Range count and key count are different
A range is an operational work unit. Because OpenPuzzle sizes assignments using observed backend speed, two completed ranges can contain different numbers of keys. “Ranges completed” therefore describes how many work units reached verified completion; it is not itself a measure of keyspace coverage.
Coverage should be calculated from the total number of keys represented by completed, non-overlapping intervals divided by the size of the defined puzzle keyspace. Keeping these concepts separate prevents a high count of small assignments from looking like more coverage than a smaller count of large assignments.
Network speed is a recent operational metric
The public network rate is derived from recent active progress. It can move as clients join, stop, change clocks or pass through heartbeat windows. It is useful for understanding current throughput, but it is not a permanent benchmark for a particular GPU and should not be used to rank participants.
What completed coverage proves
A correctly completed interval is negative evidence: the target key was not found among those candidates. That is valuable because the coordinator can avoid assigning that interval again. Aggregate coverage is therefore a record of eliminated candidates, provided the completion rules and non-overlap guarantees are trustworthy.
What coverage does not prove
Coverage does not reveal where the key is within the remaining space. A large completed fraction does not create a magical “hot zone” unless the puzzle itself provides additional structure, and OpenPuzzle does not assume such structure. The remaining candidates must still be treated as candidates in the remaining set.
Why cautious wording matters
Public dashboards can easily imply more certainty than the mathematics supports. OpenPuzzle deliberately describes coverage as evidence and avoids estimated discovery dates. Search throughput can estimate time to process a defined amount of work at a given rate; it cannot predict the location of an unknown key.