Dominoes

Opening

Dominoes does exactly one thing and does it properly: classic block dominoes, double-six set, human versus a computer opponent that is actually trying to beat you. It's part of Frank Force's LittleJS Arcade collection, and it carries that project's usual house style — tiny footprint, LittleJS engine underneath, no dependencies to speak of, MIT licensed, done.

The AI is the headline feature

Most browser dominoes implementations either play randomly or follow a simple greedy heuristic. This one runs a real alpha-beta search (dominoGame/the bundled gameAI.js) over the legal move tree, using an evaluation function that weighs pip counts, blocking potential, and endgame position. It shows: the AI reliably takes the tile that denies you the most options rather than just the first legal one it finds, and it opens with the highest double in hand exactly as competitive human players do. It's a small thing, but it's the difference between a toy and a game worth playing more than once.

Controls, confirmed by testing rather than guessing

The interaction model is tap-to-select, tap-to-place: tap a tile in your hand that has a legal move, and a highlighted (or crossed-out, if it doesn't fit) drop zone appears at whichever end(s) of the chain it can go. Tap the valid zone to commit.

This reviewed it properly rather than taking that description on faith. Reading the live game state during a real Chromium touch-emulation session (CDP-dispatched touch, not synthetic click events) confirmed: tapping a tile that has no legal move correctly does nothing (by design — no state change), tapping a legal tile sets selectedTileId and reveals the real drop-zone screen coordinates, and tapping that drop zone correctly grows the chain, shrinks the human hand, and hands the turn to the AI. Every step of that pipeline was verified against actual game-state values, not screenshots alone. It works the same way with a mouse click on desktop. The one rough edge: tapping an illegal tile gives no feedback at all — no shake, no denial sound — so a first-time player might tap around for a few seconds before realizing which tiles are live. Everything downstream of a correct tap is solid.

Presentation

It's plain on purpose: a green felt background, grey-backed tiles in hand, pips rendered in LittleJS's simple vector style, doubles drawn perpendicular to the chain the way a real domino set would lay them out. There's no music, and sound is limited to a placement click and win/lose stingers. That's a fair trade for a project this size, and nothing about it feels unfinished — it's just minimal.

Final Verdict

A small, honest, well-built dominoes game: a real thinking opponent, correct rules, and touch controls that were tested (not assumed) to work on an actual phone. Presentation is bare-bones and there's no in-game hint for illegal-tile taps, but neither issue gets in the way of a satisfying quick match. Recommended for anyone who wants classic dominoes with a genuine AI to beat.