Learning Many Small Skills Instead of Getting Stuck

I see people get stuck while trying to learn philosophy. I'll tell you what they don't do: incrementally learn a dozen different small skills, of multiple types, successfully, with some skills building on previous skills, and practice each skill.

You might think "Of course they don't. That sounds hard. But they're stuck." But what I see people do instead is harder.

A dozen sounds like a big number to people, but if you're doing learning right, you should eventually be learning thousands of things. You can do a dozen in one good day. Getting stuck at less than a dozen is really low – it's getting stuck at approximately zero.

You should figure out what you can already do successfully – your baseline – then make incremental progress from there. You should try to learn one thing at a time. People get stuck because they try to do stuff that's too advanced. It's complex, it involves many new parts, and they make many different types of errors, which is hard to deal with, so they get stuck. Incrementally working on one new thing at a time lets you succeed much faster. If there's basically only one new thing to worry about where you might go wrong, that will reduce the errors you make and have to deal with.

Having two things to worry about at once is twice as hard as dealing with one. Having three things is more than three times harder than one. The difficulty scales up exponentially, not linearly, with the number of things you're dealing with at once.

So figure out a small step forward, from your baseline, and repeat. How do you know if a step is "small"? Do a bunch of steps successfully (and some unsuccessfully) and get past experience to help train your intuition. Or consider whether you can do the new step with only a couple errors along the way. Zero errors can be good but might not involve something new. Two errors while completing a step is good. Twenty errors is a lot, so it's not a small step.

Another useful view is to figure out the easiest, fastest way to get successes and do that. Prioritize success. Each success must involve some sort of progress, but it's fine if it's tiny progress. There's no way to cheat by making it too easy or small. Prioritize getting as many successes as you can, as fast as you can, with as little work as you can. That's efficient. Keep in mind that tiny success that take a lot of time and effort aren't efficient (and also involve a higher risk of making mistakes and failing). High effort for tiny progress is only appropriate when you're working on innovative new ideas at the cutting edge of a field – and even then there's usually a better way.

I don't usually see people get a bunch of quick successes and then somehow get stuck. They tend to get stuck by failing at stuff that's hard, long, complicated, high-effort or error-prone.

There is something I see that looks kind of like quick successes then getting stuck. It's initially doing stuff that they basically already know how to do or make almost no errors when doing. Then as soon as they get to the first thing where they make some mistake that requires non-trivial problem solving, they get stuck on it. Then they often switch topics to something else they can do easily, so they can keep up the illusion of progress. What they're doing is quickly "succeeding" at stuff that doesn't require doing effective error correction or problem solving. Also, the initial "successes" don't usually involve practice or objective tests to confirm successful learning, and the knowledge or skills often don't hold up under later scrutiny. They read some essays and watched some videos, thought they learned the material, started using some phrases that sound like they learned it, but they never actually understood it well. And they repeated that for several topics, switching each time there was any difficulty, until that strategy stopped working for them, at which point they were just stuck.

Requirements for Progress

In the first paragraph, I mentioned needing successes of multiple types. If you only ever do grammar, but don't work on any skills related to math, text analysis, debate, logic or philosophy concepts, then you're probably stuck overall. Making progress in grammar is the special exception. What you need is to be able to make progress in most areas. There can be a few special exceptions where you're stuck, and you can figure out work-arounds as long as you're able to make progress on most topics. But if you're stuck on most topics, that needs to be fixed.

I also mentioned needing successes that build on each other. I see people work on sentence grammar but never do paragraph analysis, let alone analyze a whole essay or chapter. They aren't working up to bigger things. Why? Generally it's because they aren't generating a bunch of small successes. They couldn't make a list of their successes (their completed learning mini-projects). They aren't doing one thing at a time. They're trying to learn sentence grammar all at once, as one giant step, so they don't finish.

There are also logical reasons you need to build on past work. To make progress, you need to not only have successes, but have some build on past successes. If you start over from scratch after every small success, and do something independent, then you'll never make more progress than one success. To make big progress and do anything advanced, your only options, logically, are either to have multiple successes work together (instead of being totally separate) or to have a really big success that does all your progress at once.

I also mentioned practice. One way to look at your learning is to list all the different things you've practiced. Try to make that list big by practicing many small steps and finishing each one quickly. It's really important to be able to develop your own practice activities. There aren't pre-made worksheets for every little small step that would be useful. If you stick to practice activities you can get from others today, you'll end up doing a smaller number of larger steps. Also, the quality of available practice materials varies by topic: e.g. it's not very good for math, but much worse for philosophy. So if you're relying on that kind of external help, you probably won't learn much philosophy.

(Note that I'm specifically talking about practice materials. I think many of my essays, like this one, are helpful, educational material. I'm not saying the external help for philosophy is worse than for math overall. But, for example, this essay doesn't come with worksheets with practice problems because I'm prioritizing other tasks like doing more research and writing more essays. Many thousands of math teachers make guided practice resources, but making practice resources is less popular among philosophers. To learn philosophy, you'll need some ability to guide your own practice without someone else designing every step. You do need that ability with math too, but you need more of it for philosophy.)

I also mentioned success. You need lots of successes; trying and failing can happen sometimes but it isn't how you make progress. And you need ways to judge when you're succeeding. If you don't understand what you're doing enough to self-evaluate your work, you can't practice effectively. That basically means you're doing something too hard where you're lost and confused. One of the ways I see people get stuck is they try to do something where they don't know how, and it doesn't work, and then they keep trying for another hour, or even days or weeks. You need to recognize when things aren't working and abort; if you wouldn't be able to recognize that, then it's a bad activity to begin now. And if a task is hard and has a significant risk of failure, you're probably better off breaking it into smaller, easier sub-steps or finding a different approach.

Focusing on quick successes can save people many, many hours used ineffectively on long, slow failures. Occasionally they get long, slow wins but that was still inefficient, high effort, and tends to alienate them from learning: it sucked so much they don't want to do it again. They might be willing to do it again if the payoff were huge – so sometimes they start chasing really big wins, which results in even longer, slower learning processes with higher failure rates.

You should aim for a success rate over 90%. At over 75%, things can still work. People tend to have a lower success rate than what they aim for. At around a 50% success rate, you're in big trouble. (All these numbers are just loose guidelines. They can't be precise because success rate depends on lot on how you count individual steps or tasks. If you count enough tiny things, then a success rate under 99% would actually be bad. Even a failed project could have hundreds of successful sub-sub-sub-sub-sub-sub-steps in it, and fail due to only a few errors. Aiming for over 90% success rate is meant for looking at things more in terms of a mini-project level, not tiny sub-steps.)

I see lots of people slowly putting a lot of effort into stuff they find hard, so they have a 20% or lower success rate on their mini-projects and sometimes basically 0% for big projects. Prioritizing and valuing success more highly would work way better.

You should try to avoid failure streaks. For every consecutive failure, you should do something exponentially easier next to try to break your failure streak. For example, if you have one failure, try to do something with one third of the difficulty next time. If you have two failures in a row, go down to a third again, which is a ninth of the original difficulty. If you have three failures in a row, repeat, so your next goal is a twenty-seventh of the original difficulty.

You can't know exactly how hard things are in advance, or what success rate to expect, but you can estimate it. With some experience, you can form reasonable intuitions that let you effectively use these rough guidelines. (You already have a lot of experience from childhood. A lot of the intuitions about learning that you already formed are reasonable, but there are also some errors mixed in which you'll need to fix.)

Here's a warning: I frequently see people's difficulty reductions after failures be way, way, way too small. That causes long failure streaks (like 10+ failures in a row) and I think this is one of the big things people do wrong. They should try a lot harder to break failure streaks and keep their average success rate above 80%. I don't fully understand why, but many people don't seem to treasure success enough and be proud of it. Instead they seem to take pride in trying to do hard things (and failing) – like they care more about working on something impressive than whether or not they succeed. (I think one cause is doing things where they can't judge success or failure very effectively. They do something so hard that they have very little idea what's going on, so they aren't even sure they're failing, but they do feel sure that they're working on hard, important problems.)

One Shoe Example

You can share your work and ask other people to check for errors. You should do that periodically. But you should understand how to check your own work most of the time. If you don't, then you're probably working on things that aren't one incremental step from your baseline. (Also, for any field that's new to you, it can help to get someone who already knows about it to help with a few error checks at the beginning so you can see some examples of how to do error checking in that field.)

Your baseline is stuff you're done learning, where you can check your own work with confidence and ease. For example, when you walk, you can check for errors like falling over, being slower than you should, going the wrong direction. You know how to judge whether you successfully walked from point A to point B or not. Because you know what you're doing, you can make one change and still mostly know what you're doing and still be able to use good judgment. For example, you could walk with one shoe on and one shoe off, which would be harder and weirder. It'd require more attention and effort than walking normally does, but you'd still be able to judge when you're succeeding or failing. Most of your knowledge of success and error – like what tripping or getting lost is – would still carry over. You might need to learn about a few new types of errors, but that shouldn't be too hard when you have a clear, effective understanding of over 95% of what's going on.

Some things, like touch typing or walking differently, can take many hours of practice to get really comfortable with. That means you should break it into smaller steps. In other words, find more milestones. In other words, find parts you can judge success and failure at individually. Don't just wait until the end of a long project to see if it succeeded or failed. You should have frequent check-in points where you can see if you're making progress. Each of those is, essentially, a sub-project, which enables you to succeed at many smaller projects while working on your bigger goal.

For an example, I'll give a list of some smaller steps, written semi-generically so you could use some similar steps for many other projects. As you read them, think about how you could judge success or failure for that individual step, and how it'd be possible to succeed at each step before doing any of the later steps. You could also think about how you could use some of the same ideas to divide a different learning project into small steps.

The example goal here is being good at walking with only one shoe. This is a goal I never thought about before and which readers won't have past failures for or already be stuck on. Using something new to me let me do real brainstorming, which I'm sharing with only minimal editing.

Try out walking with one shoe and see what it feels like.
Evaluate what problems you're having.
Come up with a plan to deal with each problem.
Stop and reconsider if this skill is really useful and worth the effort to learn.
Do it very slowly and pay close attention to how you're walking.
Record video of you walking with one shoe.
Review the video for walking errors.
Compare what the video shows to your memories of that walk. Try to find issues where you think you're walking one way while you're walking, but you're actually not.
Figure out what you're unsatisfied with.
Come up with some adjustments.
Practice and record more video until you're walking (slowly) in a way that succeeds at your best current understanding of your final goal (you can still revise your final goal some later). (A final goal means an overall goal rather than a sub-goal.)
Practice more to get some repetition and speed up a little.
Practice with one shoe on your other foot.
Have other people watch you walk and look for errors.
Start doing one shoe walking in daily life so you can get a lot more experience with it.
Do practice sessions weekly to make sure you aren't forming any bad habits while walking around in daily life.
Start doing tests to check your comfort and mastery. Record a practice walk with a distraction, such as chewing gum, reading on your phone or juggling. See if your skill holds up to distractions.
You should be getting a feel for when you make mistakes, even when distracted. Predict whether the video recording will show an error and check whether you were right. Keep doing this to get your prediction accuracy above 80%.
Practice walking on more advanced surfaces, like hills, snow, hot coals, tar pits, cracked sidewalks and obstacle courses.
Repeat many prior steps (and invent new ones) for each advanced surface as needed (you might find your skills work on an advanced surface almost immediately, or that more practice is needed almost like relearning everything from the start, or something in between).
Enter amateur one-shoe-walking competitions to see how your skills compare to others.
Consider whether you're satisfied with your skills and whether to set some more goals or be done.

Every single one of those steps could itself be divided into multiple steps. (You could try actually brainstorming and writing down sub-steps for a few of them.)

Also, it shouldn't be very hard to take some of those steps and rewrite them for some other topic while keeping the main idea from this step. I purposefully wrote them semi-generically to make that easier. You could practice translating some of them to other topics. For example, instead of a video of yourself walking, you could take a video of yourself touch typing. Instead of walking on advanced surfaces, you could work on your chemistry glassware cleaning skill by doing advanced cleaning jobs (cleaning glassware that's dirty with something extra dirty, sticky or hazardous). Instead of doing one shoe walking in daily life to get practice, you could use Idea-Goal-Context charts in daily life for making a few decisions each day.

Practicing

Every time you read one of my essays, you could practice at least one thing discussed in the essay or a sub-skill that would help build towards it. If you want to practice something related to an essay, but can't think of anything, you can ask for suggestions on the Critical Fallibilism forum. After you get examples a few times, you'll hopefully get better at figuring stuff out yourself. And sharing what you get stuck on gives me information about what people need help with (I'm then more likely to cover that in essays).

My suggestion above to brainstorm and write down sub-steps for a few of the one-shoe-walking steps would be a way to practice. But I'm not going to create an actual worksheet where I pick several of the steps, put them on the worksheet with blank lines under them to write your brainstorming on, and put instructions saying to brainstorm sub-steps and write them on the blank lines. I have higher priorities to use my time on like essay writing. If the lack of a formal worksheet like that makes a crucial difference that prevents you from practicing, you've got some problems to solve.

You could also practice by brainstorming steps like mine but for another topic besides one-shoe-walking. But I'm not going to make a worksheet with some topics (e.g. debate, typing, chess, cooking, baseball, epistemology, algebra, negotiating car prices) with blank space under them. You can actually pick good topics for you to practice brainstorming about better than I can since I'm writing generically for many readers at once.

You can also email feedback and questions to me, but I don't want to give practice suggestions privately to individual people without being paid. I want many people to be able to benefit from my suggestions. One thing you could do is email a question and also say you give permission to share the question publicly, so then I could answer on my forum or blog. You can also specify what name you'd like to be referred to by, including anonymous (using your real name isn't relevant to my goal of giving answers where other people can read them too). Sometimes I'll write and answer a generic question similar to something someone sends me, but sometimes I do prefer to quote a real person's actual question.

Sub-Goals

People often already do more steps than they realize, many of them successfully, but they don't recognize their successes and give themselves credit. That can contribute to being demoralized. And because they tend not to count easy, quick wins as a step or as progress, they focus on doing hard things and get stuck. What they should do instead is figure out what additional easy things they could do to reach their goal. And if it's a ton of things, create sub-goals. And if it's still a ton of things, take one sub-goal and make sub-sub-goals for it. And if it's still a ton of things, make sub-sub-sub-goals. And so on.

With the one shoe walking example, each thing I listed is a sub-goal. Someone with a lot of successful learning experience might find a list like that good enough because they can break each sub-goal down into smaller parts as they do it without writing those steps down (or if they're having some trouble, they'll recognize that and write more down when they need to). Someone with less of a track record of success should probably write down sub-sub-goals for each part when they get to it. But don't try to write down sub-sub-goals for each part now, all at once, before you get started – that's too much work. And you might get stuck on step 3 and quit the project so you'll never need a breakdown of step 8. Most of the time, it's important to get started pretty quickly, not spend too long on planning. (Multi-million dollar projects at work, or space flights, merit more planning due to higher stakes, more risk, more resources being used, large project size, etc. Inexperienced people should not lead projects like those. They should ideally be led by people who have already successfully planned and completed big projects. People can work up to big projects by e.g. increasing the size of their next project by 10% after each success.)

At the start, it's good to have a rough outline of your sub-goals (rougher and shorter than my example can be fine) with the understanding that you'll break things up more as you get to them. This gives you some idea of how you'll actually get to your final, overall goal, so you can see some path and success looks realistic. (Sometimes writing out some sub-goals will make success look unrealistic because e.g. there's a problem in the middle that you have no candidate solutions for. Planning can reveal a problem that you think needs attention before you begin, which might indicate you shouldn't do this project rather than being something you can just solve as you get to it). You can also guess which sub-goals you're most likely to get stuck on and break them into sub-parts before starting to give yourself a better idea of whether you'll be able to succeed at them or not.

The bigger the projects you do, the more you'll want to plan at the start, because you don't want to get stuck and fail when you're 50 hours into the project. If you don't have a ton of past successes and a bunch of confidence, then stick to smaller projects that require less planning. You can incrementally increase your maximum project size and maximum amount of planning as you build up successes.

What if you want to learn philosophy? It's big. Should you avoid that early on? Or do you need to divide it into sub-goals? If you had to do everything by yourself, then you should avoid it early on to get some smaller successes finished. And you should at least roughly outline some sub-goals before starting to learn philosophy. However, you aren't alone. You can find a philosopher who has already figured out some sub-goals or sub-sub-goals and you can work on some of those without knowing everything about how they fit into the overall goal of learning philosophy.

You can also do an exploration project when you aren't ready for a learning project. In that case, your goal is to experiment and explore, and gather information to help decide whether you want to do it later and help you plan it if you do. If you aren't trying to make progress as a philosopher, that's easier, so it can be done with less prior skill and knowledge.

For a quick example of using sub-goals from someone who already learned about a topic, here are some smaller topics within Critical Fallibilism philosophy that you could work on learning about: trees, breakpoints, bottlenecks, automatizing ideas, integrating ideas, practice, local optima, qualitative and quantitive differences, debate methodology, text analysis, evolution, fallibilism, all evidence being interpreted with ideas, what an error is, or what a decisive argument is.

Without loss of generality, let's discuss one example from that list a little more. Local optima can be looked at in many ways. You can draw different types of diagrams to show them. You can write short essays that explain them in different ways. You can recognize them in many different contexts. You can write many different arguments that make a point about local optima. Those are all different ways of breaking the concept of local optima up into smaller parts. You can treat each type of diagram, essay, context or argument as a separate sub-project. That would easily let you do over a dozen different sub-projects within the local optima learning project, and that didn't even include the part where you learn what local optima are. One person might already have some knowledge about local optima and complete each sub-project quickly and easily, while another might need to break them down into sub-sub-projects, or might need more projects developing prerequisite skills before working on local optima.