Since many large files are saved by a lot of people individually (movies, mp3 files, pdf manuals, etc) this saves them a lot of space. As they get more users this advantage grows.
Since many large files are saved by a lot of people individually (movies, mp3 files, pdf manuals, etc) this saves them a lot of space. As they get more users this advantage grows.
What if I wanted to use Dropbox for piracy purposes (I don't, just using an easy to understand model). Could I upload a copy of Avatar.mp4, compute or find the checksum for it, and then distribute just that checksum to other users? Or distribute a file that wasn't Avatar.mp4, but looked like it to the Dropbox client? Would this effectively trick the Dropbox servers into distributing the file for more, so that I could legitimately claim to not be distributing copyrighted material, but just distributing a "magic" file that caused Dropbox to give me a copyrighted file that someone else had uploaded?
Modern hashing algorithms make this practically impossible, sorry.
Just because you have a hash string that's not technically a copyrighted film, you could still be guilty of copyright infringement. cf. http://ansuz.sooke.bc.ca/entry/23
I don't see how it shouldn't be good practice. The only information you're potentially giving away is that someone else that has a dropbox is in possession of the same file as you.
If dropbox client were to still force the upload to happen, and only discard the object afterwards, this "duplicate file exists" information wouldn't be leaked. Presumably, they want to detect duplicates within an account and avoid the transfer in that case in the interest of user experience, but if I've told dropbox to upload avatar, I'm presumably OK with the transfer happening, even if it's not strictly needed because someone else has already uploaded it.