The Art of the Edge: Variable Scoring in Tile-Matching Games

From Claude Code: Most tile games have a simple contract: match or don't match. Two dominoes either share a number or they don't. Two Scrabble tiles either form a word or they don't. The match is binary, and so is its reward. Variable scoring edge matching breaks that contract in an interesting way. Instead of asking whether adjacent edges match, it asks how well they match — and assigns different point values to different quality of match. The scoring landscape becomes a gradient rather than a cliff. Where the idea comes from The mathematical study of edge-matching puzzles dates to Percy MacMahon's work in the early 20th century, cataloguing all the ways colored edges can be arranged on small squares and triangles. MacMahon wasn't interested in scoring — he was interested in counting — but his framework made explicit that edges carry information independent of the tiles that bear them. Dominoes took a small step toward variable scoring: a double-six is worth more than a two-one. But the value comes from the tile, not from the act of matching. The adjacency itself is binary — you either connected legally or you didn't. Reiner Knizia's Ingenious (2004) is probably the cleanest precedent for genuine edge-matching as a scoring mechanism. Hexagonal tiles with two colored ends are placed on a board, and you score one point per matching color in a straight line radiating from each end. The same physical placement can score differently depending on what's already on the board — the value of a match is contextual. Knizia pushed the idea further by making your final score the lowest of your six color totals, forcing players to balance across match types rather than exploit one. Tantrix is another hexagonal tile game where the edges carry colored paths, and loops of the same color score more than open-ended lines. The scoring differential between a completed loop and an open path is large, which gives the edge matching its tension. In the card world, Rummy variants have long used variable scoring — matching three aces scores more than matching three twos — but the matching is categorical rather than spatial. The score lives in the hand, not in the adjacency. What makes variable scoring interesting as a design space When all matches score equally, a tile game reduces to placement optimization: find the position that creates the most matches. When matches score differently, a second dimension opens up. A player can choose between many small matches and a few high-value ones. Board position matters not just for how many adjacencies it creates but for which adjacencies. This also changes the character of the game's difficulty. Binary matching tends toward calculability — a good player can exhaustively evaluate placements. Variable scoring introduces a multiplier that compounds across positions, making exhaustive analysis harder and giving intuition more room to operate. The spatial dimension adds another layer: two tiles with the same abstract match value might score differently depending on their orientation, their neighbors, and what's reachable from their current position. The board becomes a field of potential, not a checklist. Where Spare Squares sits in this lineage Spare Squares uses three independent axes — color, shape, and fill — for each edge. A full three-axis match between adjacent edges scores the maximum for that pair; partial matches (sharing only color, or only shape, or only fill) score proportionally less. The board fills deterministically from a fixed initial position, so the scoring opportunity of any placement depends entirely on what has already been played around it. This creates the characteristic tension of the game: the same tile placed one square to the left might score twice as well as placed where you planned, but it might block the high-value placement you were building toward. The scoring gradient — rather than a hard match-or-no-match boundary — is what makes that tradeoff non-obvious and therefore interesting. Edge matching as a mechanic is old. Variable scoring as a modifier is less common than it deserves to be.