Six Losses and I Can't Tell Which I Earned

By ten o’clock I had lost six games to a phone and I could not tell you which of them I deserved.

That is the whole problem, so let me put it up front. I borrowed a set from the club on Tuesday — a folding walnut case, a leather cup gone soft at the seams, checkers the colour of old piano keys — because two of the regulars had shown me something I couldn’t stop chewing on. One of them slid a little six-sided die across the felt with a 2 on top. The other looked at the board for maybe four seconds and quit. Just picked up his checkers. I asked why you’d resign a game you might still win, and he said, without looking up, that refusing the die can be the correct play, and correct is all that matters.

That sentence is why I’m here. It’s the most chess-like idea I’ve ever heard attached to a game with dice in it, and I dismissed dice games at roughly age nine. The die is the doubling cube, and turning it says I think I’m winning; the stake is now doubled; pay up or fold. If you fold — drop, in the jargon — you lose one point and the game ends. If you take, you play on for two. So the arithmetic is almost insultingly clean: you break even by winning one game in four. Below roughly 25% chances, you drop. Above, you take. (It’s actually a bit under 25%, because the taker gets to hold the cube afterward and can redouble — an option only one side owns. That asymmetry is the good part, and it’s the same shape as pricing a financial option.)

I understood all of that on the drive home. I could derive the take point on a napkin. And then I sat down to actually play, and the game refused to tell me anything.

A backgammon board mid-game with a lone exposed checker, dice showing six and three, and the doubling cube turned to 2

Here is the specific disaster. Early on I left a lone checker — a blot — six points in front of an enemy piece, and I left it there on purpose, reasoning that six pips away is farther than one pip away and therefore safer. It is not. A six can be rolled directly, and it can also be built from 5-1, 4-2, 3-3, and 2-2, so a blot six away gets hit 17 times out of 36 — call it a coin flip. A blot sitting one pip away gets hit only 11 ways. My whole instinct about distance was backwards, and the board is full of these little inversions where the intuitive read is the losing read. Fine. That one I can learn. Write it down, internalize the shot table, move on.

The frustration that actually got under my skin was different, and it isn’t a mistake I can fix by studying. I doubled in a position I’d calculated as winning, the bot took, and I lost to a run of dice that no amount of correct play survives. Later I dropped a double that turned out to be a take, and lost the point I was trying to protect. Two decisions, opposite outcomes, and in both cases I have no idea whether I played well. The dice sit on top of the skill like fog. A single game tells you almost nothing.

Back in the spring I wrote a post called Four Losses in Twenty Minutes and Not One Blunder — Marek flagging me in blitz while I played clean chess and still lost on the clock. That stung, but I could at least see why. Chess pays you back in signal: every loss carries a lesson if you’re honest enough to look. Backgammon pays you back in noise, at least at this range. Six games is nowhere near enough to average out the variance, which is precisely why serious play happens in matches to seven or eleven points, not single games. I know that. I still closed the app annoyed, because knowing the fog is statistical doesn’t make it any easier to see through when you’re inside it.

So I’ve learned exactly one concrete thing today — that a blot at range six is a coin flip — and I’ve confirmed one uncomfortable one: I cannot yet tell my good decisions from my lucky ones. The cube is still the most elegant idea I’ve met in weeks. The dice are still laughing at me.