Battleship vs Computer: How the AI Hunts (and How to Hide)

Direct Answer: A Battleship computer hunts in two phases: hunt mode (checkerboard probing that finds ships fast) and target mode (sinking a found ship by testing neighbors, then finishing along its axis). Beat it by breaking up your fleet's geometry — never touch ships diagonally-adjacent, hug edges on Hard, and scatter so no two ships share a hunting parity. Placement is 80% of the game.

Hunt Mode: The Checkerboard You Cannot See

Strong AI does not shoot randomly — it fires on a parity pattern (like the dark squares of a chessboard), because every ship of length 2+ must cover one. That halves its search. Your counter from probability targeting: there is none in placement — parity hits everything eventually. The defense is making what it finds worth as little as possible, which is a placement problem (see best ship placements for the three families).

Target Mode: How It Sinks What It Finds

First hit → the AI shoots the four neighbors → second hit reveals the axis → it walks both directions until the ship sinks, then returns to hunting. Two exploits: ships that touch (even diagonally) merge into one long target — the AI sinks both without re-hunting. And a lone hit with all misses around it still costs you tempo while it confirms the miss. Never let ships touch. Ever. Our win-every-time guide makes separation rule #1 for exactly this reason.

The 3 Placement Rules That Survive Hard

1. Zero adjacency — one full square of water around every ship, diagonals included. 2. Split parities — put some ships mostly on "dark" and some on "light" squares so no single hunting pattern sweeps your fleet. 3. Edge-anchor the big ships — edges have fewer neighbor squares, so target mode gets less information per hit. Then test layouts against Hard using the program in placements that survive Hard.

Difficulty Ladder: When to Step Up

Do not start on Hard — you will learn nothing except losing. The climb that works: Easy until you win 3 straight with zero-adjacency placements; Medium until you win while trying a new layout each game (this forces real placement understanding instead of one memorized fleet); Hard with a loss journal — note which ship gets found first every game, because your most-found ship is misplaced.

The pattern in your journal is your curriculum. Carriers found early means you anchor big ships too centrally; destroyers living to the end means your scatter is already good and your hunting needs the work — steal the machine's checkerboard from the section above and run it yourself.

Attacking Back: Hunt Like It Does

Steal its algorithm: open with your own checkerboard (every other square), and when you hit, finish the axis before resuming the pattern. Track remaining-ship lengths — a 4-wide gap cannot hold the battleship, so stop probing it. Patience beats brilliance: players who finish axes before chasing new contacts win far more than "intuitive" shooters.

Conclusion — Hide Better, Hunt Smarter

The computer is not psychic; it is systematic. Break its system with separated, parity-split, edge-anchored placements — then turn its own checkerboard against it. Do both and Hard stops feeling hard.

Your Next Step

Rearrange your fleet with zero adjacency tonight and play 3 Hard games. Count how many extra turns each ship survives — that number is your placement grade.

---