The Rules in Full

Hex is played on an 11x11 rhombus of hexagonal cells - the size used here and the classic size favored by the game's inventors, Piet Hein and John Nash. Each player owns one pair of opposite edges. On PlayVersusAI, Red owns the top and bottom edges and moves first; Blue owns the left and right edges. The colored borders on the board show you which edges belong to whom.

Players alternate placing one stone of their color on any empty cell. Stones never move once placed, and they are never captured or removed. Red wins by completing an unbroken chain of red stones linking the top edge to the bottom edge; Blue wins by linking left to right. Cells count as connected when they share a hexagonal side - each cell touches up to six neighbors. The four corner cells belong to both adjacent edges.

There are no draws in Hex, ever. This isn't a house rule - it's a mathematical theorem: any completely filled Hex board contains exactly one winning connection. One player's chain must cross, and a wall that fully blocks your opponent is itself a winning connection for you. This is also why defense and offense are the same thing in Hex: every blocking stone is simultaneously a building stone.

How a Typical Game Flows

Games open in the center. The middle cells serve both players' connection plans, so early stones cluster there - an opening stone in or near the central hex is so strong that serious Hex uses a swap rule to balance it (not used on this site, so enjoy your first-move advantage).

The midgame is a fight over routes rather than territory. Each player sketches a path between their edges using loose, bridge-connected stones, while probing for the weak links in the opponent's sketch. Positions rarely look like walls; they look like scattered constellations whose connections only become visible when challenged. When one player pushes through a contested area, play often erupts into a ladder - a forced sequence of adjacent stones racing along a diagonal - whose outcome was usually decided by a stone placed many moves earlier.

The endgame is about converting near-connections into real ones. A strong position resolves itself: every intrusion your opponent makes into a bridge or template has a prepared answer, and the connection completes by force. Most games between experienced players end with a resignation well before a chain physically touches both edges, because the remaining connection is provably unstoppable. Against the AI here, the game ends the moment a full chain forms, and the winning path lights up in green.

Core Strategies, Named and Explained

1. The Bridge

The most important pattern in Hex. Two stones placed a knight's-step apart - diagonal, with exactly two empty cells between them that both touch both stones - are effectively connected. If your opponent plays one of the two linking cells, you calmly take the other. Bridges let you advance two rows per move instead of one; build your main path almost entirely from them.

2. Block by Building

Never play a purely defensive stone. Because a full wall equals a win, the best block is always the one that extends your own connection. Before answering a threat, ask which of the blocking cells also shortens your own path - there is almost always one, and finding it is the core skill of the game.

3. Ladders and Escapes

Don't start a ladder you haven't already won. When you push along the side of an enemy wall, each of your stones forces one reply, driving the sequence toward the edge - where you usually arrive one row short and lose the race. The remedy is the ladder escape: a stone placed in advance along the ladder's future track that lets your chain jump to the edge. Place escapes early; recognize when your opponent has one and refuse the ladder entirely.

4. Edge Templates

Stones near your edge connect themselves. A stone on your second row with the two cells beneath it empty cannot be cut off from the edge - that's the simplest template, and larger ones exist for rows three and four. Practical consequence: don't waste moves playing all the way to your own edge. Stop at a templated distance and spend the saved moves in the contested middle.

5. Path Counting

Evaluate like the AI does. After every move, estimate the minimum number of stones each side still needs to complete a connection, treating bridges as already connected. If your number is lower, press forward; if higher, your next stone should damage their count more than it improves yours. This single habit will improve your play faster than memorizing any pattern.

6. Play the Obtuse Corner

Corners are not symmetric. Each player has two "obtuse" corners where their edges meet at a wide angle - for Red, the top-right and bottom-left on this board. Routes through your obtuse corners are shorter and easier to defend than routes through the acute ones. When your central path gets blocked, bend it toward an obtuse corner rather than fighting straight through.

Facing Each AI Difficulty

Easy plays nearly randomly, favoring cells near existing stones. It will not punish anything, so use it to drill mechanics: build a full bridge path from top to bottom, then practice defending the bridge cells when the AI stumbles into them.

Medium plays a greedy shortest-path heuristic: every stone it places is the one that most improves its own connection or damages yours, measured by Dijkstra path analysis where its stones cost nothing and yours are walls. It has no lookahead, so beat it with two-move ideas - moves whose real point only appears after the follow-up, like ladder escapes and double-purpose stones it undervalues.

Hard adds a two-move search on top of the path heuristic: it considers its best candidate stones, then checks your strongest reply to each. One-move traps stop working, and it defends bridges correctly. Play for long-term structure - lay ladder escapes and corner routes early, before its two-ply horizon can see why they matter.

AI Training searches three moves deep over the top candidate cells with a strict two-second budget, and its evaluation understands bridge connectivity. It punishes loose play immediately, so every stone must do two jobs. Your realistic edge is the long game: choices whose payoff sits four or more moves out - overall path shape, obtuse-corner strategy, and early escape stones - are beyond its horizon. Expect to lose while you learn; even a close game against this level means your Hex is genuinely improving.

Frequently Asked Questions

Why can't Hex end in a draw?
Because of the board's topology. The Hex theorem proves that any fully filled board contains a winning chain for exactly one player - blocking your opponent completely requires building an unbroken wall between their edges, and that wall is itself your winning connection. Fill the board and someone has won, always.
Is moving first a big advantage?
Yes - John Nash proved the first player has a theoretical winning strategy via the famous "strategy stealing" argument, though nobody knows the actual strategy for 11x11. Competitive Hex neutralizes this with the swap rule, where the second player may steal the first move. On this site there's no swap and you always move first against the AI, so you hold the theoretical edge - use it.
Can my stones be captured or moved?
Never. Hex has no captures, no movement, and no removal - every stone stays where you placed it for the whole game. That permanence is what makes early moves so important: a stone placed on move five can decide a ladder on move forty.
What's the fastest way to get better at Hex?
Master bridges before anything else - build entire games out of them and practice answering intrusions into the linking cells. Then learn to count paths: after each move, estimate how many stones each player still needs. Those two habits, plus refusing ladders you haven't prepared, will carry you past the Hard AI.