Rewarding the people who keep turning up

Every giveaway has both kinds of viewer in it. Someone who has watched for two years and enters everything, and someone who arrived nine minutes ago from a raid. Treating them identically is defensible, and it is also the reason regulars stop entering: the fifteenth loss in a row to a stranger is a good argument for not bothering.

The obvious fix is the one that gets you in trouble

The instinct is to give subscribers better odds, or to let people spend channel points on extra entries. Both are common and both move a free prize draw towards something regulators have opinions about, because the extra chances can be read as bought. Subscriptions cost money. Channel points are earned by watching, which sounds free until you notice a new viewer cannot get any.

The version that stays clearly on the right side is weighting by attendance. Everyone can still enter, entry is still free, and the bonus accrues to anyone who keeps showing up — including the person who arrived nine minutes ago, eventually. Nothing is for sale.

What the ladder looks like

In Kloot it tops out at twice the chance, and it gets there in four steps rather than all at once. A first-timer holds the base share; each previous giveaway of yours they entered adds one, up to four extra. So the ladder runs 1×, 1.25×, 1.5×, 1.75×, 2× and then stops.

Two is a deliberate ceiling. Enough that a regular can feel it, small enough that a newcomer is not entering a lottery someone else has already half-won. It is counted in whole shares underneath rather than in decimals, which matters only to anyone recomputing the draw afterwards — and the point of the next section is that they can.

Everyone has to be able to see it

This is the part that makes the whole thing legitimate rather than merely legal, and it is easy to get backwards. The temptation is to show the bonus to the people who have it, as a reward. Do the opposite: show every viewer their own multiplier, including the ones sitting at 1×.

A weighting nobody can see is indistinguishable from a rigged draw. It is, in fact, exactly what somebody accusing you of rigging would describe — some entries counting more than others, for reasons only you know. Saying it out loud on the entry page, to the person it does not yet benefit, is what makes it a rule rather than a secret.

The written terms have to carry it too, in the same plain sentence as everything else: that the multiplier is based only on how often someone has taken part, and never on paying, subscribing or spending anything.

And it has to be checkable

A fairness receipt that only covered unweighted draws would quietly stop meaning anything the moment weighting existed. So the shares go into the receipt alongside the entrant fingerprints: whoever recomputes the draw is recomputing it with the same weights the server used, and a streamer cannot hand a friend extra shares after the fact.

Receipts published before any of this existed are untouched — an unweighted draw is hashed exactly the way it always was, so nothing that verified last month has stopped verifying. That was a constraint on the design rather than a happy accident: a receipt that stops verifying looks like proof of a rigged draw, which is worse than never having published one.

How the counting works, and where it errs

The count is per device rather than per name, because a name is typed and a device is not. It rises only when an entry is actually accepted, so a duplicate that gets refused earns nothing. And when the system is unsure — a slow write, a read it could not confirm — it under-counts rather than over-counts.

That direction is chosen. A viewer occasionally getting slightly less luck than they earned is a disappointment; a viewer getting more than they earned is the thing the whole feature is trying not to be.

What it does to your regulars, and what it does not

Be realistic about the size of the effect. Doubling one person’s share in a pool of three hundred moves them from a very small chance to a slightly less small one; nobody is going to feel the arithmetic. What they feel is being counted — the line on the entry page that says the channel knows they have been here before.

That is the actual mechanism, and it is worth saying because it changes how you talk about it. This is recognition with a small material component, not a rewards programme. Presenting it as "loyalty pays off" oversells it and invites someone to check the maths; presenting it as "the people who keep showing up get a bit of an edge" is both true and enough.

Whether to switch it on at all

It is per giveaway rather than per channel, which is the right granularity. A weekly draw for the people who are always there is a good candidate. A launch-day giveaway aimed at strangers is not — the entire audience is at 1×, and the only thing the setting achieves is a line on the page explaining a bonus nobody has, which is a strange first impression to make on a room you are trying to win over.

And if you are running a giveaway a sponsor is paying for, think twice regardless. The brand is buying reach among new viewers, and weighting the draw towards your existing ones is quietly working against what they bought. It is not dishonest — the terms say what the odds are — but it is worth deciding on purpose rather than leaving a setting switched on from last week.