Chess is our benchmark for depth. A quiet design choice makes WordChess deeper still.
In 1950, Claude Shannon, the father of information theory, estimated how many different games of chess were possible. His answer, roughly 10120, became the Shannon number, and it has anchored our intuition ever since.1 It is a figure so large it embarrasses the physical universe, which holds only about 1080 atoms.6 You could give every atom its own chessboard and still not have enough boards to play out every game.
Chess earns this honestly. From the opening, White has 20 moves; Black replies with 20, and there are already 400 positions after a single exchange. Six half-moves in, the count passes 119 million; by the tenth it reaches 69 trillion.4 Players call this the branching factor, the number of legal choices each turn. In chess it averages about 35.2 That modest number, compounded move after move, is the engine of the game's mystery. Over the first twenty moves it produces on the order of 1060 games. The source of chess's depth is not the pieces. It is the branching.
Chess's early move counts are known exactly. WordChess's are estimates, but the two games diverge so fast that the gap is unmistakable within a single turn.4
| After move | Chess, exact 4 | WordChess, estimate 7 |
|---|---|---|
| 1 | 400 | ~1012 |
| 2 | 197,281 | ~1018 |
| 3 | 119,060,324 | ~1024 |
| 4 | 84,998,978,956 | ~1030 |
| 5 | 69,352,859,712,417 | ~1036 |
Chess figures are exact move-generation counts (perft).4 WordChess figures assume roughly a million legal opening placements per side and a conservative thousand thereafter, see the method note.
WordChess looks like the gentler cousin, a word game on a grid, closer to a crossword than a knife fight. That impression is exactly wrong, and a single line in its rules is why: every player holds the entire pool of one hundred tiles.7
There is no seven-tile rack, no luck of the draw, no waiting for a vowel. On any turn, a player can reach for nearly any of the 148,941 words in the dictionary, words up to twenty-five letters long, and look for somewhere to put it.7 Scrabble, throttled by its seven random tiles, offers a branching factor of roughly 35, about the same as chess.5 WordChess removes that bottleneck entirely.
The consequence is violent. The very first turn opens to somewhere between one and two million legal placements, a word, an orientation, and a spot on the wide-open 25×25 board. When both players have moved just once, the game has branched into something like a trillion positions. Chess, after the same exchange, has four hundred.3
The rules are simpler. The space of possibility is not.
Each rung is ten times taller than the one below. On this scale, WordChess's first twenty moves climb clean past the number of atoms in the universe, and land exactly where an entire game of chess sits.1
As the board fills, chess's branching factor drifts up toward 35 and holds. WordChess's stays in the thousands, every word already played becomes a new anchor to hook onto, and the full tile pool means the only real limit is which crossings the dictionary allows.7
Run that forward. At a deliberately conservative thousand legal moves per turn, WordChess reaches 10120, Shannon's number, the complexity of an entire game of chess, within its first twenty moves. Allow ten thousand moves per turn, still reasonable, and twenty moves climbs toward 10160: a margin of forty to a hundred orders of magnitude over chess's 1060.1
Shrink the estimate until you assume a player finds only three hundred legal moves a turn, a fraction of the true number, and twenty moves still yields 1099. Still forty orders of magnitude beyond chess. The conclusion survives every pessimistic assumption you can hand it.1
Chess's numbers are the product of decades of exhaustive computation; they are known. WordChess's are careful estimates, drawn from its real parameters, a 25×25 board, a 148,941-word dictionary, and the full-pool rack, and they carry wide error bars. What is not in doubt is the direction and the scale of the gap. Every assumption in this piece was chosen to be conservative, and the gap is still enormous.
Chess constrains you: a knight moves as a knight, a pawn creeps one square, and your options, though rich, are finite and familiar. WordChess hands you the whole language and the whole board and asks you to choose. That is the trade the design makes, and it is the reason the friendly grid conceals a combinatorial ocean.
None of this makes WordChess harder to play well, a larger search space is not the same as a deeper strategy, and chess's genius is how much meaning it wrings from its narrow branching. But anyone who imagines a word game as the lightweight option has the mathematics precisely backwards. For its first twenty moves, WordChess makes the great game of kings look almost small.
Method. "20 moves" means 20 by each player, 40 half-moves, the chess convention. Chess: game count ≈ b40 with b ≈ 30–35 → ~1060. WordChess: opening branching estimated from (playable words that fit through the centre) × (placements per word) ≈ 106 per side; later turns held at a conservative 103–104 → b40 ≈ 10120–10160. The 1099 floor uses b = 300. These are estimates, not proofs; see "A note on certainty."