On Developing Interests
Another rant on motivation
(Note: I wrote this because of an argument I had with someone over how to manage procrastination, and they gave a very wrong answer about the importance of planning and "finding your obsession". At the same time I wasn't able to convey my understanding of procrastination well, so I got annoyed enough to write this. I recommend reading my much better written essay on motivation first, link below, since this "essay" is just a laundry list of additional practical tips.)
A Rant About Motivation
Part 1 - Grit in the Engine I hate the idea of "grit". In theory, it sounds about right. Kids wrongly believe school success is correlated with intelligence, when really it's correlated with inputs, like doing homework. So we don't want to tell kids that they're intelligent (even if they are), otherwise they get in trouble once intelligence alone is not …
Interests do not arise ex nihilo, as if by magic, they are cultivated:
There did not exist cavemen with an inherent disposition towards programming, if only they were born in the present century
A ten year old who fails to learn calculus is not bad at math. The fact that most people in the Middle Ages could not read does not mean most people cannot learn to read. In the same way price cannot be considered without quantity in economics, the difficulty of learning a task cannot be considered without how well the environment teaches the task
China failed to industrialise during the Great Leap Forward, and failed so badly tens of millions of people died from famine. This does not mean China is inherently incapable of industrial production. The development of an economy and the development of a person's interests are both complicated processes involving multiple sub-processes with complicated dynamics. Repeated failure is expected before success
If a person begins with the impression that interests are inborn, they will find easy and trivial things interesting and difficult but meaningful things boring and wrongly conclude that they are only interested in easy and trivial things
Interests, not obsessions, are worth cultivating. A person can have many interests, but only one obsession. Breadth(-first search) is depth:
Knowledge is infinite; It is not possible to know everything
Useful knowledge is applicable; We should strive to learn useful knowledge
Knowledge which is useful tends to be applicable across domains
There is no guarantee that a problem first discovered in a particular domain can be solved with knowledge within that domain; The longer the problem has existed without being solved, the stronger the evidence that a solution does not exist within that domain
Quoting Feynman: "You have to keep a dozen of your favorite problems constantly present in your mind, although by and large they will lay in a dormant state. Every time you hear or read a new trick or a new result, test it against each of your twelve problems to see whether it helps. Every once in a while there will be a hit, and people will say, 'How did he do it? He must be a genius!'"
Remembering good problems and checking similar-ish solutions intuitively seems more appropriate than fixating on arbitrary solutions and checking arbitrary problems. The former seems to return a good solution for a good problem on a match, while the latter seems to return awkward solutions to questionably important problems
It does not matter which interest a person starts with, as long as they start, since which interest is being focused on will change later based on feedback from the environment. Feedback from the environment limits the risk of developing only shallow interests in trivial things, as long as a person responds to this feedback
Overly-focusing on one domain is akin to digging deeper and deeper into a depleted mine; Too much effort for too little insight
Learning is easier the more you learn, not harder:
Chess grandmasters can remember chess boards more easily than beginners; Learning the Chinese language from reading about history in Chinese is easier if you are already familiar with history
More than simply learning engineering more easily if you already have a background in maths, there is something akin to a network effect e.g. encountering an eigenfunction in engineering reinforces what you already know about eigenvectors in maths
Learning involves a large set of tacit knowledge and skills which are not well-explained by any single source. Learning different things in different contexts makes us more aware of them e.g.:
The learning curve does not gradually increase; There is a sharp spike at the beginning before it drops sharply, then gradually increases. Think of learning a new programming language if you already know how to program: The most annoying part and failure-inducing part is setting up the environment. Mistaking the initial difficulty spike as representative of overall difficulty is demotivating and wrong
Forgetting is expected. The solution is not to attempt to remember anything, by adding everything we have ever learnt to Anki, any more than the solution to website latency is storing all information on cache. Some information is more useful than others. If a particular piece of information seems difficult to remember, it is probably not useful. Just delete it from Anki/your memory
On the other hand, some information, like long maths proofs, are difficult to remember but still require occasional access. You don't need to remember them, but store them on disk/in your notes somewhere so you can access them without wasting time re-computing/re-deriving everything
There is a tradeoff between write/read. Summarising, linking, organising all cost time and psychic energy. It might be easier to find a piece of information in a set of well-organised notes, but it is probably almost as fast to search for a keyword in a pile of poorly-organised notes and much faster to write
A sense of familiarity or unfamiliarity with a task, while logically irrelevant to whether the task is worth doing or capable of being done, has in practice a large psychic effect. Any time a new task has to be done, multiply the expected time cost by 10x from inevitable procrastination. Celebrate even trivial accomplishments effusively if they are new; This matches the actual psychic cost
Memorisation is much easier than we might expect. If a person is fluent in a language, they have already trivially memorised at least ~20,000 words. There are only 195 countries in the world, 118 chemical elements, 46 US presidents etc.
There's more I can add, but I don't want to make this section too long
Interest is a measure of usefulness, as well as feasibility:
It does not matter if a problem can technically be solved, if we personally cannot solve it
Interest-based gradient ascent may lead away from an initial problem, but it gives us more tools to solve it later; Backtracking and side tracking is always allowed
If a piece of news is important, you will hear about it even if you don't obsessively read the newspaper. If a problem is important, it will occasionally re-appear and re-engage your interest
Caveat: There will be things which we are forced to learn regardless of interest; This advice is targeted more towards developing our own interests. But the things we are forced to learn generally come with structures that make it easy to learn them. We might not find geography very interesting in school, but we are given geography teachers, forced schooling hours, peer pressure from classmates etc.
Because repeated failure is expected before success, it is important to keep trying, and even more important to develop systems to make you keep trying:
Believe sincerely that failure is logically inevitable, but that even failure is beneficial, and especially more beneficial than not doing anything (If we're not happy with the status quo in, there is no opportunity cost)
Confidence to fail cannot be purely in the mind, or the mind will not believe it. This is easier if e.g. you have parents who will support you even if you fuck up (probably true if you're middle class), have in-demand skills that you can get a decent low-prestige job very easily (e.g. even if you get fired you can be a software engineer for some non-tech firm), already have good friends so you don't really care if you mess up an interaction with a new group you just met
If a person does not have those things, they should be aware that even trying to try will be harder and less efficient for them, but that they should keep trying anyway. Awareness of a common failure mode of a task reduces the strength of the evidence that therefore the task should not be done, reducing the psychic cost of failing
Momentum is the North Star. It is much easier to be productive when we are already productive, since each success generates positive feedback, which generates more motivation, which generates more productive action. This works in reverse; It is much easier to be unproductive when we are already unproductive, since everything will seem hard and distraction more attractive. Anything that breaks momentum should generally not be done
The motivation generated by doing one task well can easily be applied to another task, so it makes sense to do something even if it is not to the point. It is much easier to steer a bicycle which is going at a good speed than to maintain balance on a bicycle which is barely moving; A standing army that is maintained by doing drills and overseas relief efforts is much more effective in war than an army which is only raised in war
If a person has momentum in a certain direction (like spending hours writing a silly essay about motivation), they should continue in that direction, rather than attempting to steer away to do "something that should be done". Attempting to steer away from something with momentum costs psychic energy, and attempting to steer towards something costs psychic energy. Why incur a high penalty twice if at some later time we'll be motivated to do "the thing that should be done" anyway?
Knowing that "the thing that should be done" will actually be done requires self-awareness of our own historical behaviour (as well as this being actually true historically...), so this will be harder for someone who is younger and does not have as much historical behaviour to reason off of
The fact that this advice is conditional does not make it less valid. I would suspect for most literate adults, this condition is true, but not being taken advantage of. Different rules-of-thumb apply at different stages in a process. This is obviously true but rarely acknowledged. The best description of this pattern I've read is in the essay Gravity Turn
Quote: "Regardless of its power, a rocket launched horizontally will quickly nose-dive into the Atlantic. Similarly, a student can rarely succeed in graduate school by solely imitating the activities of established researchers. The student must engage instead in certain activities, such as studying fundamental background material and actively networking, that are mostly orthogonal to a research professor’s day-to-day. (...) (I)t is an equally enormous error to dip your nose cone towards the horizon too late, and spend too much fuel accelerating vertically. Once you break the atmosphere, all excess vertical velocity is wasted motion. At some point during graduate school, the student must transition away from activities that only grant temporary altitude."
Attempting to force "something that should be done" is counter-productive. The advice for people with insomnia is to not keep lying on their beds and associate lying on their beds with negative feelings. Repeatedly forcing "something that should be done" associates negative feelings with that thing and makes it impossible to return to it later. Whereas I find that if I return to a thing after a while, I remember all the good reasons to do the thing and have forgotten the feelings of not wanting to do the thing, so I'm willing to try again
The Pomodoro technique is silly and counter-productive. Why on earth would anyone want to interrupt their focus? If we're too tired to be productive we'll generally be able to tell. The body is a much better measure of the body than a timer. The only exception is if a person is prone to overly-focusing on trying to solve a problem and making no progress for multiple hours/days, but I haven't quite figured out how to solve that problem yet...
Scope-cutting should be done aggressively and without shame as soon as we start to stall due to high psychic costs. It is much, much more important to start/keep going and there is little risk of being too unambitious since scope tends to naturally creep up
Paying for someone to teach you is one of the most effective things you can do to learn, since a teacher will help you to learn quickly and get lots of positive feedback, avoid psychically expensive guessing games about whether you understand something or not, pace you so you're forced to prioritise learning more important things, unstick you when you get stuck so you don't completely stall out, and provide social proof that something is worth learning
A teacher is especially important in the beginner-immediate stage where the very basics (which is generally trivially easy) have been self-learnt, but there is a large chunk of things to learn that aren't useful until the whole chunk is learnt e.g. when learning a language, most books suitable to a beginner's language level are incredibly boring. Reducing the difficulty of this chunk or moving through this chunk quickly is important for keeping up momentum
Paid teachers provide structure that meeting other people to "learn" will not. In informal groups, people will come and go as when they feel like it so there's no regularity or continuity or implicit social pressure; get bored and distract themselves by talking about random unrelated things; be interested in different parts of the same thing, so we don't get to focus on the specific thing we want to learn etc.
Paying a language teacher on iTalki for $20/hour is probably the cheapest way to create structure for ourselves; At $80/month it's still cheaper than seeing a therapist for an hour, more effective in making us feel good about ourselves, and we even get to learn a new language!
It is inevitable that all momentum will be lost at some point, and that this state will continue for some duration. This will happen over and over, but with decreasing frequency and decreasing duration. It is important to know how to start back up from this state:
First, acknowledge that things which were once trivial will be difficult again. This is fine. It will never be as difficult as the first time
Motivation does not arise, ex-nihilo, as if by magic. If we are more motivated one day than the previous, the difference must come from fluctuations in our mood from unobservable bodily processes and fluctuations in our environment whether "random" or caused by our previous actions. "Self-motivation" is logically incoherent, and "hard work" is just handwaving the management of motivational processes
Because our mood and environment naturally fluctuates, at some point both fluctuations will be in our favour, and we'll be productive again. Even if we're addicted to video games at some point we'll get bored of the particular game we're playing. Try to build momentum in these favourable circumstances
Hobble along as best as possible until circumstances are favourable; Just having a job or being in school will give us the structure to keep us at some baseline level of functionality and allow us to take advantage of good moods; Constantly trying to do small tasks and get small wins also helps
Develop a good Youtube/email newsletter feed. It is much easier to find something interesting and worthwhile to do if we don't have to go out and hunt for it. Be careful of not being temporarily over-ambitious and making your Youtube front page full of hour-long lectures you will never watch and end up wasting time on Reddit instead
Youtube has a great feature of inserting random items in your Watch Later list in your main feed, which is great for prompting to you spend 15 minutes learning about Byzantine emperors or modern road construction techniques. Delete items from your Watch Later list that don't seem interesting anymore, but don't worry too much about never being able to watch all of it. Peoples build houses beside rivers in order to have easy access to water, not in order to oblige themselves to drink all of it
The best part about Youtube is that you're already on Youtube to listen to music and watch silly videos, which makes it easy to "get distracted by" harder stuff (assuming you're in the right mental state). Whereas you would never go to Coursera for fun, and so you never do it
Distractions are not inherently a problem; They are only a problem when we don't have enough momentum that we find whatever we're doing so much more interesting than the distraction. Attempting to block distractions only ever works temporarily unless the real problem is solved; If we're smart enough to block sites from our hosts file, we're smart enough to unblock them
There seems to some notion of symmetry and asymmetry at play. Momentum strategies/feedback loops are inherently asymmetric - it's easier to follow momentum than to go against it. I have this intuition (that I cannot properly elaborate on) that exploiting asymmetries is the only way to get more output than input. Whereas it's as easy to make fake goals as it is to break them
One failure mode is thinking that because we have failed to follow through on previous plans, we therefore have to (somehow) redouble on following through on subsequent plans to make up for lost time, like a gambler who keeps increasing their bets after a loss. This just traps us in a cycle of negative feedback
There's a lot of trial and error involved in learning (including attempting to determine if a task is even possible). If we're just copying what someone else has done, we get to skip all of that, so there's little reason to believe we can't "catch up", and therefore there's no reason to rush
Planning beyond one or two immediate tasks is generally a waste of psychic energy. We tend to plan tasks which are unpleasant, which we probably should not be doing anyway, and just thinking about/planning unpleasant tasks has a psychic cost. Having a plan also means we can fail to follow the plan, which incurs another psychic cost. Plans should be small and extremely flexible, more of a mental dump to clear the head, or else a very broad consideration of long-term goals just make sure we're not going in completely the wrong direction. Not an (inevitably grossly inaccurate) short- to mid-term expectation of what should/will happen
Planning is overrated in general. They work very well in some cases, but those cases don't apply to someone with no momentum, and so incurs cost for no benefit:
Planning works when the optimal strategies for a system are well understood. In developmental economics, low-cost textile manufacturing has been the stepping stone for industrialisation for England, China, Bangladesh etc. across different time periods and cultures, so it makes sense for a generic developing country to do low-cost textile manufacturing. Someone with no momentum has no understanding of which tasks are both feasible and useful stepping stones; Making shit up and trying to stick to it is just a way to get stuck in shit
Planning works when the planner is capable of implementing said plans. Someone with no momentum/motivation is by definition unable to follow through, so why make a plan that is not even going to be followed?
Planning works when there are multiple parties to coordinate, such that the cost incurred by having to wait around for someone else to do something is greater than the cost of planning. But for someone who's trying to motivate themselves, there's no difference whether they learn physics at a certain time spontaneously or at a certain time as planned, or at a different time entirely. Precision is not free, and optimisation often involves discarding qualities we don't care to improve qualities we do care about
If we don't plan, how do we measure results? By chance, I recently saw a title on my Youtube feed that went something like "TO FEEL A SENSE OF PROGRESS, YOU HAVE TO KEEP TRACK OF WHAT YOU'VE ACCOMPLISHED". And, like, that's not true? If someone wants to lose weight, they don't need to measure the amount of weight they lose every day (which is variable and includes a lot of noise like water weight), they just need to observe that they are thinner now than they were three months ago, and that's good enough?
If a process is well thought-through enough that we're confident in it (parachutes have never undergone RCTs!), or efficacious enough that results are eye-bleedingly obvious (like Ozempic for weight loss, or for me personally being able to undertake 4 hours a week of private lessons in Chinese/maths/electrical engineering while spending many hours more a week consolidating those lessons, for almost two years now - I'll add a PDF of one of my notes for proof), there's much less of a need to faff about with ever more precise measures of tiny effects
