Alright, @vgr, I'm down.

1 like = 1 take about play; up to 100. https://twitter.com/diviacaroline/status/1207119068001636352

1. We refer to many apparently different activities as "play" -- play a game, play an instrument, playing around, play the fool, play someone *for* a fool.

This is not a coincidence; we use the same word because the underlying structures are deeply related.

2. The common core of play - the center of playfulness - is to hold a goal that you really do want, to be nonetheless unsure of the outcome, and to thus hold the goal responsively.

3. Often, you can only hold the goal responsively, rather than firmly, because you know that the formulation of the goal that you're holding does not quite capture everything you value -- the current goal is neither necessary or sufficient to achieve the unspoken want.

4. We only really take on challenges when we have a clear, imaginable goal in mind, either implicit or explicit. So, being *able* to hold some sort of goal is critical for learning, taking on creative work, or achieving much of anything satisfying.

5. But holding your goal rigidly is the death of any activity that's difficult in both planning and execution, because rigid, uninformed goals set you up to take on impossible goals or subgoals, and just fail.

6. Novel goals, especially, are at least as difficult to plan as to execute - because the planning is a more abstract, less-intuitive activity. So you'll get new-to-you, difficult goals at least somewhat wrong. Even if you're following good instructions, you may misconceive them!

7. And so playfulness is the right attitude towards essentially any novel, difficult, creative activity. You will get plan wrongly, you will learn, and you might muddle through -- but you will only be able to do this if you can get there through play.

8. In particular, spending lots of time in play is indispensable for mastery of any deep skill (or combination of skills), and it's indispensable for solid learning.

(I don't say "teaching yourself" -- play is not really teaching. But it's constantly learning.)

9. Play is impossible too close to anything that doesn't feel safe. Wherever you're not safe, avoiding the unsafe outcome is a goal that you *must* hold and *cannot* responsively drop.

10. Any activity where you unresponsively demand that your own outcomes meet the goal you're holding is *work*.

To the extent that you can responsively put down the goal, it's play.

11. To link those first examples back: this is why we say that we "play" games. We call something a game, basically, when it's a limited system with an ultimate goal that doesn't matter by itself. The goodness of the game's goal is not in winning, but in the side effects of play.

12. But we'll hold the goal of a game lightly for that very reason. If we're caught in a dull, dull game, then there's joy, not harm, in responding to that dullness by putting down the goal.

This is why it's good form to resign a lost, boring game!

13. A game can only yield play when everyone in the game feels safe to play it. The "magic circle" of a game is, most of all, the shared cultural understanding of the limits on the costs and stakes of the game - e.g, time, money, social status.

14. When the upside of an uncertain goal is too low, you just do something else. When the downside is too high, the game is unacceptable to lose. For play to work well, you need the downside to be acceptable and the upside to be intuitively valuable - and uncertain.

15. The upsides and downsides controlling whether play is acceptable and worthwhile are your underlying valuations of the thing, not necessarily your explicit values. These may change a bunch over time.

16. One strong result of all this:

You can only get sort of OK at something if you fear failing at it a lot. You can absorb painful failures in practice and count yourself nobly virtuous; but you won't become great if you can't play, and you can't play if you don't feel safe.

17. Put another way: anywhere you want to become great, first learn how to play. Anywhere you want to play, first learn how to fail without getting hurt.

18. To watch super clear, extremely obvious, only-slightly-structured group play among kids and adults, watch people at a juggling festival. Almost anyone that likes to juggle is quite good at play and comfortable with moderate failure.

19. Among the jugglers I've known, a common pastime was Wait, Could We Learn *This*? The ability to hold a silly goal lightly - but maybe for hours - is central to inventing and learning new, strange tricks, at least half of which turn out practically impossible.

20. Repeating the same "Wait, Could We Do This?" loop is an astonishingly fast way to, say, bootstrap knowledge of programming languages and tech. Or find tunes and techniques on a musical instrument. Or prove new mathematics. Or compose an essay. Or reinterpret great text.

21. It's a great way to induce much-more social skills, too, though those will, again, require either a psychological constitution or a social environment where repeated failures are actually acceptable.

22. This is "fiddling":

Do something that (a) you're learning, and (b) produces a visible effect.

1. Would any other concrete effect feel satisfying?
2. Imagine and cause *that* effect.

If you succeed, fail, or learn it's impossible, aim to understand how or why.

Repeat.

23. For the play-impaired, the Fiddling Procedure is a loose container for instructive play in a fast-feedback domain.

If the "feel satisfying" part seems dumb instead of intuitive, you're probably attempting work. It's not going to work. Stahp.

24. I expect almost everyone to be pretty thoroughly play-impaired.

Why? Schooling, and because play can't be incentivized: Play is too illegible to reward directly. Its results are too heterogenous to reward indirectly. Getting play by punishing its absence does *not* get play.

25. If play is so important, and you can't incentivize it in a System-2 way, how do you play?

You can gather its time, space, and resources. Pre-accept lots of failure. Prime it by Fiddling. Mostly, repeatedly draw your attention to satisfying, new goals in the domain of play.

26. You can also play more in a domain by getting sucked into a social group that plays in the domain - you pick up the play intention. Adults are mostly bad at play, so you have to filter. Seek a group where excitement and skill growth are high-status and failures are welcome.

27. If you're looking for a group to play with in a domain, deliberately filter out groups where hardcore-ness or steady success are high-status.

Those aren't bad on their own terms, per se, but they are bad for play.

28. A specific, actionable filter for a group:

Ask them about what people in the group have done, and notice what stories they tell and how they feel about them. It's probably a good group if people are proud of each other's successes *and* their interesting failures.

29. So, play is critical to learning goals that maybe you already understand. Play is also a central way to achieve those of your goals that you can't articulate well enough to deliberately plan for.

30. To achieve a goal that you can't plan for, you have to (a) get into a situation where you can notice that goodness is available, and (b) hold flexible enough plans to achieve that goodness once recognized. (a) demands exploration, (b) demands an active but flexible mindset.

31. Good play is an extremely efficient exploration of your goals. Good play gets you into novel situations that are likely to be of intuitive value to you, and good play leaves you the flexibility to pursue freshly-recognized goals that you've got.

32. Good play lets you explore and learn about your preferences, in dialogue with reality, is it's both effective for navigating an unfamiliar domain and effective for refining your model of your own goals. How you feel in a new situation? Excellent feedback on what you want.

33. The underlying motivation behind lots of deliberate mediocrity is to keep play from becoming work. That's a critical goal! Avoiding the underlying perfectionism instead would work with lower side effects.

34. You can't go from work to play just by willing an unacceptable failure to be acceptable. You can, though, figure out why it would be bad, and why that would be bad, and so on, and fix your plan or your belief somewhere in that chain.

35. After safety, the major factor determining whether or not a domain supports play is the speed of feedback - after you've taken an action, how quickly do you find out whether you're pleased with it or not?

36. Instantaneous is vastly better than in seconds. In seconds is vastly better than minutes. In minutes is probably too slow to support play.

37. When you can't build faster feedback, you can learn faster preferences. An experienced painter immediately knows whether or not they like the last few strokes, even though it'll take hours to learn how the full painting comes out.

38. You can learn faster preferences by guessing at a strategy, playing inside that strategy, and seeing how it turns out in the end. Hold the strategy itself as a bunch of goals, with the same responsive lightness as the overall goal itself.

39. Well-honed "faster" preferences get themselves into your sense of satisfying outcomes, so even though the process should formally recurse, you shouldn't really need to over time.

40. If you imagine your implicit mind is basically a neural net, satisfying play is more or less running its built-in, already-well-honed training algorithm.

41. From this model of play, you should expect young kids to get bored quickly of toys or games that provide inflexible affordances, like any stuffed-animal-with-buttons. They might keep messing with it, but the sense of play will vanish unless they just ignore the buttons.