The variation in megabytes stored per vault is slightly worse than the variation in chunks stored. This is roughly in line with intuition from ‘compounding’ randomness by adding random chunk size on top of (somewhat) random vault naming and random chunk names.
As previously established, variation in chunks stored is about 2x difference between the vault with most vs least chunks, and about 1.5x difference between the third quartile and first quartile.
Variation in megabytes is about 4x difference between the vault with most vs least MB, but still about 1.5x difference between the third quartile and first quartile.
It’s really good to see that most vaults will still be within 1.5x of each other both for chunks and megabytes stored.
I find it very difficult to pick much difference from looking at the charts, but the stats do indicate MB stored is more variable than chunks stored.
Slight tangent: based on the distribution of chunk sizes, 1M chunks with 8 copies ended up being about 2050 TB total stored by the section.
Note these two charts are from different seeds so the exact shapes are unrelated. It’s more the broad shape / jaggedness that’s of interest.
