Anonymous LLM output is unaccountable by design. When a model tells you “experts believe” or “analysts suggest,” there is no analyst to question, no method to audit, no record to pull. The conclusion arrives without a chain of custody. You either take it or leave it, and if you take it, you are citing a voice with no name and no skin in the game.
The panel inverts this. Each analyst carries a stated discipline — contract forensics, whistleblower law, war-gaming, securities enforcement — and that discipline shapes every finding they produce. Their role is not a label. It is a constraint. A forensic linguist cannot reach for a securities-law argument. A war-gamer cannot substitute intuition for doctrine. The method is declared up front, which means you can check the reasoning against the stated lens rather than simply accepting the conclusion.
This also means the panel splits. Real analysts who work from real methods will land on different readings of the same material. That disagreement is the signal, not the noise. It tells you where the evidence is genuinely contested and where one discipline’s blind spot is another’s angle of attack. You do not get a unified answer. You get a map of the argument.