In the workshop
Stelletta
Star Battle, on irregular regions. The grids exist; the game does not, yet.
Place stars so that every row, every column and every region holds the same number of them, and no two stars ever touch — not even corner to corner. The regions are irregular shapes, which is where the difficulty lives.
- 2,004grids produced and checked
- 12solving techniques, in five levels
- 364daily challenges, a year per weekday
- Five shapes of puzzle
- From 8×8 with one star per region up to 14×14 with three. The largest are the hardest to make: a single 14×14 grid at the top level takes about six and a half minutes of computation, and two tries in five are abandoned.
- One solution, proved by a second opinion
- Every grid is passed to a SAT solver that shares no reasoning with the logic solver and counts the solutions. A grid with two never enters the bank. The stored answer is then checked against the one the SAT solver found, and the difficulty is recomputed from scratch.
- The level is earned, not assigned
- A grid's level is the hardest of the twelve techniques the solver had to use, easiest first. The top level never appears on its own — it has to be forced, and only the larger multi-star shapes can reach it at all.
- No technique that leans on the answer
- The solver never argues that the puzzle has one solution, therefore this. Two of the twelve do test an idea in order to refute it, which is a different thing from keeping a guess.
There is no app. Not a line of Swift, no project, no icon. What exists is the content and the machinery that grades it: 2,004 verified grids, a solver that does not guess, and a year of daily challenges ready for the day the game is built.