Can your friends tell when you're lying?

Caught means the liar got more votes than anyone else. Anything less and they got away with it. In 17% of rounds, someone innocent tied the liar for the most votes. In 10%, someone innocent got more. And in 27%, nobody voted for the liar at all.

That last one is worth a second look. In 27% of rounds, a whole table looked straight at the person lying to them and didn't give them a single vote.

Does table size matter?

More people should mean more eyes on the liar. It also means more suspects and more ways for the votes to split, and a split vote helps the liar. If the liar gets one vote and so does someone innocent, that's a tie, and a tie lets the liar walk.

At three phones, nobody innocent ever gets more votes than the liar, and not because three-player tables are sharp. With only two voters, an innocent player can't outvote a liar who got any votes at all.

These are 3,000 rounds of The Lobbyist played between Oct 10, 2026 and Nov 3, 2026, rounded down to the nearest hundred. In every round, everyone picks a side of a question in secret, one player is told to argue the side they didn't pick, the table argues, and everyone else votes for the person they think didn't mean it. When a round ends, the game records how the vote came out, how many phones were playing and how long the table argued before voting. Only the votes of players who weren't lying count, as in the game's scoring. Everything here is a total across many games, so no single player or table can be picked out of it: a table size needs at least 200 rounds to show up here, and a question needs at least 20 rounds from 15 different games.

The best-known number on lie detection comes from a 2006 review by the psychologists Charles Bond and Bella DePaulo in Personality and Social Psychology Review ("Accuracy of deception judgments"). They pooled research covering 24,483 judges. Asked whether a statement was a lie or the truth, people got it right 54% of the time, against 50% for a coin. They recognized 61% of truths but only 47% of lies, because people lean toward calling what they hear true.

It's a loose comparison, because the tasks are different. In a typical lab study, one judge watches one statement, often from a stranger, and calls it true or false. At a Lobbyist table, the players argue face to face, question each other, and then vote on which one of them is lying. That's harder in one way: there are several suspects, not two answers, so a player guessing blind at a table of five has one chance in four. It's easier in another: you usually know these people, and you get to question them before anyone votes.

So the two numbers don't line up exactly, and they aren't meant to. But even with a full argument and a vote, 54% of liars at real tables got away with it.

Tables argued for 79 seconds at the median before they voted. Split by outcome, rounds where the liar was caught ran 61 seconds, and rounds where they got away with it ran 93.

That's a pattern. It doesn't say why, and it reads two ways. Maybe a longer argument gives a liar time to blend in and pick up allies. Or maybe it runs the other way: when someone slips early, the table spots it and calls the vote, and when nobody has a clue, everyone keeps talking. A long argument might be a sign the liar is already winning, not the reason.

These are the Fuzzy Logic questions real tables argued about longest.

If you're the one lying

54% of liars before you got away with it. Some things that help:

• Don't rush the vote. Rounds the liar got away with ran longer. Whatever the reason, you have nothing to gain by hurrying.
• Aim for a tie. You don't need the whole table to trust you. You need one innocent player to collect as many votes as you do.
• Stay ordinary. The liars who got no votes at all didn't have to win the argument. Nobody looked twice at them.

If you're trying to catch them

• Agree on a suspect before you vote. Scattered votes are how a liar walks away with a tie. Say who you're voting for and why, out loud, while there's still time to change minds.
• Ask for details early. A made-up position runs out of story fast. Ask why, then ask for an example.
• Don't ignore the quiet ones. A liar nobody votes for wins just as much as one who argued brilliantly.

There's more on both jobs in our bluffing tips and our guide to arguing a side you don't believe.

Your friends might be better than average. Or worse. The only way to find out is to lie to them. In The Lobbyist, everyone picks a side of a question on their own phone, one player is told to argue the side they didn't pick, and the phones reveal who it was after the vote. It runs in the browser, and Fuzzy Logic, the deck these questions come from, is free. Here's how to play.