Solving Logic-Grid Puzzles Systematically

Five people sit in a row; each has a distinct favourite drink, pet, and profession. Clues like "the coffee drinker sits immediately left of the trader," "the person with the cat is not at either end," and "the analyst drinks neither coffee nor tea" are given, and the puzzle asks who has the fish. Logic-grid puzzles look intimidating because there are dozens of possible combinations to track, but they are entirely mechanical once you build the right bookkeeping structure — no flash of insight is required, only careful, exhaustive elimination.

The core idea: a grid turns "who could have what" into a solvable constraint satisfaction problem

Every pair of categories (person × drink, person × pet, drink × profession, and so on) gets its own small grid of yes/no/unknown cells. Each clue is translated into one of two operations on these grids: an elimination ("the analyst doesn't drink coffee" marks that cell impossible) or a link ("the coffee drinker is the trader" marks a cell certain, which then lets you eliminate every other cell in that row and column). The puzzle is solved by applying every clue's direct implication, then repeatedly propagating: whenever a row or column has only one remaining possible cell, that cell becomes certain, which triggers new eliminations elsewhere, cascading until the whole grid is filled.

Working example: three people, three drinks, one relational clue

Amy, Ben, and Cara each drink a different one of coffee, tea, or juice. Clues: (1) Amy doesn't drink coffee. (2) Ben doesn't drink juice. (3) The tea drinker's name is alphabetically before the juice drinker's name.

Start with a 3×3 grid of all nine (person, drink) cells marked "possible." Apply clue 1: eliminate (Amy, coffee). Apply clue 2: eliminate (Ben, juice). Now check each row: Amy can still be tea or juice; Ben can still be coffee or tea; Cara can still be all three (no direct elimination on Cara yet).

Apply clue 3 by testing consistency rather than direct elimination: if Amy drank juice, the tea drinker would have to be alphabetically before "Amy" — impossible, since Amy is first alphabetically among the three. So eliminate (Amy, juice). Now Amy's row has only one possibility left: (Amy, tea) becomes certain. Propagate: since Amy has tea, eliminate (Ben, tea) and (Cara, tea) — the tea column is now used up. Ben's row now has only (Ben, coffee) left (juice was already eliminated, tea just got eliminated) — certain. Propagate again: eliminate (Cara, coffee). Cara's row now has only (Cara, juice) left — certain. Check clue 3 against the finished grid: tea drinker is Amy, juice drinker is Cara, and "Amy" is alphabetically before "Cara" — consistent. Solved: Amy–tea, Ben–coffee, Cara–juice, entirely by elimination and propagation, no guessing.

What this means in practice

The grid-and-propagate method is a concrete rehearsal of constraint satisfaction — the same technique used to reconcile trade breaks against multiple possible explanations, or to narrow down which of several data sources is stale by cross-checking a handful of consistency constraints. Interviewers use these puzzles to check whether a candidate can build and maintain external bookkeeping (the grid) rather than trying to hold the whole deduction in their head, which is exactly the discipline needed when debugging a live system under time pressure with many interacting constraints.