Why random shooting loses — and what to do instead.
The game looks like pure luck. It isn't. Between a player who fires at random and one who searches systematically there is a gap of dozens of shots per match — and in Hard mode, where ammunition runs out, that gap decides the game outright.
The most common mistake is firing at neighbouring cells to "cover an area". It's waste. The smallest craft in the game, the boat, occupies 5 cells within a 3×3 box. Nothing in the game fits into a space smaller than 3×3.
So there is no reason to fire at two cells one square apart: if something were there, a single shot would have found it. Search on a lattice, every third square, until you hit something. You cover the same ground with a fraction of the shots.
The moment you land a hit, the game changes completely. You are no longer searching — you found something and must finish it, especially since a hit keeps your turn.
From that point, fire around the hit rather than in some far corner of the map. Every craft has its cockpit somewhere inside it, and craft are compact: the cockpit is close to what you just struck. A player who finds a hit and then wanders off throws away the very advantage they just earned.
Your hits tell you what you found. Two hits on the same row one cell apart look nothing like two hits on a diagonal. Shapes are fixed and there are only six of them — after two or three hits you can usually tell which craft it is and roughly where its cockpit should sit.
It pays to study the shapes before a match. Whoever knows them by heart fires their third shot where someone else fires their seventh.
This is where matches are won defensively. Most players place their craft in the middle, because "there's room there". The result: the middle is the first place everyone searches.
In Medium mode you get a single 3×3 scan for the whole match, and using it costs you your turn. The temptation is to spend it in the opening "for a head start". That's waste: early on the grid is empty and every region is as likely as any other.
Radar is worth most when you already hold partial information — a craft hit but not located, a region you suspect. Then the scan confirms or kills a hypothesis, which is genuinely worth a turn.
140 shots sounds generous on a 20×20 grid — but the grid holds 400 cells, so you cannot cover even half of it. Random shooting loses mathematically in Hard mode; it isn't bad luck.
Here lattice search isn't advice, it's a survival condition. More importantly, every curiosity shot into an area you've already cleared is a shot you won't have at the end, when you need it to finish the last craft.
The computer opponent has three levels. The strongest doesn't fire randomly: it searches systematically and, once it finds something, pursues the target with priority on the cockpit — exactly what's described above. On a 10×10 grid it finishes a match in roughly 17 shots.
To beat it, play the defence: maximum spread, edges, never two craft side by side.