Tonight one room does. Every claim scored, every dodge caught, every hit taken off the HP bar.
| # | Fighter | Bouts | Wins | Win % | KO | Upsets | CPU level | Yap | On-target | Sharp | Clear | Receipts | Dodges |
|---|
Ranked wins come from the judges' cards. CPU bouts count toward your level, not your record.
After every turn the analyst reads the exchange: what the claim was, whether the reasons hold, what it does to the other side's last point, and how much it takes off the HP bar. A bar at zero is a KO. At the end both positions get the same six marks, side by side: logic, punches landed, evidence, eloquence, timing, and how much of the position is still standing. With judges in the room, the cards decide the ranked result and the analyst's read is the commentary; with no judges, the read is the result.
Underneath, a second model tags each turn with things it was tested on: 76 real arguments, rated twice with the speakers swapped, correlated against length, hit with five injection attacks. Those tags are the small chips.