How to play Bridge (and why attacking and defending are one move)

Bridge is a free connection game you play in the browser. A rhombus of hexagons sits between four coloured rails: two in your colour on the left and right, two in the bot’s along the top and bottom.

You are trying to link your two rails with an unbroken chain of your own hexes. The bot is trying to link its two. Tiles are never removed, never moved and never captured — once a hex is yours, it is yours for the rest of the game.

No download, no account, no app. It runs on a phone on a bad connection, and it keeps working with the network switched off.

The rules, in two sentences

Take turns placing one tile on any empty hex, and win by joining your two edges with a chain of touching tiles of your colour.

That is the whole game. There is nothing else to learn — no captures, no special pieces, no second kind of move.

Why blocking and building are the same move

This is the idea the game is built on, and it is worth being precise about.

In most two-player games attack and defence are different actions. You either develop your own position or you interfere with theirs, and part of the skill is knowing which to spend your turn on.

In a connection game that choice does not exist. Look at the board: your chain runs left to right, and the bot’s runs top to bottom. For the bot’s chain to get from top to bottom, it has to cross the corridor your chain needs. Any hex on that crossing is a hex you both want.

So when you take it, two things happen at once, from one move:

1. Your chain gets one hex closer to the far rail.
2. Their chain has to go around, which costs them at least one hex.

You never choose between building and blocking. The only question is which shared hex is worth most — and that is the same question either way you frame it.

A worked example

Picture a small board where your chain has reached the middle column and the bot’s chain is coming down two columns to the left of your leading tile.

You could place ahead of your own chain, continuing straight for your far rail. Or you could place in the gap the bot’s chain needs next.

Now notice that on a hex grid those are often the same cell, or two cells that touch. The bot’s route down and your route across have to cross somewhere, and the cells near that crossing do double duty. Placing there extends your chain and forces the bot into a detour in one action.

When they are genuinely different cells, the tiebreak is simple: take the one that is worth more to whoever is closer to winning. If the bot is one hex from connecting, the blocking cell is also your best building cell, because a game you lose next turn is not a game you were building in.

Why this game can never end in a draw

Fill every hex on the board, in any pattern at all, and exactly one player will have a completed connection. Never both. Never neither.

That is a proven mathematical property of connection games on this grid, not a rule bolted on to avoid stalemates. There is no tiebreak, no sudden death and no draw screen, because there is no position that needs one.

It has a practical consequence worth internalising: there is no such thing as a purely defensive game. You cannot block your way to safety, because if you have stopped the bot from connecting, you have connected. Blocking every route they have is completing a chain of your own.

The three board sizes

The board and the opponent change together.

• loose · 5×5 — a small board. The bot plays next to its own tiles most of the time and reads nothing else. Good for learning what a chain looks like.
• tuned · 7×7 — the real game. The bot works out how many tiles each side still needs and takes the cell that helps it most while costing you most.
• ruthless · 9×9 — the same eye on a much bigger board. More room means more ways to go around, and far more turns in which to go wrong.

The bot’s judgement is the same at tuned and ruthless. What changes is the board: on 9×9 a mistake in the opening is still a mistake twenty moves later.

Frequently asked

Why can this game never end in a draw? Because a completely filled board always contains exactly one winning connection. It is a proven property of the hex grid: if you have blocked every route your opponent could use to cross the board, the tiles you blocked with necessarily form a connected chain between your own two edges. Blocking perfectly and winning are the same outcome, so there is no drawn position for the game to handle.

Is blocking or building more important? They are the same move, and that is the whole idea of the game. Your chain and your opponent’s chain need to cross the same region of the board, so a hex that advances your connection is usually a hex that lengthens theirs. You are never choosing between offence and defence — only choosing which shared hex is worth the most right now.

What is a good first move? Near the centre. The middle of the board is on more potential routes than any edge cell, so a central tile does more for your own chain and sits in the way of more of your opponent’s options. Corners are the weakest opening: the two corners off your own rails contribute almost nothing to a crossing.

Does the first player have an advantage? Yes, honestly. On a symmetric board the player who moves first has a theoretical edge, and in our own simulated matches between two identical opponents the opening player won roughly seventy percent of games. Bridge does not currently alternate who starts between matches — you always move first against the bot, so the advantage is yours. Competitive play usually balances this with a swap rule, where the second player may claim the first player’s opening move as their own; Bridge does not implement that rule.

Which board size should I start on? Start on 5×5 to see how chains connect and how the rails mark your goal, then move to 7×7 as soon as you win a couple. 7×7 is the size the game is balanced around. Go to 9×9 when you want the opening to matter.

Where Bridge sits

Bridge is the deepest two-player game in the SETPOINT set, and the cleanest statement of what the whole portfolio is about: one rule, no hidden information, and a decision that does not get easier once you understand it.

If you want something with the same no-draw guarantee but a much shorter game, try overload or closure. For deduction against a hidden target, there is filter. For pure nerve, try graze or deadband.

Play bridge — it runs in the browser, free, no account.