← back to the board
SHIPPED experiment 01 · 9 Mar 2026 · still live

Whist Scorer, started at the table

A ski-trip scoring tool that went from spoken rules to a live, usable app in the time it took to deal the next hand.

Status
Shipped · live
Time to first use
~20 minutes
Spec format
Spoken aloud
Rewrites since
Zero
[ whist-scorer · hero screenshot · 1600×900 ]
the live scoreboard, mid-game — every phone at the table looking at the same running total.
The build log

How it actually happened

DAY 0 · 21:40 · THE CHALET TABLE

Described the rules out loud

No spec doc, no wireframe. I just talked through our family whist variant — the bidding, the doubling rule, who deals, how ties break — while the cards were still out.

“It’s like whist, but the bid is a contract and you lose double if you miss it by more than two.”
DAY 0 · 22:05 · FIRST RUN

A working scoreboard, first try

The first build already tracked players, rounds and running totals. It got one edge case wrong — the double penalty — which took exactly one more sentence to fix.

[ setup screen · add players ]
add-players screen — the only setup you need before dealing.
DAY 0 · 22:30 · AT THE TABLE

Everyone opened it on their phone

The real test: passing the URL around. Four people, four phones, one shared game. Watching non-technical family members just... use it, minutes after it existed, was the whole point.

[ live scoring · round 4 ]
round four — the app quietly settling an argument about who was actually winning.
WEEKS LATER · STILL RUNNING

Four game nights, zero changes

It kept working. No crashes, no feature requests, no maintenance. It does one thing, it does it at the table, and it gets out of the way — which is exactly the bar I want every experiment on this desk to clear.

The verdict

The fastest spec is a conversation, and the best test suite is a competitive family. Four game nights later, not a line changed.