Download Me (Windows)

A Simple Twist

On top of being a language enthusiast, I'm also really into chess. I watch tournaments, follow grandmasters' Twitch streams, play on some chess apps, (try to) solve chess puzzles, etc. I inevitably started contrasting chess pedagogy with language pedagogy, and I found some possible issues:
  • Like bad language instructors, chess instructors teach a lot of theory and analyze games.
  • Chess lessons involve curricular development approaches similar to situational syllabi.
  • Private chess centers appear to run on a student retention model.
  • Instructors routinely attribute failures to students and successes to themselves.
  • Most people who pursue chess mastery fail, and the aforementioned likely don't help things.
So, I thought on this question: How would a naturalist language instructor craft chess lessons? What would he keep? What would he omit? How would he organize them? I programmed some ideas, tested them on myself, and noticed myself getting better. No theory. No analysis. Just time and manageable input.

How did you build it, and how does it work?

I treated the material like the lexicon of a language, and I devised 108 boards that eliminated a certain amount of material. Limiting the "lexicon" in this way impacts the "expressiveness" of the chess language, "expressiveness" being the number of possible positions. For those who know FEN's, you can think of what kinds of FEN strings become unavailable once certain material is gone. I then used Stockfish 11 to repeatedly assess those board positions. Those assessments were used to rank the boards from easiest to hardest.

The rest works like a standard chess program. You play chess against Stockfish 11 with a certain material advantage until either (a) you win more games than you lose or (b) you win three games in a row. At either point, you advance a level and play further games with a slightly worse material advantage. As you progress, that material advantage reduces to zero, and you just play standard games.

How is this any different from chess apps already out there?

Most chess apps, in my opinion, do something pretty stupid. They train you against a strong chess engine, but give you levels by dumbing that engine down. Bad human play is really not something these engines are designed to do. Why not let the engine always play at its maximum level? What if, instead, we handicap its material so that we can "decode" the board and "track" play and counter-play up to our mental limits? That would match competent language instruction -- comprehensible, yet incrementally complex "messages". Hence the twist.

It's not just about what I put into SpotChess, though. What I left out -- move recommendations, visible move recording, game scoring, etc. -- also matters. I want players to visualize their play more and to analyze it less. Only the board's dark squares indicate your quality or state of play:
  • Winning positions are bluer.
  • Losing positions are redder.
  • Mates are gray.
Further, (as of version 1.1) clicking any square with a legally movable piece imposes the touch-move rule (see FIDE Laws of Chess Article 4.3a), which shows by highlighting the clicked square in yellow. You can still see your available legal moves hovering over your pieces.

Where can I download and play SpotChess?

The Windows version is here.

The Android version is in late-stage development.

Shut up and take my money!

No need to tell me twice.

No comments:

Post a Comment