A board for the move before the obvious one
Most Gomoku games are decided before either player can point to a threat.
The finished five gets the credit. The earlier empty intersection usually did the work: the point where two loose pairs could meet, where an ordinary extension became two questions at once, or where a defender covered the nearest stone and left the important space open.
This 15×15 freestyle board is for learning to notice that earlier moment. Play the AI when you want a calm opponent. Pass the device when someone is sitting beside you. When the position turns, undo and replay help you find the decision that mattered instead of admiring the stone that ended it.
For the player who keeps getting surprised
The danger is often an empty point, not a row of stones.
A pair of stones is rarely dangerous on its own. The useful question is whether one empty point could belong to more than one future line. Near the edge, a stone has fewer jobs available. Near the middle, it can still serve horizontal, vertical, and both diagonal plans. That is why good openings lean central: not because the centre is magic, but because it keeps choices alive.
Before every move, ask two questions: “What does this extend?” and “What else could this same point join two turns from now?” The second question catches the forks that make a loss feel sudden.
For the player who sees a three and panics
Not every row asks for the same answer.
A closed three can wait.
If one end is already blocked, a long-looking row may still be slow. Read the whole board, but do not spend every turn answering a shape that has only one way forward.
An open three changes the turn.
Two empty ends give the attacker options. Let it become an open four and one defensive stone can no longer cover both winning points.
A double threat has already won.
One stone creates two problems on different lines. The defender has one move and two required blocks. The five arrives afterward; the fork was the winning move.
For the player who wants an AI game to teach something
Play the computer for a question, not just a result.
The point of an AI opponent is not to prove that a number is “strong.” It is to give you a position worth reading at the right speed. Choose the setting that lets you see the mistake before the board tells you about it.
Use it to make the board legible.
Easy leaves room to test a plan and watch its shape develop. It is useful when you are learning what an open end or a crossing point looks like without every loose stone becoming urgent.
Use these to test your scan.
Before every move, name the opponent’s strongest shape and one shared empty point. Medium gives that routine time to form; Hard punishes the moments when the scan becomes a guess.
Use it when you want the position examined.
Master is for deliberate games, not a promise of a formal rank. Give yourself time, use replay after a forced sequence, and look for the first turn where your plan left a second direction undefended.
For the player who wants a useful post-game
Review the first forced reply, not the final stone.
After a loss, do not begin with the winning five. Go to the first move that forced an answer, then go back one more turn. Which loose pair did you leave alone? Which empty point linked two groups? That is the position worth carrying into the next game.
Use Hint as a second opinion, not an oracle. It can name a candidate point; the board teaches why it matters. Start at Medium until you can name the opponent’s most advanced shape before you play. Move to Hard when that scan feels automatic. Save Master for the games where you want every loose crossing examined.