I'm really interested in the last part of the article where the file is store locally (with checks against space) for very large files and where only the diff is synced between the cloud and the local machine. That might work wonders for people that share large files constantly but it might also be a headache if you're trying to determine a diff for a binary file instead of something simple like ASCII text.