Why "puzzle" is the least useful word on a store listing
Of all the category labels an Android game can carry, "puzzle" tells a reader the least. It covers a game where you clear coloured tiles with your thumb while waiting for a kettle, and one where you write a small program to route a robot through a factory. Those two share a genre tag and nothing else — not the session length, not the failure state, not the kind of thinking they reward.
So our desk stopped treating puzzle as a genre and started treating it as a shelf with six families. What follows is the map we use internally when a candidate enters the queue. The borders are soft, but it gets you to the right shelf faster than a store page will.
The six families at a glance
| Family | Core demand on the player | Typical session | Fails you by |
|---|---|---|---|
| Match-and-clear | Fast visual pattern recognition | One to five minutes | Repetition, difficulty walls |
| Physics puzzles | Estimating force, angle and consequence | Two to ten minutes | Imprecise aiming, unclear rules |
| Logic grids | Pure deduction from fixed constraints | Ten to forty minutes | Guessing being viable |
| Tile-sliding and rearrangement | Planning moves under spatial limits | Three to fifteen minutes | Fiddly touch targets |
| Escape-room | Observation and lateral connection | Twenty minutes to hours | Pixel-hunting, arbitrary logic |
| Programming puzzles | Abstraction and debugging | Half an hour and up | Poor feedback on what went wrong |
Match-and-clear: recognition, not reasoning
The clearing family asks one thing: see the pattern before the timer or the move counter runs out. Good examples are honest about that — a board you can read in a glance, a clear reason why a move is legal, and a visible consequence when one clear cascades into another.
This family dominates phone screens for a mechanical rather than mysterious reason — it survives interruption. Testers rate it well when the board stays readable on a smaller handset and when difficulty rises through new board rules rather than through fewer moves. It rates badly when progress stalls at a designed wall and the only variable is patience.
Who it suits: readers who want a game that fits into gaps in a day and never demands a plan.
Physics puzzles: aim, release, watch
Here the puzzle is not on the board, it is in your estimate. You judge a force, an angle or a cut, you commit, then watch a simulation resolve without you. The pleasure is the half-second of suspense between release and result.
What separates a good one is tolerance. A physics puzzle that only accepts a single perfect input is a reflex test wearing a puzzle costume. The ones we rank highly accept a band of near-correct answers, so a solution you understood but executed loosely still lands. Their rules stay consistent enough that your second attempt is informed by your first rather than another guess.
Who it suits: readers who like being wrong in an interesting way, and who would rather feel a rule than read one.
Logic grids: the deduction family
Logic grids — number placement, cell-marking picture logic and constraint satisfaction of every stripe — are the strictest of the six. There is exactly one solution, reachable by reasoning alone, and no part of it should ever require a guess. That last clause is the whole quality test: when a tester can only move forward by trying a value and seeing whether the board contradicts itself later, we mark the puzzle broken, not hard.
The good ones are also generous about bookkeeping. Pencil marks, an undo that goes back more than one step, and a highlight showing which constraint a cell belongs to are not hand-holding. They remove clerical work so the reasoning stays the difficulty.
Who it suits: readers who want a quiet, sustained sitting and dislike time pressure of any kind.
Tile-sliding and rearrangement: working inside a cage
The rearrangement family gives you a small space, fewer free positions than pieces, and a target configuration. Everything interesting comes from the constraint: to move the piece you want, you must first move three you do not.
This family lives or dies on touch. A one-piece mis-drag in a fifty-move plan can sour a whole session, so our protocol pays unusual attention to target size, to whether a piece can be nudged by a tap rather than a drag, and to whether undo is instant and unlimited. The strong entries teach a technique in an easy layout, then demand you apply it under a tighter cage — the cleanest kind of difficulty curve there is.
Who it suits: readers who enjoy planning several moves ahead and like a puzzle that fits on one screen.
Escape-room: observation and connection
An escape-room puzzle hides its rules inside a scene. You look, you notice, you connect a thing in one room to a lock in another. At its best it produces a feeling the other five families cannot: the moment when something you glanced at twenty minutes ago suddenly means something.
The characteristic failure is equally specific: when the connection is arbitrary rather than inferable, the game stops being a puzzle and becomes a search. Our marker is simple: after being told the answer, could a reasonable player say "I should have seen that"? If the honest answer is no, the design has failed.
Who it suits: readers who want atmosphere with their thinking and do not mind a slow start.
Programming puzzles: the abstraction family
The smallest and most demanding family asks you to build a procedure rather than an answer. You assemble instructions, run them, watch them fail, then read the failure. That final step is the genre — these games are debugging exercises with friendly graphics.
They reward a phone least in screen space and most in session flexibility, since a half-built solution is a perfectly good place to stop. We look for feedback quality: a visible execution pointer, controllable speed, and an error state that shows where the logic diverged rather than simply reporting a loss.
Who it suits: readers who like systems, and who are not discouraged by a screen full of their own mistakes.
Reading the shelf
Once you know which family you are in, most of the question answers itself. A reader who wants five-minute sessions should not be talked into a programming puzzle, and one who wants a long quiet evening will find the clearing family thin by the third hour. Our per-family criteria and their weightings are set out on our methodology page, the current standings sit on the ranking, and the rest of the genre guides are indexed on the blog. The editorial score The Night Notebook assigns is our own judgement rather than the Play Store star average, and on puzzles in particular the two often disagree — a game can be widely liked and still be guessable, and we say so.