Flood

How Flood works

Flood is a puzzle of coloured territory. Your area begins as the single cell in the top-left corner. Each turn you pick a shade, your whole area changes to it, and any touching cells already wearing that shade are swallowed into the blob. Turn the entire board into one tone before the move counter runs out and you win. The rules take ten seconds to learn; the move limit is where the game actually lives, because it is tight enough that two or three lazy picks in the middle of a run are usually fatal.

The greedy trap

The obvious tactic is to grab whichever shade adds the most cells right now. That is a fine reflex and a mediocre plan. Cell count measures the past, how much of the board happened to border you already; what wins games is frontier, how much new border a pick opens for the turns that follow. Suppose grey would absorb six cells that form a dead pocket along the top edge, while black absorbs four that stretch your border toward the far corner and touch three fresh regions. Black is the stronger move despite the smaller bite. Strongest of all are the picks that merge two separated pools of one shade into your blob at once, because merged regions bring their entire perimeter with them, and perimeter is future growth.

Count the bands

Here is a trick that turns vague anxiety about the move limit into arithmetic. Trace any path from your corner to the bottom-right cell and count how many times the colour changes along it. Your blob adopts one shade per turn, so each of those bands costs at least one move: the count is a hard floor on what you still need. If the counter shows nine moves and your best path crosses eight bands, there is exactly one move of slack left in the whole game, and spending it on a two-cell nibble is how you lose. Check a couple of different routes, take the smallest count as your true budget, and let it tell you how greedy you can afford to be. When budget and bands are equal, every remaining pick has to clear a band, no exceptions.

Openings and endings

Openings reward shape. From the corner you want to grow as a broad wedge hugging both the top and left edges, because a blob spanning two full edges touches the maximum number of regions at once; a thin snake wandering into the middle looks fast and plays slow. In the endgame the logic flips from expansion to bookkeeping. Count which shades survive outside your blob. If only two remain, your last moves are forced. If three remain, order matters: absorb the two smaller ones in whichever sequence lets the final pick sweep the whole remainder in one stroke, and you will save a move that greedy play would have burned ten turns earlier.

Flooding tips

• Judge picks by the new frontier they open, not the cells they grab; the far corner is the finish line.
• Prefer the shade that merges two clusters into your blob over the one with the biggest bite.
• Push toward the bottom-right early; it is almost always the last region to fall.
• Keep your frontier even, since a single stripe left behind can cost three closing moves.
• Near the end, identify the final shade and steer the last few picks toward it.
• When two picks look equal, take the one whose shade is most common in the parts you have not reached.

From paint bucket to hard problem

The engine under Flood is the flood-fill algorithm, the same seed-fill routine behind the paint-bucket tool in every image editor: pour at a pixel and the colour spreads through everything connected that matches. Turned into a puzzle, it became Flood-It, a quiet classic of the early smartphone era around 2009 and endlessly cloned since. There is also a genuine reason the game resists a formula. In 2010 a group of computer scientists proved that playing Flood-It optimally is NP-hard, in the same complexity class as the famously intractable problems, and that this stays true with as few as three shades. The same paper showed the greedy strategy can be made to perform arbitrarily badly on constructed boards. So when the biggest-bite move feels wrong, trust the feeling: no reflex solves this game, and a few seconds of lookahead is worth more than any rule of thumb.