Trigame is not only something to play. It is also a place to build personalities,
shape strategies, and help expand the world of the rabbits who inhabit it.
It is a growing geometric strategy game played on triangular tiles. The board is built as you play, paths and tokens interact, and scoring comes from presence, subs, and full tiles rather than from simple elimination.
The board expands as players place tiles. Paths become pressure, mobility, and scoring lanes. Sub-areas and full tiles matter on top of token count, and each rabbit in the fluffle can develop a different style.
Getting in is simple. The entry cards on the right let you start immediately.
Fastest path in. Play immediately. Use your own name or let the page assign Guest-1, Guest-2, and so on.
Sign up with only a user name, optional email, and password. Enough for identity without friction.
Saved rabbits, avatars, queueing, and online matches can grow from the same identity system.
Trigame began as a hand-drawn terrain game idea and evolved into something more abstract, more constructive, and more strategically alive. What started as tiles on paper became a living board: tokens create paths, paths create tokens, subs are captured, full tiles are won, and every move changes the shape of the next problem.
Along the way, the project picked up a whole rabbit ladder of opponents, a symbolic move language, a Game Record system, AI training tools, and a help/manual structure fit for curious players who want to look under the hood.
Earlier versions focused on colored terrain tiles and physical placement. As the game moved into the browser, coding “accidents” sometimes revealed better rules than the originals. Tile overlap became a feature. Logging became infrastructure. The AI stopped being a difficulty setting and became a cast of personalities.
The current design keeps the spirit of the old board experiments while leaning into the strengths of a digital game: score tracking, AI ladders, move notation, automatic path effects, training matches, and archives of the project's long development trail.
Enter a name if you like, or leave it blank and let the page assign one. Then head either straight into solo play or into the Green Room to play with others.
Trigame's Green Room is where folks go to join in games with others playing online.
Create a simple member account. Email is optional, though adding one makes it easier to recover your account later.
Return with your user name or email and your password.
If you added an email address when you signed up, you can log in with that email instead of your user name.
If you do not remember your password, or if you never added an email address, the simplest path for now is to sign up again.
The built-in opponents are not merely “easy” through “hard.” Each of Rusty Rabbit's kin can favor a different habit: Kit loves tiny fireworks, Newb-Bble likes more tiles, Lop-Sided scrambles for one neighborhood, Burrow Scout notices lanes, and stronger rabbits will push deeper into denial, conversion, and endgame control.
The long-term goal is not just stronger AI, but more memorable AI.
To this end, these personalities are handled by individual JavaScript files. We are considering the option of allowing players to create their own AI game characters, so we have included what we are doing now in case anyone wants to see in advance.
Fluffle character personalities JavaScript info..
Trigame has a long visible history of versions, from SVG and early Canvas boards to scoring, sound, Game Record notation, Rusty Rabbit, and the current UI. That archive is part of the project's charm and may eventually grow into a structured lesson series.
Read more on the about page →
Rusty Rabbit is the name ChatGPT chose from working on another related project in which he will be creating a series of lessons on a variety of subjects including writing code. The process of developing this game step by step will serve as examples for his classes. ...Stay tuned!
Part of this ongoing educational aspect of Rusty's is this problem we call the Kit problem. Looking for the best possible moves for a single tile game. See more here:
The Kit problem for graduate level math heads.
Trigame combines two distinct layers of complexity: a finite local problem and an expanding global problem. A single tile may be treated as a closed combinatorial puzzle with fixed internal structure, while the game as a whole behaves like an open recursive network in which each placement can create additional future placements.
In the discussion that informed this summary, the internal logic of a single tile was described as having approximately 39,916,800 possible permutations. That figure refers only to the bounded logic of one tile or one local sub-problem. The wider game becomes far more complex because each tile placement opens additional sockets, allowing the board to continue expanding outward rather than remaining confined to a fixed grid.
| Layer | Type | Description |
|---|---|---|
| Single Tile | Closed / Finite | A bounded combinatorial puzzle governed by fixed internal elements and a finite state space. |
| Whole Game | Recursive / Expanding | A growing network of placements in which each move can reshape the future board and enlarge the move tree. |
In the same discussion, the game was described as gaining a net of 6 new sockets per tile placement when one socket is consumed and seven new ones are opened. Under that model, the number of available future placements grows rather than shrinks, making the overall game tree increasingly wide and deep as play continues.
| Game | Board Bounds | Termination | Complexity Character |
|---|---|---|---|
| Checkers | Fixed 8×8 board (32 dark squares in play) | Yes | Large but finite; smaller than Chess or Go in standard estimates, with roughly 5 × 1020 reachable positions in classic solving analyses, and solved as a draw under perfect play. |
| Chess | Fixed 8×8 board | Yes | Extremely large but finite game tree; often represented by the Shannon estimate of about 10120. |
| Go | Fixed 19×19 board | Yes | Vast positional and game-tree complexity, often estimated around 10360 for 19×19 Go. |
| Trigame | Expanding field | Not fixed by board size alone | Locally finite but globally divergent: complexity grows through recursive placement and expansion. |
The discussion also proposed rough comparison points suggesting that:
These figures should be read as illustrative comparisons, not as formal proofs, but they communicate an important point: Trigame does not merely have a large number of possible positions; it also has a growth mechanism that can continue enlarging the space of possible play.
In bounded games such as Chess, strong engines can rely heavily on deep search because the board is fixed. In Go, Monte Carlo Tree Search and neural evaluation work well because, although the state space is vast, the board is still finite. Trigame places a different burden on intelligence: the challenge is not only to calculate well, but to manage the growth of the board itself.
For that reason, Trigame naturally favors:
Chess and Go are deep games played inside fences. Trigame is a deep game in which the field itself can continue to grow. That makes it locally finite, but globally explosive.