Fundamentals
What logic grid puzzles are
You will have seen them under many names: Einstein riddles, zebra puzzles, logic problems, logic grid puzzles. They all describe the same creature: a problem where several categories are matched against each other and a handful of clues is enough to reconstruct the exact combination.
The mechanics, in one sentence
You have a set of positions — five houses, four columns, six days of the week — and several categories distributed across them. Each element of each category occupies exactly one position. You do not know which. The clues do not hand you positions directly: they hand you relationships.
"The doctor lives next to the person with the cat" does not tell you where either of them is. What it does is eliminate every combination in which they are not neighbours. And that is the key to the whole genre: each clue is a filter, and solving consists of applying filters until only one combination survives.
Where they came from
The riddle everyone knows as "Einstein's" was published in Life magazine in 1962, with no attribution. A later version that appeared in Scientific American included a zebra among the pets, which is where the alternative name "zebra puzzle" comes from. The Einstein connection is folklore: there is no document linking him to it, and the line about "only 2% can solve it" is a promotional addition that spread by email in the nineties.
The format predates those publications. Pastimes that matched categories using clues were circulating in crossword magazines from the 1940s, and the grid of ticks and crosses used to solve them with a pencil was standardised around then. It is a genre born on paper, and that explains many of its conventions.
How it differs from sudoku
It is the inevitable comparison, and there is an underlying difference that most people never quite put into words.
In sudoku, all the information is on the board. The rules are fixed and you know them by heart: each number once per row, per column and per region. What varies from puzzle to puzzle are the starting numbers. Solving means looking at the grid and finding the gap where only one digit fits.
In a logic puzzle, the board starts empty and the information arrives from outside, in sentences. Each clue brings its own rule, different from the others, and the main mental work is not searching the grid: it is translating. Turning "the mill and the cow share a column" into a set of crossed-out cells.
That is why many people who burn out on sudoku take to these instead: the effort feels closer to interpreting witness statements than to filling in a table. And it is also why they are harder to get started with — you have to learn to read the clues — but less repetitive in the long run.
What makes one of these puzzles good
They can be generated by the thousand with a program. Making them good is another matter, and it comes down to three properties.
A unique solution
This is the bare minimum. If two different combinations fit all the clues, the puzzle is broken: the player can arrive at a correct answer and be told it is wrong.
Solvable without guessing
This is the property that separates a good puzzle from a frustrating one, and the one most often neglected. There are puzzles with a unique solution that you can only reach by assuming something, following the chain and checking whether it collapses. Technically they have a single answer; to the player, they are indistinguishable from luck.
Verifying it requires writing a solver that only makes direct deductions, with no hypotheses, and checking that it reaches the end. If it stalls, the puzzle is discarded. It is an expensive step that many automatic generators skip.
No redundant clues
A puzzle with redundant clues is easier, but it is also duller: there are too many routes and none of them feels necessary. Good generators make a final pass removing clues one at a time and checking whether the puzzle is still solvable. What remains is the minimal set, where every clue genuinely earns its place.
Why they work so well as a game
There is a specific psychological reason: progress is irreversible and visible. Every cell you cross out is a conclusion you will never have to revisit. In a game of skill you can lose ground; here you cannot. That clean accumulation of certainties produces a sense of advancement quite unlike almost any other genre.
And there is a second, subtler reason: the cascade moment. There comes a point where you confirm one element and it unlocks another, which unlocks another. You made one move and the board solves itself for a few seconds. That instant is the hook, and it is completely different from the relief of finding a number in a sudoku.