Does it just feel unfair, or is it? Audit your random picks over time

Anyone who runs repeated random picks — daily standup orders, weekly chore assignments, the classroom name jar — eventually hears the complaint: "it always picks me." Sometimes that's just how randomness feels from the inside. Sometimes the process really is skewed, because someone's name is in the list twice or half the team was missing from Tuesday's draw. The only way to tell the difference is to stop debating impressions and look at the actual record. If your randomizer keeps a timestamped log of every draw, that record already exists — you just have to read it.

Why your gut is a terrible fairness detector

Human memory is spectacularly bad at judging randomness. We remember the mornings we got called first at standup and forget the weeks we coasted. Genuine randomness is also streakier than intuition expects: in a team of five, the same person opening standup twice in a row happens about once every twenty-five days by pure chance, and three times in a row will eventually happen too. When it does, it feels rigged. A fair process and a broken one can look identical over a handful of draws — they only separate when you count across dozens. That's exactly the evidence a draw-by-draw log gives you and your memory doesn't.

The log is already being kept for you

Every tool on this site quietly records its recent results in your browser — what was picked and precisely when. The randomizer history & export page collects all of it in one place, grouped by tool, and turns it into a CSV file with one click. The export has three plain columns: the tool that made the pick, an ISO timestamp, and the result itself. That's a dataset, and a small one is all you need. One thing to know before you start: each tool keeps a rolling log of its ten most recent draws, with older entries dropping off as new ones arrive. An audit that spans weeks is therefore built from a habit — export regularly and append the new rows to a running sheet, and the ten-draw window stops being a limit.

A ten-minute audit in any spreadsheet

Open your accumulated CSV in any spreadsheet app and the analysis is genuinely simple. Filter the tool column to the draw you care about — say the standup order randomizer — so you're looking at one process, not everything mixed together. Then count how often each name appears in the result column; a pivot table does it instantly, and in a pinch COUNTIF does the same job. What you want to see is everyone's tally in the same neighbourhood. After twenty draws in a five-person team, four picks each is the ideal; a spread like 2-3-4-5-6 is completely unremarkable. What should raise an eyebrow is a tally that's wildly out of line and stays out of line as more draws accumulate — the person with twelve picks out of twenty when chance says four, not the person with six.

The timestamp column earns its keep here too. Sort by it and you can answer the questions that actually come up in disputes: was there really a draw every day that week, or did someone skip Thursday and quietly assign themselves the easy job? Did the "re-roll" happen thirty seconds after the first result — the classic tell that somebody didn't like the original answer and spun again? A log doesn't just tally outcomes; it shows the process.

When the audit finds something

If one name really is over-represented far beyond what chance explains, the cause is almost never the random number generator — it's the input list. The usual suspects are wonderfully mundane: a name entered twice (once with a typo), a stale list that still includes someone who left, or a chore roster where one person appears under a nickname and a full name and so holds two tickets in every draw. The audit's real value is that it converts "this feels off" into "Sam appears in the log nine times and everyone else four — let's check the list," which is a conversation with an ending. And when the tallies come back level, that's just as useful: you can show the complainer the counts, sympathise that randomness is streaky, and move on without changing anything.

Publishing the receipts settles it in advance

For one-off draws with something at stake — a raffle or giveaway with a real prize, or a single pick that decides who presents to the board — the strongest move is to export the row immediately and share it with the announcement. A result published alongside its timestamp is very hard to argue with, and much of the arguing never starts, because everyone can see the receipt exists. The habit costs one click at the moment of the draw and saves the whole relitigation later.

Count, don't quarrel

Fairness disputes about random picks are rarely about the randomness — they're about the absence of evidence either way. A running log closes that gap. Everything stays on your own device, the export is a click, and the analysis is a filter and a tally in the spreadsheet you already know. Next time someone insists the universe has it in for them, open the history & export page, download the CSV, and count together. Either you'll find the duplicate name, or you'll find that the universe is innocent this time — and both answers end the argument.