If you've played Anubis Knows: Footballers and watched it correctly guess your footballer in just eight or nine questions, you might assume there's some kind of artificial intelligence behind the curtain. The truth is more interesting — and it's a technique that's been used in guessing games for decades, long before modern AI existed.
The "20 Questions" Principle
At its core, Anubis Knows: Footballers is built on the same logic as the classic "20 Questions" parlor game. The idea is simple: with the right sequence of yes-or-no questions, you can narrow down even hundreds of possibilities surprisingly quickly.
Here's the math behind it. If Anubis has 200 footballers in his database and asks a question that splits them perfectly in half, he's left with 100 candidates after just one question. Ask another perfectly balanced question, and he's down to 50. After about eight ideally-balanced questions, in theory, he could narrow 200 candidates down to roughly one name.
In practice, no single question splits the database perfectly in half every time — but Anubis always picks the question that comes closest, at every single step.
Entropy: The Secret Ingredient
Each time Anubis needs to pick his next question, he doesn't choose randomly. Instead, the game runs a calculation across every remaining unanswered question and works out which one would divide the current pool of candidates most evenly between "yes" and "no" answers.
This concept comes from a branch of mathematics called information theory, where it's known as maximizing entropy, or information gain. In plain terms: a great question isn't one that's interesting — it's one that eliminates the most possibilities, regardless of how you answer it.
This is exactly why Anubis sometimes asks questions that feel oddly specific, like whether your footballer has worn the number 10 shirt at a major club, or whether they've played for a Brazilian or Argentine club at some point. These aren't random trivia — they're chosen because, given everyone left in the running, that particular split was the most efficient one available.
Why It Sometimes Gets It Wrong
No system is perfect, and Anubis will occasionally land on the wrong name. This usually happens for one of two reasons:
- An ambiguous answer. If you answer "Probably" or "Don't know" to a question, Anubis can't narrow the pool as efficiently, which sometimes lets the wrong candidate slip through to the final guess.
- A database gap. If the footballer you're thinking of isn't in our database yet, Anubis will always land on his closest match instead — which can sometimes look like a genuine wrong guess.
We're continually expanding the database to reduce that second scenario, and we always recommend answering as decisively as you can (Yes or No rather than Probably) for the sharpest possible guesses.
No AI, No Internet Required
One detail that surprises a lot of players: Anubis Knows: Footballers doesn't call any external AI service while you play. The entire engine — the footballer database and the question-picking logic — runs locally in your browser. There's no waiting on a server response, no API costs being passed on to you through excessive ads, and no personal data being sent anywhere just so a jackal-headed god can guess your favorite striker.
It's a simple idea, executed carefully: the right question, asked at the right time, can do more work than people expect.