Skill Acquisition

post by Screwtape · 2023-03-21T00:49:21.506Z · LW · GW · None comments


No comments

Summary: Juggle! Draw an egg! Sing! Learn some sign language! The ideal rational agent should be capable of picking up new skills quickly and with a minimal amount of training. So practice practicing things!

Tags: Large, Repeatable

Purpose: To learn how to learn, studying how you acquire a new skill and experimenting with different ways to do so.

Materials: Varies. See Topics.

Announcement Text

I'm going to try and learn a new skill in an attempt to figure out how we learn things. You're all invited to hang out and try and learn a new skill as well!

No, I'm not trying to figure out the neurology of learning- not today anyway. The thing I want to practice is the process of learning a new thing, and how to learn things faster or easier. 

We've all learned things over the course of our life, sometimes deliberately and sometimes incidentally. Usually when we’re trying to pick up a new skill, there’s a particular need for it. Whether it's a new code framework for my job, a new equation for a test, or a new dance move before a date, I'm not paying as much attention to how I'm learning as I am to what I'm trying to learn. Since I don't have an urgent need to be a juggler, I'm hoping this will give me some space to experiment with how to learn best.

And, like every good experiment, I could use a bigger sample size. Come hang out, practice something you've wanted to learn, and we'll compare notes on what seems to be working. 


1. As people arrive, pass out the tools necessary for the activity. Ask if they have any previous experience with it. Write down their name, the experience level, and how they plan to learn.

2. If the activity is best practiced in pairs, practice with someone when there’s an odd number of attendees and match new arrivals with your current partner if there’s an even number of attendees. Basically, either count or don’t count yourself as needed to make it an even number.

3. Encourage people when they make progress.

4. Towards the middle of your time, pull everyone together and ask how they’re faring. Ask what approach they’ve been using to learn. Suggest switching how they’re trying to learn, if they haven’t already. Write this down.

5. At the end of the time, ask for a demonstration if that’s feasible. Write down how much progress people felt like they made, and what approaches they used to learn.


By default, you can just throw yourself at the new skill and figure it out from trial and error. This works pretty well for juggling and not at all for a new language. (You need at least a dictionary or something to learn the words from!) If you have a teacher, that’s one way to learn. If you have a book on the subject, that’s another way to learn. If you have a couple youtube videos on the subject, that’s yet another way to learn. It helps to have as many different ways to learn available and around as possible, so people can experiment.

One variation is to pair up after a short while and try to help each other. Watch the other closely, point out what they might not see.

By default, you’re just trying to see how far you get using different methods. Aiming at specific thresholds in a minimum amount of time is an option, such as keeping three juggling balls in the air as soon as you can.


In general, quick physical skills with your hands are good, languages are good, and music with cheap or free instruments are good. If there's a skill that someone in your group is excited about teaching, that's also a good reason to pick that topic!

Notes: Skill Acquisition, more than most meetup-in-a-box activities, looks different depending on the topic. The ideal topic is something which 1. You fail at safely when you fail, 2. You fail at unambiguously when you fail, 3. You fail at quickly when you fail, and 4. You fail at cheaply when you fail. Juggling is the paradigmatic example. If you fail to juggle you will fail in a couple of seconds when the balls fall harmlessly to the ground where you can pick them back up again. Reciting a poem from memory likewise will usually fail in about a minute and can be reset simply by rereading the poem and starting over.

Backflips fail to be safe. If you screw up a backflip you can hurt yourself quite badly. 

Composing poetry fails to be unambiguous. It’s not obvious if a rhyme is good or bad.

Writing a novel fails to be quick. You can’t do it, fail, and try again a dozen times in an afternoon.

Cooking fails to be cheap. You have to buy ingredients for each attempt.

Some activities are cheap to fail at once you have an initial payment. These can be borderline for Skill Acquisition. Once I purchased a guitar, each individual failure to play a song became very cheap. Having a guitar available for each of your attendees is likely cost prohibitive, and taking turns slows down learning too much. Adjust the topic to your financial inclination; a sack full of juggling balls, a small pile of ocarinas, or a multipack of shoelaces for knot tying can be had for the price of a pizza or two. 

Lastly, I wish to emphasize that the priority is figuring out how you learn. Try different things! Write down what you tried and how fast progress was made. Compare notes with each other! 

Credits: This owes its impetus to How To Become A Thousand Year Old Vampire [LW · GW], as well as a slightly literal take on Cup-Stacking Skills [LW · GW].

None comments

Comments sorted by top scores.