I was reading the code for flumeview-reduce yesterday - there were comment about how it wait for a minute of no new messages then writes (or similar)
https://github.com/flumedb/flumeview-reduce/blob/master/inject.js#L57-L63

flume uses seq (sequence) and upto to track wherre it's up to in terms or reducing (or materialising) a view. In scuttlebutt this might be a the timetamp that's being used. seq is (also) the keyword used for the location of the data in the store - in the case of the flumelog-offset it's a location based on bytes ! so your seq progresses like : 9, 27, 102, 344, 392, ... (based on how big the things were you logged)