[coob]

Edit: I now find out that wladimir has had this idea and shipped code on the same idea: https://github.com/driverdan/dropship

I have wondered for a while if it could be done without the API by hooking into whatever checksumming is done before uploading.

For example, if you were to find a popularly pirated file (for example, a widely distributed MKV of an episode of Break Bad) and copy it to your Dropbox, the upload is instant as there is a high probability that someone has already uploaded it.

If you were able to somehow able to hook into Dropbox's checksumming functions it may be possible to report a checksum on a 'fake' non-existent file that is the same as what would be provided by the real copy of the file. The file then appears in Dropbox where you are free to download it at top speed having only known the hash of the file previously.

I wonder have safeguards Dropbox have to prevent this.

[rubenfonseca]

Dropship hack was immediately rendered useless by Dropbox by requiring that a random part of the file be uploaded, even if Dropbox already has many copies of the file in their system.

[toomuchtodo]

So use the Dropbox as the golden copy/SHA reference, and fetch the file from peers over Bittorrent.

DB is always going to be able to be used like a resilient tracker/datasource, and DB devs can only put so many walls in the way of using it like that before it starts to hinder your "average" user.

Example: Get magnet hash, get data from Bittorrent swarm, and then fetch from both swarm and public copies available via Dropbox. Verify against Dropbox copy if necessary.