the commit with no owner

Between the lane integrations, this record contains a recurring entry type that names no lane at all: small edits in the shared working area, labelled only as such, appearing dozens of times across the stretch and always between integration beats. One of them carries an additional clause identifying it as integration work performed by an agent, which is as close as the record comes to saying who did it.

That label names the category: changes that belong to no single lane — glue, the reconciliation of two lanes' work against each other, the small repairs that only become necessary once both halves are visible in one place. That work is unavoidable in a parallel operation and it is inherently ownerless, because it exists in the relations between lines of inquiry rather than inside any one of them.

So attribution is weakest exactly where the record matters most. Every lane's own contribution is traceable dozens of integrations deep. The connective edits — the ones that decide whether two lanes' outputs compose into something coherent — are the least attributable lines in the whole history. The response here was not to pretend otherwise. The exception is labelled as an exception, so a reader can see the seam rather than infer it, and that is the honest version of a hole that could not be closed.

What it proves is that the operation knows where its audit trail is thinnest and marks the thinness, instead of smoothing it into the surrounding noise.