Weighted Draws Are Honest, Rigged Ones Are Not

Unequal odds are frequently the fair answer

Equal chances is a default, not a principle. There are plenty of situations where it is the wrong answer: raffle tickets bought in different quantities, a chore rota correcting for who did the bad job last month, a ballot weighting applicants who were unsuccessful in previous years.

In each of those, an equal-chance draw would be the unfair option. It would ignore information everyone agrees is relevant, and it would do so while claiming to be scrupulous.

So weighting is not a compromise of fairness. It is often what fairness requires, and treating unequal odds as inherently suspect gets the analysis backwards.

Where it stops being legitimate

The line is whether the people affected know. A weighted draw whose weights are published is a design choice people can evaluate and argue about; the same draw with the weights concealed is indistinguishable from a rigged one, because the only difference is information you withheld.

That framing is worth taking seriously, because the practical test is not about intent. An organiser who weighted a draw for entirely good reasons and did not mention it has produced something an entrant cannot distinguish from manipulation, and cannot rebut afterwards.

The corollary is that disclosure has to happen before the draw. Explaining the weights after announcing the winner is explaining why the result you got was legitimate, which is a much weaker position.

How to publish weights so people can actually use them

Publish percentages, not raw weights. Three entries weighted 2, 3 and 5 requires the reader to normalise; 20%, 30% and 50% does not, and most readers will not do the arithmetic.

State the rule as well as the numbers. "Applicants unsuccessful in the previous three years receive four entries" tells people how to think about next year; a percentage table alone does not.

And show the table before the draw button rather than in a footnote. Odds nobody saw in advance are indistinguishable from odds invented afterwards, which is why every weighted tool on this site prints the table above the control.

The cases that are genuinely dishonest

Weighting a draw toward a specific individual rather than toward a stated criterion is rigging, whatever the weights are called. The test is whether the rule could have been written down before you knew who was entering.

Adjusting weights after seeing the entrant list is the same thing with extra steps. A criterion chosen because of who it favours is not a criterion.

And describing a weighted draw as random without qualification is a lie of a specific kind — technically the draw is random, and the claim the reader took from it is false. That is worse than an ordinary inaccuracy because it was accurate enough to defend.

What disclosure should actually look like

Saying the weights exist is not disclosure. A note that entries are weighted by contribution tells a reader that something is happening and not what, and it is the form most weighted draws use.

Real disclosure is the resulting probability, per entry or per band, published before the draw. "Long-standing members receive additional entries" is a policy; "a member of five years holds 3.1% and a first-year member holds 0.8%" is a disclosure, and only the second lets somebody decide whether they think it is reasonable.

The difference matters because the first form is what an organiser writes when they have not worked out the numbers themselves. Being able to publish the percentages requires computing them, and computing them is where a weighting scheme's actual effects — which are frequently more extreme than intended — become visible to the person who designed it.

The weighting that is genuinely hard to justify

There is one category where even full disclosure does not settle it, and it is worth naming because it comes up more than the clear cases.

Weighting by something correlated with an attribute you are not allowed to select on is not solved by publishing the weights. A draw weighted by length of membership in an organisation whose membership has only recently opened up is, in effect, weighted by something else, and saying so plainly does not stop it being true.

That is not an argument against weighting. It is an argument that the disclosure has to cover the criterion's effects rather than only its definition — publishing that long-standing members get four entries is different from publishing what that does to the composition of the winners.

The check worth running is to compute the weighted odds by whatever grouping you would be uncomfortable selecting on directly. If the answer surprises you, the weights are doing something other than what the stated criterion says, and disclosure alone will not fix it.

Frequently asked questions