Designing Your Own Grid-Capture Game Variant

The grid-capture genre — the family of games that includes dots and boxes, Dot Clash, hex variants, and territory games like Go — has been generating new variants for over a hundred years. Most of them are forgotten. A few stick around. Designing a new variant is harder than it looks, but the design space is wide and there's room for new ideas.

This post is a designer's primer on grid-capture games. What dimensions can you vary? What stays the same across all viable variants? How do you test whether your design works?

What stays constant across all grid-capture games

Some properties seem load-bearing. Variants that change them tend to break:

• Two players, alternating turns. Three-player versions exist (see multi-player dots and boxes) but most don't sustain interest.
• A discrete grid. Continuous boards lose the tactile quality.
• Territorial scoring. The basic loop of "make moves to claim territory" is what makes the genre coherent.
• Perfect information. Both players see everything. Hidden information turns the game into a different genre.
• Finite game length. Games end when the board is full or a target is hit.

Variants that violate these rarely succeed. The genre's coherence depends on them.

The dimensions you can vary

Within the constants, here are the design knobs:

Knob 1: Grid topology

• Square grid (classical dots and boxes, Dot Clash): default, well-understood.
• Triangular grid: each "box" is a triangle. Different chain structures, different parity calculations. See hex and triangular variants.
• Hex grid: hexagons as the unit. Smoother adjacency, often combined with edge-drawing.
• Cylindrical/torus grid: edges wrap around. No corners. Eliminates corner-strategy advantage; rebalances the game.
• Irregular grid: shapes vary. Hard to teach, but interesting variation per game.

Knob 2: Turn / capture rule

• Free turn after capture (dots and boxes): the chain rule emerges from this.
• No free turn after capture: each move is a single line, no chains. Simpler tactical structure.
• Multi-line moves: each turn places multiple lines. More tactical depth per move.
• Move-then-capture: you move, then resolve all captures at once. Different timing.

Knob 3: Scoring

• Box count (classical): whoever has more boxes wins.
• Score target (Dot Clash): first to N captures wins. Game can end early.
• Weighted scoring: corner boxes worth more, or boxes near edges worth more.
• Negative scoring: every box you don't capture costs you points. Encourages denser play.

Knob 4: Information structure

• Full information: both players see everything (default).
• Fog of war: each player only sees their own side or recent moves. Changes the game profoundly.
• Asynchronous moves: both players move simultaneously, conflicts resolved at end of round.

These last two break the "perfect information" constant we discussed above. Variants that try them tend to feel like a different game entirely.

Knob 5: Capture mechanic

• Box completion: drawing all 4 sides claims the box (dots and boxes).
• Region enclosure: closing a region claims everything inside (Dot Clash).
• Capture by surround: like Go — boxes are claimed when surrounded, lose them when captured back.
• Stone placement: like Go — you place markers, not lines.

Knob 6: Player asymmetry

• Symmetric: both players have identical capabilities (default).
• Asymmetric pieces: one player has long-range moves, the other short-range.
• Asymmetric goals: one player tries to capture territory, the other tries to prevent it.
• Handicap: one player starts with extra material or restricted area.

Asymmetric variants are harder to balance but produce more strategic variety.

Combining knobs

A new variant is a combination of choices on each knob. Some combinations:

• Hex grid + score target + region enclosure = something like Dot Clash but on hexes. Probably interesting.
• Square grid + free turn + weighted scoring (corners worth 3x) = makes corner strategy hyper-decisive. Probably too one-dimensional.
• Cylindrical grid + box completion + box count = dots and boxes without corners. Plausibly fun, definitely different.
• Square grid + multi-line moves (2 lines per turn) + box count = "double dots and boxes." Higher variance, lower decision quality per move.

Most combinations don't work. A few do. The job of the designer is to find them.

How to test a new variant

The standard playtesting process:

Step 1: Self-play

Play 5 games against yourself. Look for:

• Are games decisive? If most games end in ties or near-ties, scoring is too symmetric.
• Is the game over too quickly? If 70% of games end in under 15 moves, the game is shallow.
• Is there a clear strategic structure? Can you describe an "opening, middlegame, endgame" arc?

Step 2: Test with a partner

Play 5 games against another person. Look for:

• Does each game feel different? If every game plays out the same way, the variant is too restrictive.
• Does the better player usually win? If outcomes feel random, the game has too much variance for skill to matter.
• Is there a learning curve? Does the second game feel different from the first?

Step 3: Test with strangers

Find 3 people who've never seen the variant. Teach them in under 5 minutes. Watch them play. Look for:

• Did the rules transmit cleanly? If you have to keep correcting them, the rules are too complex.
• Did they figure out a strategy? Or did they play random moves the whole game?
• Did they want to play again? This is the single most important signal.

Step 4: Iterate

Almost no variant survives playtesting unchanged. You'll iterate on rules, sometimes substantially. Don't fall in love with your first design — it's a draft.

What's been tried and what hasn't

Many variants have been published over the years. A non-exhaustive list of what's been tried:

• Pen and paper: dots and boxes, sprouts, Boxes (variant), Squares.
• Digital: Dot Clash, various app implementations of dots and boxes.
• Hex variants: Hex, Y, *Star.
• Surround variants: Go, Othello (different mechanic but same genus).

What hasn't been tried much:

• Time-asymmetric variants where one player has more time per move than the other.
• Score-decay variants where captured boxes lose value over the game.
• Drafting variants where players choose initial positions for their dots before play begins.
• Cooperative variants where two players against an AI or against the clock.
• Network grids rather than full grids — sparse boards with explicit connections.

Each of these is a potential design space. None has been thoroughly explored. If you want to design something new, this is where there's room.

How Dot Clash fits in

Dot Clash is itself a variant. Compared to classical dots and boxes:

• Larger grid (25×25 vs. 5×5).
• Score target rather than full-board scoring.
• Region-enclosure capture rather than box completion.
• Time controls (per-move limits).

Each of these design choices has been tested by thousands of games. The design holds — the game is fun, has strategic depth, and produces decisive outcomes most of the time.

For the deeper analysis of why these choices work, see grid-capture game design principles and Dot Clash vs. dots and boxes key differences.

Why design new variants?

Two reasons.

One: design teaches you the game. Even if your variant doesn't catch on, the act of trying to design one forces you to understand what makes the existing variants work. You'll see the chain rule, corner strategy, and parity dynamics more clearly because you'll see what happens when you remove them.

Two: occasional variants stick. Sprouts (1967) is now a classical game. Dot Clash (2026) might be a classical game in 50 years. Most variants don't make it, but the ones that do enrich the genre.

If you have an idea, prototype it. The cost is a notebook and a willing playtest partner.

In short

• The genre constants are two players, discrete grid, full information, territory scoring, finite length.
• The design knobs include grid topology, turn rules, scoring, information, capture mechanic, asymmetry.
• Playtest in three stages: self-play, partner, strangers.
• Most variants fail. A few succeed. Try anyway.

Every variant is a hypothesis about what makes grid-capture games fun. The good ones teach us something. Even the failures sharpen our understanding.