An agent collective is more auditable than a single model of equal capability, because its coordination has to pass between agents through channels that can be logged, rate-limited and read, while a single model's deliberation stays in hidden states; the advantage holds only where messaging is mandatory and standardised and hidden encoding is guarded against.

qualifiesUntitled source

Importantly, this should account for the potential of [steganography](https://arxiv.org/abs/2507.02737).

groundsUntitled source

Contrast an organization that does business in hallway whispers versus one required to do everything in writing

groundsUntitled source

(a) legibility, because in a single agent you could have deception as a "hidden" state, whereas different agents with standardization enforced on messaging will need to communicate, and this can be monitored, logged, rate-limited and so on.