init draft "Dance, Dance"

diff --git a/content/drafts/dance-dance.md b/content/drafts/dance-dance.md
new file mode 100644 (file)
index 0000000..d97d080
--- /dev/null
+++ b/content/drafts/dance-dance.md
@@ -0,0 +1,27 @@
+Title: Dance, Dance
+Date: 2027-01-01
+Status: draft
+Category: psychology
+Tags: dance
+
+As a brash but anxious young programmer in the before-time, I affected a standoffish attitude on the topic of _games_. Why would I learn to play a game as a human, I said, when I could just write a program to do it? [When an office chess game got started at my workplace](/blog/2019/May/minimax-search-and-the-structure-of-cognition/), I [wrote](/blog/2016/Jan/ideas-have-expirations/) [a chess engine](https://github.com/zackmdavis/Leafline) (a standard minimax search with α–β pruning) rather than learn anything about strategy or tactics. When my insufficiently requited love started hosting a meetup to play Zendo—a deduction game in which one player makes up a rule to positively or negatively classify arrangements of plastic pyramids, while the others submit arrangements to be classified to help them guess the rule—I wrote [a program](https://github.com/zackmdavis/Mezzanine) to efficiently propose examples to deduce a true rule from dozens of candidates. (Given a distribution over hypotheses, it generated an arrangement whose positive or negative classification would eliminate half the remaining probability-mass.)
+
+Now, in my old age (and the later dreamtime in which custom software is no longer such a flex), 
+
+[TODO—
+
+I no longer feel the need to pretend not to understand the appeal of games. (I even learned chess in 2024.)
+One game that's captured my interest is trivial from the standpoint of my write-a-program for it impulse.
+
+Or, it doesn't even translate into the abstract terms of "write a program for it" (in the way that chess is a logical game and the implementation of physically moving pieces doesn't matter at all), translated into abstract terms, it becomes trivial—at this-and-such time, send this-and-such arrow input. It's _only_ interesting as a "robotics" problem. The appeal is that it's a Moravec-proof game!
+
+And rather than writing a GOFAI program (as I did with chess or Zendo) the learning method is modern— the judgements give you a very clear gradient. (StepManiaX Perfect!!/Perfect/Late/Early/Miss has Late/Early as first-class judgements, but there's a config option to make DDR show Late/Early, too—that's important for gradient information. And the less we say about the South Korean game with the inflated grades and timing windows, the better.)
+
+Contrast to chess where ratings can be flat for years of play—you have to study; you don't necessarily learn enough from experience. Contrast to gymnastics where I didn't have enough of a gradient to learn a handstand. 
+
+There are a few verbal tips that can be given (try to alternate feet; at the beginning, I tried to treat the center as "home base"), but mostly, you play, and the judgements give you a gradient, and the levels give you a very smooth loss progression (patterns show up, then blue notes at level 6, and you get used to seeing the patterns as chunks and loading them as one motor program). You don't linguistically reason about where to put your feet, you just get a sense of it. (The pattern-chunking isn't a taught technique, it's a natural solution to the learning problem.)
+
+You can tell that you're not reading individual notes, because when you get out of sync with the beat, you miss a lot of consecutive notes correlatedly (as contrasted to the individual Gaussian error that tightens up and lets you get Great Fullcombos when previously a regular Fullcombo seemed hard). 
+]
+
+[TODO: not sure how to tie off the piece?]