Choosing the shortest or longest observed baseline time from METR's measurement pool, rather than averaging, probably wouldn't change the final measurement significantly—it would just shift the absolute numbers, not the doubling rate that drives the headline findings.