In a standard British cryptic grid, roughly half the letters of any answer are also letters of some other answer. Wikipedia states it flatly: unlike American crosswords, where nearly every square is checked, only about half the squares in a cryptic crossword are checked. Most people read that as a fact about aesthetics, about white space and how much black a grid can carry. It is better understood as an error-detecting code, drawn in advance, sitting under the puzzle the whole time you solve it.
Write a wrong word into a grid and the grid tells you. Not immediately, and not in its own voice, but eventually, through a clue you have not reached yet.
Minute Cryptic gives you one clue a day and no grid at all.
What it is
Minute Cryptic began as short explainer videos, one cryptic clue taken apart per video, posted by Angas Tiernan on TikTok and Instagram until enough people asked for something to play. The Australian setter Liam Runnalls came on as clue writer. The product is one clue, no grid, free, with a video explaining the wordplay once you are done.
The numbers are not small. An April 2025 case study by Will at Hold! recorded 64,000 daily players as of September 2024, about 20,000 YouTube views a day on the explainers, roughly 264,000 Instagram followers, and around 4,000 paid subscribers at $9 a month. By the time the founders spoke to Palatinate in June 2026, Instagram was over 300,000. Whatever else it is, it is the largest front door the cryptic has ever had.
Three routes, and the one that was cut
A cryptic clue is already redundantly encoded. Ximenes' formulation, quoted everywhere: a good clue contains a precise definition, a fair subsidiary indication, and nothing else. Two independent paths to the same word, which is why the genre feels like proof rather than trivia. You arrive at the answer twice by different means and the two arrivals agree. Afrit's older injunction says the same thing from the setter's side: you need not mean what you say, but you must say what you mean.
The grid adds a third path. Crossing letters get filed under hints, or under scaffolding, which undersells them badly. They function as a parity check on the other two, running in both directions, so that every answer you commit is simultaneously testing and being tested by its neighbours.
Which means a clue inside a grid does not have to be individually sufficient. It can be ninety percent determinate. Thirty clues at ninety percent, plus crossings, converge on a single consistent filling, and any one of them can be locally underdetermined without the puzzle failing. The grid is a system that tolerates loose clues because it resolves them globally.
Take the grid away and every clue has to stand on its own two encodings, alone, with no third opinion.
The cost, described by the person paying it
Runnalls told Palatinate that having one clue instead of the usual thirty forced him to "further perfect each one."
I have gone back to that sentence several times. It is offered as craft pride, the pleasure of lavishing attention on a single object. It is also a specification change, stated precisely and without alarm. In a grid, ninety percent determinate is fine. Standing alone, ninety percent is a coin toss between two readings with nothing available to break the tie, and the solver who picks the wrong one has no way to discover it and no reason to look.
So the clue has to absorb the work the crossings used to do. Every day, in one object. There is no easy neighbour to hide behind, which is the other thing the grid was quietly providing: a difficulty budget spread across thirty items, letting a setter bury one savage clue among twenty-nine merciful ones. On a one-clue-a-day schedule the difficulty of the puzzle is the difficulty of the clue. Nothing is distributed. Everything is load-bearing.
Tiernan's observation that controversial clues reliably generate the most engagement follows from this rather than sitting beside it. When there is only one clue, an argument about whether it is fair is an argument about the whole puzzle.
The checked square, sold back one at a time
Minute Cryptic has a hint ladder. You can reveal a nudge, and you can reveal a letter.
Look at what a revealed letter is. It is a checked square. The grid used to hand you those for nothing, in a pattern fixed before you sat down, whether you wanted them or not, arriving as a consequence of solving something else. Here the same information exists, but it has been unbundled from the structure and turned into a decision you make about yourself, in the middle of the solve, while your ego is watching.
I do not think this is cynical. It might be the more honest arrangement, since a grid never asks whether you would rather have worked it out unaided, it simply fills your squares in. But it converts an automatic structural gift into a confession, and those are not the same experience even when the letter delivered is identical.
Where the confirmation went
It survived the removal of the grid. What changed is when it arrives: afterwards.
The explainer video is genuinely the best thing about the product, and it is doing the job the crossings used to do, one step too late to change anything. In a grid, confirmation arrives during — a down entry lands its letter into an across answer you wrote ten minutes ago and the two agree, mid-flight, while you are still working. The verification is interleaved with the solving. On a single clue, you finish, and then you are told.
Which also means the grid gave you somewhere to be provisionally wrong. Pencil in a plausible mistake, carry on, and three clues later something contradicts it. The error has a location and a moment of discovery, and the discovery is itself a solving move. Being wrong inside a grid is an event with a shape. Being wrong on a single clue is silent. You simply stop, satisfied, and the video arrives later to tell you what you were doing.
I would like to know whether solvers who learned this way transfer to a full grid — whether trusting a crossing letter is a skill you can only build by having had to lean on one, or whether it turns out to be nothing much, a convenience you pick up in an afternoon once the two hard encodings are already in your hands. Three hundred thousand people are running that experiment. Nobody, as far as I can find, is measuring it.