Why player counts on server lists are usually wrong

Open almost any game server list and you will find servers claiming numbers they do not have. It is the oldest problem in this corner of the internet, it is worse than most players realise, and it is almost entirely a consequence of how the lists are built rather than how dishonest the owners are.

Here is where the numbers come from, which ones can be trusted, and how to tell.

The three ways a list gets a player count

The owner types it in. The list has a text field on the submission form and prints whatever is in it. This is still how a surprising number of toplists work, including some large ones, and it is exactly as reliable as it sounds. Nothing verifies it, nothing ages it, and a number entered in 2019 is still sitting there in 2026.

The server reports it. The list gives the owner an API key and the server pushes its own count. Better, because it updates, but still self-reported — a server can send any number it likes. Rate limits and sanity caps help; they do not make it true.

The list queries the server. The list connects to the game server itself, using whatever status protocol the game speaks, and reads the count from the response. This is the only method where the number is not chosen by the person it flatters.

Most lists mix all three and label none of them, which is the actual problem. A player has no way to tell whether "4,102 players" was measured or typed.

What can genuinely be measured

Quite a lot, as it turns out, because most games ship a status protocol:

• Minecraft answers the Server List Ping — the same request your client makes on the multiplayer screen. It returns online count, max slots, version and MOTD.
• Rust, Counter-Strike, Garry's Mod and most Source-engine games answer Valve's A2S query protocol, which returns players, max players and the current map.
• FiveM exposes an HTTP endpoint with the player list and queue depth.
• Tibia has official world data through TibiaData; open Tibia servers usually expose their own status JSON.
• Steam games report global concurrents through Steam's own API.
• Twitch, Kick and YouTube report live viewers through their APIs.

That covers the large majority of what a general toplist carries. The genuine exceptions are custom emulators — RSPS, most Metin2 servers, bespoke private servers — which have no standard status protocol at all. For those, self-reporting is the only option, which is why we built a push endpoint and a script rather than a text box: the number still comes from the owner, but it comes from their live server every five minutes rather than from their imagination once.

How counts get inflated

The methods, roughly in ascending order of effort:

Typing a bigger number. Where the field exists, it gets used.

Counting the wrong thing. A server with 200 registered accounts, 40 of which logged in this month, reporting "200 players". Not technically a lie, and completely misleading.

Counting bots. Idle clients, AFK accounts, or NPCs that occupy player slots. This one is common on servers that query correctly, because the protocol cannot distinguish a bot from a person.

Padding the query response. For self-hosted status endpoints, simply adding a constant. A server that reports real + 150 looks healthy at any hour, and there is no way to detect it from outside.

Peak instead of current. Showing the day's or the month's highest figure as though it were live. Technically a real number, presented to mean something it does not.

How to tell, as a player

You do not need any tooling. Three checks catch nearly everything:

Check at an odd hour. Every real server has a daily cycle: European evening peak, a trough around 05:00 UTC, a smaller American evening bump. A server showing 400 players at 4am and 400 players at 8pm is not measuring anything. This single check is the most reliable one there is.

Compare the count to the community. A server claiming 800 concurrent players will have a Discord with visible activity, a chat that scrolls, and people complaining about something. If the numbers say 800 and the Discord has nine people online, the numbers are wrong.

Join and look. The count says 300. Are there 300 people? On a Minecraft server, run the player list. On an RSPS, stand in the main bank for thirty seconds. This is unanswerable and it takes a minute.

If a listing shows a player-count graph, look at its shape rather than its height. Real population is a wave. Fabricated population is a flat line, or a sawtooth that resets at exactly the same time each day.

What we do, and where it stops

Every count on VoteRank that can be measured, is. We poll Minecraft over Server List Ping, Rust and Counter-Strike over A2S, FiveM over its HTTP endpoint, Tibia through TibiaData, and Steam and Twitch through their APIs — every few minutes, from our servers, not the owner's. If a server is down, the listing says offline instead of showing yesterday's number.

Where a game has no status protocol, the owner pushes the count with an API key, and that is self-reported. We cap it, rate-limit it, and it ages out if the reports stop — but we are not going to pretend it is verified, because it is not.

Two other things worth stating, since they are where most lists quietly cheat:

Player counts do not affect ranking. Rank is votes cast this month; live population only breaks ties. A server cannot climb by inflating its count, which removes most of the incentive to try.

View counts are unique visitors, one person per day. Not raw hits. It makes our numbers look smaller than a list counting refreshes and bots, and it means the number is comparable to something.

Why any of this matters

A player choosing a server is making a bet with their evening, and sometimes with their money. The population number is the main input to that decision — it is a proxy for whether the world will feel alive, whether the economy works, and whether anyone will be around next month.

Getting it wrong wastes the player's time and, in aggregate, poisons the whole category: after enough empty servers with impressive numbers, people stop believing any of them, and the honest servers lose too.

That is the whole argument for measuring rather than asking. It makes our biggest lists look smaller than the competition's. We would rather have the number mean something.