The Structural cause of file size distributions
- Allen B Downey, October 2000
http://rocky.wellesley.edu/research/filesize/
The author finds the same thing as me, that file sizes follow a LogNormal distribution.
The author proposes the multiplicative generating model I described on the LogNormal page as the mechanism causing the distribution.
What's cool about this paper is that the author actually finds a little bit of evidence for the multiplicative generating mode, by comparing the sizes of old versions of files with the current version.
--NickChapman