Everyone on the grid is secretly a criminal or an innocent, and exactly one line-up fits every statement on the board. You work out which, one provable verdict at a time. What separates Griductive from most deduction games is the thing it refuses to let you do: you cannot guess. Not because guessing is discouraged — because the board will not accept a verdict you have not earned.
Written by Shawn Liu, founder & puzzle designer.
Sweep, don’t stare. New solvers pick an interesting-looking person and try to reason them out; that person is usually not the one the clues have already decided. Every grid always has at least one verdict sitting there fully forced, so the opening question is never “who is guilty” — it’s “who has the board already told me about, whether or not I noticed?”
Those are the openers. The five patterns that actually force a verdict — and the difference between “neighbours” and “connected”, which decides more grids than anything else — are on the strategy page. The clue glossary, the ground rules and a grid walked through end to end are on the how-to-play page. The interactive tutorial runs it on a real, small grid.
Most logic games let you mark whatever you like and tell you at the end whether you were right. Griductive does something stricter. When you call a verdict, the board works out whether the clues on the table actually force it — and if they do not, it declines the call and says so.
It also separates the two ways a call can be refused, which turns out to matter enormously. Either the verdict is not yet provable, in which case you could not have known and nothing is your fault; or the verdict is provable and you called it the wrong way, which means the inference is sitting there and you have misread something. Neither outcome locks the card, ends the run, or costs you anything. The rejection is the feedback.
Knowing what is forced is not cheap. The board answers it by enumerating the solutions that remain consistent with the clues, which is exponential work, so it runs off the main thread in a background worker — the grid stays responsive while it thinks, and a hint never comes back empty. That machinery exists so that the no-guessing rule is a property of the game rather than a promise on a page. The rule itself is written up in no-guessing puzzles.
A new grid drops every day at midnight Eastern, on a ladder that runs easy on Monday, medium through midweek, tricky on Thursday and Friday, then hard on Saturday and outright evil on Sunday. The full range unlocks on 1 August 2026 — before that the ladder is capped at tricky, so the opening weeks never open brutally while people are still learning to read the clues.
The last seven days stay free in the archive. Grids older than that, back to launch in June 2026, come with a subscription. If you like the deduction but want it set in a house full of suspects rather than on a grid, Alibi is the same instinct in a different room.