjoyed it! It sounds like you might find my post on social initiative interesting, where I elaborate a bit on how I think about social systems.

With regards to "optimize willpower" vs "seek obsessions", I think that's a super important question I'm somewhat confused about. I find that for me personally, often things I find genuinely fun, valuable and rewarding still take willpower to start doing, and the mindset I've outlined in this post is really helpful to ensure I actually do those. But I think there's also a skill of noticing in the moment when an obsession comes to mind, and running with it, even if it involves violating some systems. I've tried outlining my thoughts on how to find obsessions, and generally manage intrinsic motivation here.

You might enjoy Lynette Bye's series with various highly-productive EAs on how they manage their productivity systems (IIRC, Owen Cotton-Barratt's stood out as obsession focused, most of them felt systems focused, though there's probably a selection bias towards people with systems being more interested in that kind of interview).

Digital calendar = Google Calendar, and surrounding systems to eg check it every morning and generally having it feel like the default to "always do the thing in my calendar"

On Systems - Living a life of zero willpower

Hmm, that doesn't feel like a significant concern to me. I see the point of a routine as making the small, everyday things go well. Eg, in the moment I want to stay up late reading and mess up my sleep, and a routine ensures I don't. While all of the concerns you've raised feel about big, life-altering things. And it seems like it's both completely fine and easy to break routines to do something big and important, and that it's also entirely possible to do big and important things with a consistent daily routine? Eg, if your routine has 8 hours of "do work" in it, you can still freely choose what work means while sticking to the routine.

I'd expect the actual bottleneck on your ability to do big, spontaneous things to be more various life commitments, like job, family, friendships, housing, finances etc.

Maybe there's a psychological barrier to breaking a routine that matters here? Eg keeping to consistent bedtimes makes me warier of spontaneously staying up till 4am because me and my friends want to do something crazy, and that does seem like a cost? My intuition says that kind of thing is fairly minor though, and can be mostly addressed by being willing to deviate from the routine where appropriate

How to learn from conversations

Glad you enjoyed it!

That's interesting - I'm mostly optimising for efficient and accurate learning, but it does also seem effective for signalling sincerity and building rapport. Turns out smart people seem to enjoy talking to someone who's clearly putting effort into understanding what they have to say!

I'm curious what you mean by jamming style conversations?

Become a person who Actually Does Things

I think the ideal here is to be able to reflexively notice in the moment "this is bad and I should fix it", and then actually doing something about it. But this is really hard to consistently pull off. For me, a major bottleneck is that it takes a lot of attention and willpower to do this reliably in the moment.

I've found that I can get a long way by systematising it - creating a regular time when I dwell on "what opportunities am I currently procrastinating about?" Or "what is a low level inconvenience that I'm not doing anything about?". I find a weekly review is a great time to go through questions like that.

I find this really, really helpful, because it's easy to make something like that a routine, and it takes much less willpower than being agenty in the moment. And it also makes it easier to track things when they happen, because I can notice and make a note, and put in the actual effort to fix it during the weekly review

Become a person who Actually Does Things

Ah, makes sense, thanks!

The intended nuance there was "this is an overly simplified and not-literally-true statement, but which I think can be a useful simplification for noticing a common mistake and overcoming it" (or, frankly, that part happened because this post was an experiment in speed-writing and didn't have much thought put into the exact wording. But that's my back-filled justification for why I like that line!)

Become a person who Actually Does Things

I'm making the empirical claim that people systematically don't take enough opportunities. Essentially that people fall too far on the exploit side of explore/exploit, thanks to a bunch of human biases that lead to paralysis. And empirically I've found that when I started taking opportunities more, some were meh, and others were really valuable. I don't think this claim is obviously true, but empirically it seems true for my experience and my observations of people.

I think there's also an important skill of prioritising and choosing the right opportunities. But people are bad at this, and I think that trying to do this often leads to excessive paralysis. I think you first need to develop the skill of taking opportunities, there will be enough good ones in there for this to feel motivating and sustainable, and then you develop the skill of selecting things and prioritising.

I'm also not arguing that you should take literally every opportunity - just that on the margin people should take opportunities more. I think it's really hard to give advice that leads someone paralysed to take too many opportunities, because their bias goes so far in the other direction. And so getting them to take the marginal opportunity naturally means they select good ones (on average).

I agree this can go wrong! Eg, somebody who signs up to a bunch of extra-curriculars, realises they don't have enough time and burns out. I'm not sure how to give advice that can help people to overcome paralysis and be good at filtering opportunities at the same time.

Become a person who Actually Does Things

Hmm, I'm wondering if the law of equal but opposite advice is applying here?

I completely agree that some people do too many things, and that moderation is important! Sky-diving without a parachute is an example of doing something, and obviously dumb.

I think the important question is, on the margin, are people better off doing things more? And in my personal life, and in the people I see around me, the answer is overwhelmingly yes. I see a lot of people paralysed by perfectionism, indecision, anxiety etc. Who always wait for the perfect opportunity, and never deviate from the path of least resistance. And I think those people have too much moderation and not enough agency, and that a post exhorting them to be more agenty is exactly what they need.

I think there are also people who are great at being agenty and really need to learn moderation. And it's approximately impossible to write a post catered to both at once.

My post is very much aimed at the people I have in mind. And I'm implicitly making the empirical claim that most people, on the margin, would benefit from being more agenty. Which is true in my experience, but I definitely live in a bubble.

I think "inherently false" is an extremely strong assertion against this post, and I'd be interested in hearing more justification for that.

Load More