An openings trainer for a vibe-based brain

I started playing chess when everyone did: The Queen’s Gambit dropped, Beth Harmon stared at a ceiling full of pieces, and a few million of us downloaded the app convinced we’d be next. Most people churned. I didn’t. I love this game in a way I can’t fully account for. The problem is how I love it.

Because an ADHD brain doesn’t want a forty-minute rapid game where it has to sit still and calculate. It wants the dopamine slot machine. So I play bullet. One-minute games, premoves, flag the opponent, queue the next one before the board even clears. It’s the same machine I wrote up in the ADHD playbook, just wearing a different skin: the thumb is going to scroll regardless, so it scrolls through chess instead of Reels. Which feels productive and absolutely is not.

Here’s what a thousand bullet games actually trained: I move on vibe. Not calculation, not theory. Pattern-match the position to a feeling, slap a piece down, react when it breaks. It’s fast, it’s fun, and it has a hard ceiling, which I have apparently hit. 1700 rapid, 1500 blitz, and very stuck. The vibes carry you to exactly the point where everyone else also has decent vibes, and then they stop.

The opening is where this bites first and hardest. By move ten in a real game I’m already improvising, burning clock and brain on positions that have been theory for a hundred years, while my opponent is still in book and spending nothing. I’m solving a problem that was solved before I was born. That’s not a talent gap. That’s a homework gap.

What I actually want to build

A web app to study openings. Not a database to passively scroll, the internet has a hundred of those and they all bore me into closing the tab. I want the thing that worked everywhere else: compile the lookup down into a reflex. Same move I made with the Swipe-to-Pwn deck in the playbook, where the front of the card is a signal you’d actually see on the board and the back is the line it demands. Spaced repetition, decision-format, so the opening stops being something I recall and becomes something I do. This is reading for meaning on the first pass aimed at a chessboard: I want the move to land without three rereads of the position.

The honest design problem is that I don’t know what the app is yet. A few shapes I’m circling:

• Spaced-repetition over a tree. Pick a repertoire (something sharp for White, one defense each against e4 and d4), and drill the lines as flashcards that resurface right before I’d forget them. The hard part isn’t the cards. It’s authoring a repertoire worth memorizing in the first place.
• Drill against the tree, not just recite it. Recognizing the right move when prompted is easy and a lie. The real test is playing it cold, from the side I struggle with, getting punished the instant I leave book. More like a sparring partner than a deck.
• Punish-the-deviation mode. Most of my rating isn’t lost when I misplay the opening. It’s lost when my opponent plays a dumb non-book move and I don’t know how to refute it, because I only ever studied the “correct” line. I want the trainer to throw the garbage at me on purpose.

All three rest on the same bet this whole garden keeps making: engineer the environment, not the willpower. I’m not going to out-discipline my own brain into loving opening theory. But I might be able to build a tool that makes the rep so frictionless the ADHD brain mistakes it for the slot machine and pulls the lever anyway.

Open threads

Wide open, honestly, which is why this is a seed and not a project with a green checkmark.

What’s the stack? Static and local-first would fit the rest of this garden, but a real openings engine wants a move tree, eval, and probably a Lichess or chess.com data pull, which is a different beast than serving Markdown. Do I author my own repertoire (slow, mine, correct for me) or import a grandmaster’s (fast, generic, possibly above my pay grade)? And the uncomfortable one: is an openings trainer even the right lever at 1700, or am I just building a tool to avoid the tactics grind I actually need? Plausibly I’m doing exactly what the last open thread warned about, dressing up procrastination as engineering.

I’ll find out the only way I ever do, which is building it in the open and logging the reps. If the rating moves, the app earned its place. If it doesn’t, at least I’ll have lost on a line I chose instead of a vibe I felt.