Why a Good Logic Puzzle Never Needs a Guess

Here's a puzzle. It takes about a minute. Put one star in every row, every column and every colored region. Stars can't touch, not even at a corner.

If you tried it, you probably started by putting a star somewhere that felt right, saw what broke, and moved it. That works on a board this small. It stops working fast. And here's the odd part: you never needed to do it at all.

A well-made logic puzzle never needs a guess. That isn't a claim about how smart you are. It's a property of the puzzle itself, and once you see why it's true, you solve differently. Let's break one to find out.

A puzzle you can't solve

This board looks almost the same as the last one. It follows the same rules. But it has two correct answers. Flip between them.

Both answers keep every rule. So when you're sitting at this board, trying to decide where the star in row 2 goes, no amount of thinking will help. There's no reason hiding on the board to find. Any argument that rules out Answer A would also have to be false, because Answer A is fine.

The only way through is to pick one. That's what a guess really is: a choice the puzzle has given you no reason to make. Setters call this a broken puzzle, and it's the one thing a good one must never be.

One answer is a promise

Now flip that around. If a board has exactly one answer, every other arrangement of stars breaks a rule somewhere. That means every wrong square is wrong for a reason, and that reason is sitting on the board, waiting. So you can stop asking "where might a star go?" and start asking a better question: "which squares can't hold one?"

Here's the first puzzle again, solved with no guessing at all. Step through it and read why each move is forced.

Notice how much of that is crossing things out. Most of the work happens on squares that will end up empty. Stars are almost a side effect: once a row has one open square left, you've already done the thinking.

What guessing costs

So how much does reasoning actually save? Let's race. On the left, a computer that does nothing but guess: it tries a star in each row, checks the rules, and tries the next arrangement. It's fast and very stupid. On the right, the number of moves you just stepped through.

That's a 6×6 board with one star per line. The real ones get much worse: a 10×10 Star Battle has two stars per row, so blind guessing has more arrangements to wade through than anyone could check by hand. Logic scales. Guessing doesn't.

One square changes everything

Here's the part that surprises most people. Uniqueness is fragile. The broken board from earlier differs from the good one by a single square: row 4, column 4 changed color.

This board is live. Pick a color, tap a square to repaint it, and watch the answer count change. See how few squares you can move before it breaks. Then try to fix it.

That's why nobody makes these by eye. Every board we publish goes through a solver that counts its answers and throws it out unless the count is exactly one. If you've got a board you're not sure about, from a book or an app, the Star Battle solver will run the same check and walk you through it.

But sometimes you do feel stuck

You will meet a hard board where nothing obvious is left. Every row has three open squares and no region is close to done. That doesn't mean it's time to guess. It means the reason is further away.

At that point strong solvers look ahead: "if a star went here, this region would have nowhere left to go, so a star can't go here." It looks a lot like guessing from the outside, but it's the opposite. A guess hopes. Lookahead proves. You try the idea in your head until it breaks, and the break is a fact you get to keep.

Harder puzzles aren't harder because they ask you to take bigger risks. They're harder because the reasons are longer. Our strategy guide walks through those longer patterns, and this post covers how the no-guessing guarantee works on Star Battle specifically.

Where you've seen this before

The board on this page uses the rules of Queens, the game on LinkedIn. Give it two stars per line instead of one and you've got Star Battle, which the New York Times prints as Two Not Touch and Inkwell Games calls Stars. Fields, Sudoku and nonograms all work on the same deal.

That deal is what makes these puzzles feel fair. Somebody built the board so that the answer is there, every step of the way, for anyone patient enough to find it. You're never being asked to get lucky. You're being asked to look harder.