One number that follows you from round to round, like a chess rating. It moves when you win or lose against real people, and only then. The judge writes your ballot; the people you beat write your rating.
Six tiers. Everyone starts at 1500, between Regular and Contender. The top of the ladder is earned against people, never against the AI.
Beating a stronger opponent moves you more. Practice rounds against the AI sharpen you without touching the number.
Plain Elo would move you 30 points on a coin flip and call it skill. The Debate Rating is Glicko-2: it carries a confidence range that narrows as you play and widens if you vanish for months. Until 3 rated rounds settle it, your rating is provisional: it shows on your profile and drives matchmaking but holds no numbered place on the public board.
Look yourself up and claim your record: the index covers hundreds of tournaments of published Tabroom results, and your rating seeds from your real wins, losses, and elim runs. Wide range on purpose, once ever, and only before your first rated round, so nobody re-rolls a bad start.
Same idea, different math. Elo assumes you play often enough for noise to average out. Debate rounds are sparser, so Debatable uses Glicko-2, which tracks a rating and how certain the system is about it. A new account and a proven account can both sit at 1500 and mean very different things.
Winning and losing rated rounds against real people. Beating a stronger opponent moves it more than beating a weaker one. The AI judge's speaker points never rank you; only results against people do.
Yes. If you have competed before, you can look up your published Tabroom results and seed your rating from your actual record. The seed carries wide uncertainty and tightens as you debate.
A rating stays provisional until the system is confident: you need at least 3 rated rounds and enough results for the uncertainty to narrow. Provisional ratings display and matchmake but do not hold a numbered place on the public board.
Newcomer, Rookie from 1300, Regular from 1450, Contender from 1550, Varsity from 1650, Elite from 1750, and Titan from 1900. Everyone starts at 1500 with high uncertainty.
The rating runs server-side and every change is recorded with a before and after, so a rating can always explain itself.