Gridlocked

Gridlocked is a sliding-block puzzle lab. Slide cars around a gridlocked

lot to free the red car and get it out the exit — the classic Rush Hour

formula — but built from the ground up as a research tool as much as a

game.

**Two boards, one engine.** Play the familiar 6×6 square grid, or the hex

variant, where every piece can slide along *three* axes instead of two.

That third axis changes everything: it gives you more ways out, and (as it

turns out) it makes puzzles solve in noticeably fewer moves than their

square equivalents at the same piece count.

**Every puzzle is real.** Nothing here is hand-placed for difficulty —

every board comes from a full solver pass over a corpus of 14,000+ puzzles

(square boards from Fogleman's Rush Hour database, plus generated hex

boards at three densities), so the "hard" ones are hard because a

breadth-first search over the whole state graph says so, not because it

looks intimidating.

**Curious what makes a puzzle actually hard?** We wrote up the full

investigation — state-space size vs. dependency depth vs. how

"counterintuitive" the optimal solution is — with interactive widgets,

including a from-scratch explainer of how the hex board's third axis

works: https://orbitope.github.io/gridlocked/