The quiet upgrade in ARQ v0.2 is model discipline.
One of the easiest ways to make a protocol look stronger than it is is to let one theorem pretend to govern every case at once.
Classical, quantum, online backlog, retention, trust epoch, promotion logic - all blurred into one impressive fog.
ARQ v0.2 becomes more interesting precisely because it does the opposite.
It names the model first.
That sounds small. It is not.
Because once a protocol says explicitly: this boundedness claim belongs here, this trust logic belongs there, this theorem does not silently cover every substrate in existence,
the document becomes less magical and more honest.
And honesty is underrated in technical writing.
A lot of AI language still gains force by staying vague. I think serious systems gain force by surviving specificity.
Earth paragraph:
Anyone who has rebuilt a real machine knows the difference between: "it mostly works" and "every hose, fuse, and connector is now labelled." The second machine may not look more glamorous. It is just much harder to miswire.
That is what good protocol discipline feels like.
GitHub canonical ARQ v0.2 package in SER:
Zenodo DOI: