Mr. Grummel Get the app
← All notes
LEARNING 6 MIN READ DRAFT — SEPTEMBER 2026

The question computer science still can't answer

Checking a finished sudoku takes seconds. Nobody has ever found a fast way to solve one from scratch — and nobody has proven a fast way is impossible either.

Hand someone a completed sudoku grid and ask them to check it's correct, and they can do it in a couple of minutes by scanning rows, columns and boxes. Hand them a blank grid and ask them to solve it, and there's no known shortcut — every fast method anyone has ever found eventually falls back on some amount of trial and error that grows explosively as the grid gets bigger. That gap, between how fast a solution can be checked and how fast one can be found, is the entire content of the most famous open problem in computer science: is P equal to NP?

Two different kinds of "fast"

P is the set of problems a computer can solve quickly — in a time that grows manageably as the problem gets bigger. NP is the set of problems where, if someone hands you a proposed solution, you can verify it's correct quickly, even if finding that solution in the first place might take far longer. Every problem in P is automatically in NP, because if you can solve something quickly, you can obviously also check a proposed answer quickly. The open question is whether the reverse is true: does the ability to verify a solution fast always mean a fast way to find one exists, even if nobody has found it yet?

Why one problem's difficulty stands in for thousands

In 1971, Stephen Cook proved something unexpected: a specific logic problem called Boolean satisfiability is "NP-complete," meaning every other problem in NP can be systematically translated into it. If anyone ever found a fast algorithm for satisfiability, that algorithm — combined with the translation — would instantly solve every problem in NP quickly, collapsing the distinction between P and NP entirely. Richard Karp extended the result the following year, showing dozens of everyday problems — scheduling, route planning, packing, graph colouring, and yes, sudoku — are all NP-complete too, secretly the same problem wearing different clothes. Solve any one of them fast, in general, and you've solved them all.

Every computer scientist believes P doesn't equal NP. Nobody has proven it. Belief and proof are not the same currency.

What we're still unsure about

Most researchers strongly suspect P does not equal NP — that a genuine gap exists between checking and solving, and it isn't just a failure of imagination. That suspicion rests on decades of nobody finding a fast general algorithm despite enormous effort, not on a completed proof. The Clay Mathematics Institute lists it as one of seven Millennium Prize problems, with a one-million-dollar reward for a correct proof in either direction, and at least one high-profile attempted proof — Vinay Deolalikar's 2010 paper — drew intense public scrutiny before the mathematical community found a fatal flaw in it. The honest state of the field is that P vs NP remains genuinely open, which is a different claim than "everyone knows the answer and just hasn't written it down yet."

This sits inside NP-Completeness & Reducibility, one of eight topics in Algorithms, one of seven domains in Computer Science, one of seventeen subjects the app can quiz you on.

Draft — not published yet.
Try the pop quiz