Choosing which stats to show, and which to leave alone

Publishing a leaderboard changes what people do on your server. That is the point, and it is also the risk — because whatever you measure becomes what a proportion of your community optimises for, whether or not that was the behaviour you wanted.

The rule underneath all of it

Anything you rank, somebody will farm.

Not most people. But on any server with more than a handful of regulars, publishing a number creates a small group who will pursue that number past the point where it means anything.

That is not a reason to publish nothing. It is a reason to think about what the ranked behaviour looks like when somebody takes it seriously, before you publish, rather than after.

What each common stat produces

Every one of those is a real behaviour that appears on real servers within weeks of the leaderboard going up.

The interesting one is kill/death ratio, because it is the clearest case of a stat that inverts the behaviour you wanted. A server that ranks K/D gets players who leave fights they might lose — which makes the game worse for everyone including them.

Stats that behave better

The ones that hold up have a common shape: they reward doing the thing rather than accumulating the thing.

• Round wins rather than kills — rewards playing to the objective
• Time in the last 30 days rather than all-time — a leaderboard a new player can enter
• Participation counts — events attended, contributions made
• Nothing at all, for the categories where the honest answer is that ranking them helps nobody

Rolling windows are the single most useful adjustment available. An all-time leaderboard is a monument to whoever was there first, and it is unreachable by definition for anybody who joins later — which makes it demotivating for exactly the players you want to keep.

Public, private, or absent

Three levels, and they are different decisions:

Public leaderboard. Visible to everyone, drives competition, creates the farming pressure above.

Private stats. A player can see their own numbers. No competition, and it turns out most of the value is here — people like seeing their own playtime without needing to be ranked against anybody.

Not collected. Some things are better not measured, because measuring them makes them visible and visible things get optimised.

The middle option is underused. A server where everybody can see their own hours and nobody can see the ranking gets the satisfaction without the arms race.

What this does to new players

A leaderboard is read by new arrivals as a description of what matters here.

If the top entry is somebody with four thousand hours, the message is that this is a place for people who have been here for years. If it is somebody with a hundred kills this month, the message is that it is competitive. Neither is wrong; both are decisions about who feels welcome in the first ten minutes.

That is worth checking against what you actually want. A survival community that ranks kills is advertising something it may not be.

The numbers on your listing

Separate from in-game stats: what a directory shows about you is not your choice, and it is measured rather than reported.

That is deliberate on any list worth using — the figures come from querying the server, not from what an owner types — and it means the count next to your name is one number you cannot optimise. The useful response is to make it accurate rather than to work around it, since inflated counts convert badly even when they raise a position.

What you can control is what your own stats page says, and that is worth keeping honest for the same reason.

Retiring a leaderboard

Sometimes you publish one and it produces exactly the behaviour above. Removing it is legitimate and worth doing directly rather than letting it rot.

Say why. "We are removing the K/D board because it was making people avoid fights" is an announcement that increases trust rather than costing it, and it is the kind of reversal that makes every other decision you announce more credible.

Expect the people at the top to be annoyed. They are a small number and they were being rewarded for something you did not want.

It is also worth updating the listing when you retire one, because a description promising a leaderboard that no longer exists is the same stale-entry problem in a different place.

Data you are keeping

Stats are player data. Playtime, activity patterns, sometimes addresses attached to accounts.

Worth deciding how long it is kept and who can see the raw version. A leaderboard is public by design; the underlying table usually should not be, and it is the kind of thing that gets exposed by accident when a web panel is set up in a hurry.

The short version

Whatever you rank, somebody will farm. Look at each stat and ask what it looks like when a determined person optimises for it — and drop the ones where the answer is a behaviour that makes the server worse.

Prefer rolling windows to all-time, prefer participation to accumulation, and consider showing people their own numbers without ranking them at all. That last option delivers most of the value and none of the arms race.