Strategy

Meowdoku Opening Moves: Scan Regions, Not Rows

A new Meowdoku board looks like a wall of color with one cat on it. Most players start by reading across the rows, because that is what every other logic puzzle trains you to do. Measured across four hundred boards per difficulty, that is the worst of the three scans available — and the first two or three cats need no scan at all.

By Meowdoku Playbook Editor·Published August 23, 2026·Updated August 23, 2026

Two or three answers are already on the board

Before you place anything, a Meowdoku board has told you some of its cats. One is the starter cat, drawn on the board when the puzzle loads and locked there so a stray tap cannot take away your head start. The others are the one-cell regions.

A one-cell region is a color that occupies exactly one square. Every region holds exactly one cat, so that square is a cat and there is nothing to work out. No board ships without at least one of them; 8×8 and larger normally get two, and Extreme is the one difficulty that deliberately takes one back, running its 10×10 board on a single one-cell region instead of a pair.

That is two free cats on Easy, Normal and Extreme, and three on Hard, before you have thought about anything. Collect them first, every time. The rest of the opening is about what they eliminate.

What those first cats already kill

A cat rules out its entire row, its entire column, its entire region, and the eight squares touching it. Those overlap heavily, but the total is still most of the board: after the opening alone, 60% of the cells on a Hard 9×9 are illegal, and on Easy it is 67%.

This is bookkeeping worth delegating. The Auto button crosses out every cell a placed cat has ruled out, using the same test the game uses to reject an illegal move — so an auto-crossed cell is never one you could have used. With it on you see the shape of what is left instead of re-deriving it every time you look up.

What is left is the interesting part. On a Hard board just over a third of the cells are still open, spread across six regions, six rows and six columns. The only question in the opening is which of those three to scan.

Scan regions, not rows and columns

The basic Meowdoku deduction is the one Sudoku teaches: find a line with exactly one legal square left, and that square is a cat. Meowdoku gives you three kinds of line — the row, the column and the color region — and reading habits put them in that order.

On this ruleset the order is backwards. Measured over four hundred freshly generated boards at each difficulty, this is how often each kind of line is sitting at exactly one legal square the moment the opening is done:

On a Hard board a region is eighteen times likelier than a row to be handing you a cat. On Extreme, rows and columns produce essentially nothing at this stage and every deduction on offer is a region. Only on the 5×5 board do the three genuinely compete, which is precisely why habits formed on Easy mislead people on Hard.

The same average, three times the spread

The reason is arithmetic rather than anything subtle about cats, and it is worth following, because it also tells you when the advice stops applying.

All three scans look at the same cells. A Hard board after the opening has six open rows, six open columns and six open regions, and every legal cell left belongs to exactly one of each. So the average number of candidates per open line is the same three times over: 4.87. An average cannot separate them.

The spread can. Every row is nine squares wide and every column nine tall, so their candidate counts cluster near that average — a standard deviation of 1.1. Regions have no fixed size; on a Hard board they run from one cell to about sixteen, and their counts scatter with a standard deviation of 2.9, nearly three times wider on the same mean.

A forced cat is a line that has landed on exactly one candidate. That is a tail event, and the wider distribution reaches its tail far more often. Rows and columns are not harder to read. They are just too uniform to fall off the end.

Smallest region first

Inside the region scan there is an obvious order, and it survives measurement. When a region is forced right after the opening, it is the smallest region still open 61% of the time on a Hard board, and on average it ranks 1.7th smallest of the six still open. Easy, Normal and Extreme all land in the same place: about 60%, rank 1.5 or 1.6.

So sort the open regions by eye, smallest first, and walk them in that order. The deduction is usually in the first or second one you check, and you skip the twelve- and sixteen-cell regions that are almost never ready this early.

That ordering is not an artefact of how the boards are built. A small region is compact, so one cat placed a couple of squares away can cut through it by row, by column and by diagonal contact at once. A nine-square row spans the whole board, and no single cat comes near most of it.

What the free cat is actually worth

The starter cat reads as a courtesy. It is not. Below is how far a board gets on nothing but the three single-candidate scans, applied over and over until none of them fires — first with the free cat, then with the same boards and the free cat withheld.

On Easy the free cat is the difference between a board that unravels by itself nine times in ten and one that does so barely a third of the time. On Normal it is more than a fivefold difference. One cat out of five, or of seven, is not doing one cat’s share of the work.

It compounds. Everything the starter eliminates makes some other line shorter, and a shorter line is a line closer to sitting at one candidate. That is also why the game never puts the free cat in a one-cell region: spending it on a square you were going to be given anyway would waste the entire cascade.

Where the opening runs out

The middle column of that table is the honest part of this article. On a Hard 9×9, single-candidate elimination places about five of the nine cats and then stops. On Extreme it places three of ten and stops.

That is the technique’s ceiling, not yours. When the three scans go quiet the board still has exactly one solution and is still fully determined; it just needs a different kind of reasoning. That a region’s remaining squares all lie in one row, so no other region may use that row. That two regions are confined to the same two columns, so everything else in those columns is dead. Those are their own articles.

What the opening buys is a board where that harder reasoning has something to bite on. Five cats on a nine-cat board is not a stall. It is the position the rest is findable from.

The opening, in order

Five steps, and the first four take about fifteen seconds once they are habit:

• Take the free cats. The starter is already placed. Add one for every region that is a single square.
• Turn on Auto. Let the game cross out what those cats ruled out. It uses the same test that blocks an illegal move, so it never crosses out a square you needed.
• Scan the regions, smallest first. Any region down to one open square is a cat. This is where nearly every early deduction comes from.
• Re-scan after every placement. One cat frequently forces the next two, and sweeping the board once leaves most of that cascade unfound.
• Only then try rows and columns. Worth one pass when the regions go quiet. On a 9×9 or 10×10 they will usually find nothing until the board is well advanced.

How these numbers were measured

Every figure here comes from the game’s own generator — the same code that builds the boards you play — run across four hundred fresh boards at each of the four difficulties. Nothing is estimated, and nothing is drawn from a sample of human play.

“The opening” means exactly the starter cat plus every one-cell region and nothing else. “Elimination” means only the three single-candidate scans described above, repeated until none fires. A player who uses anything more advanced gets further than the table shows, which is what “Where the opening runs out” is about.

The figures describe the boards the generator currently produces, and it has been retuned before. So the test suite that ships with this article re-derives the central comparison from freshly generated boards on every run: if regions ever stop out-producing rows, the page stops building.

Try the opening on a real board

Start a Hard 9×9, collect the three free cats, turn on Auto, and work the regions from smallest upward. Count how far you get before the scans go quiet: around five cats is exactly where the measurements say you should be.

Boards are generated in your browser. Nothing about a game you play is sent anywhere.