Altering Text In eBooks To Track Pirates 467
wwphx writes "According to Wired, 'German researchers have created a new DRM feature that changes the text and punctuation of an e-book ever so slightly. Called SiDiM, which Google translates to 'secure documents by individual marking,' the changes are unique to each e-book sold. These alterations serve as a digital watermark that can be used to track books that have had any other DRM layers stripped out of them before being shared online. The researchers are hoping the new DRM feature will curb digital piracy by simply making consumers paranoid that they'll be caught if they share an e-book illicitly.' I seem to recall reading about this in Tom Clancy's Patriot Games, when Jack Ryan used this technique to identify someone who was leaking secret documents. It would be so very difficult for someone to write a little program that, when stripping the DRM, randomized a couple of pieces of punctuation to break the hash that the vendor is storing along with the sales record of the individual book."
So... (Score:5, Informative)
Normal book publishers have been doing this for decades, inserting the occasional misspelling here or there. Later, they inserted correct spellings, but of the wrong word, to get around auto-correction in scanner software.
So...no, they can't patent it.
That's not how traitor-tracing algorithms work (Score:5, Informative)
They don't hash the whole shebang into one number. Rather, they take a (random) number and use that to generate a set of mutations and then probe for that set of mutations in the leaked document. So now, even if you alter the document further, you probably didn't undo the mutations in question. Even if you did, you probably didn't undo all of them and you almost certainly didn't produce a high-confidence result that it's somebody else's copy.
Similar to something Amazon patented (Score:4, Informative)
There was an article [slashdot.org] about it here a few years ago. A followup [slashdot.org] someone made to a comment I wrote to the article mentions some work being done by some guy from Purdue that sounds a lot like what's being done here. IBM also seems to be doing work [slashdot.org] on canary trap-based ideas.
Re:So... (Score:5, Informative)
There would be no need to reverse engineer a pristine copy of the work. Simply proofreading a single copy and correcting some of the existing errors, while at the same time, introducing a few new errors of the same type
I didn't read the article because I had seen it earlier in another news source, so I don't if this is mentioned in the one mentioned here, but proofreading may not do it in this case. The source I read mentioned two specific types of change that do not introduce any typos (I'm choosing the exampled myself):
- One of them was reordering of nouns when the order does not matter, e.g. "Peter and John went for lunch" vs "John and Peter went for lunch";
- The other was playing with negatives: e.g. "something is unclear" vs "something is not clear"
Since there are no actual typos, it's hard to spot the identifying bits. You'd have to change the text substantially, in order to have a good chance of being free from discovery. Adding your own typos may not serve any purpose, since the company selling can focus just on the changes they made, not looking for other changes introduced after.
Of course, if there is a concerted effort to release documents, all pirates would need to do would be buying a few copies and diffing the documents. You may not get the original back, but if the changes are randomly put in a specific set of words, you certainly can end up with something close to the original than any of the sold copies and still free from pirate identification.