3 ms·

why even bother with this? just randomly change punctuation the same way drm is and they wont be able to match it with the correct account

by marme 13y ago

drchiu 13y ago
That was my first thought as well. Whoever came up with this idea may have more countermeasures, however. For instance, every page may provide a unique enough punctuational fingerprint to identify the buyer.
jerf 13y ago
They know where to look. You don't. You'll be throwing up a lot of chaff in places they aren't looking in the hopes of catching where they are looking in the crossfire, which is an approach that will lead to a degraded book. You can mutate the hell out of the paragraph I just typed, but if you don't know that the only bit I've hidden in that paragraph was "chaff" vs. "noise", you may not hit a thing that I'm looking for. And bear in mind that's two short sentences; if they hide enough of these things in there, they'll have plenty of leftover bits to track you down. columbo's idea is much more solid. It directly identifies the changes, and while it's not even all that challenging to put enough bits into these books to identify which 3 books the pirate copy comes from[1], you are making much greater strides towards defeating the DRM. And if you learn enough from the comparisons to begin reverse-engineering the algorithm, you may be able to entirely defeat it. ("Aha, they always treat "fire" and "flames" as one bit, I'll just use the same one all the time." String enough of those together and you win.) By the way, to any who might question whether students would do this... I used to work on a learning content management system primarily used for highly randomized physics problems. It is absolutely staggering what students will do to cheat. They'll happily generate Excel spreadsheets fully exploring 48 permutations of a problem, each of which may be missing a random 2 out of 5 variables to be solved for, and share it far and wide. Though whether the utility of "not being caught pirating" rises to the level of "hope to score with members of the desirable sex by helping them cheat through a mandatory class" (the only possible motivation I can imagine for some of what we saw) is dubious. [1]: In fact I'd consider it an obvious elaboration to also fingerprint other information into the books, such as destination University, and so on. Basically you're encoding a bit string into the books; you can do all the usual things you do with such bit strings, like parity, encoding various bits of information, etc. It's just steganography being put to a new use, and we know that can encode arbitrary information with arbitrary characteristics.

why even bother with this? just randomly change punctuation the same way drm is and they wont be able to match it with the correct account

by marme 13y ago

why even bother with this? just randomly change punctuation the same way drm is and they wont be able to match it with the correct account

drchiu 13y ago

That was my first thought as well. Whoever came up with this idea may have more countermeasures, however. For instance, every page may provide a unique enough punctuational fingerprint to identify the buyer.

jerf 13y ago

They know where to look. You don't. You'll be throwing up a lot of chaff in places they aren't looking in the hopes of catching where they are looking in the crossfire, which is an approach that will lead to a degraded book. You can mutate the hell out of the paragraph I just typed, but if you don't know that the only bit I've hidden in that paragraph was "chaff" vs. "noise", you may not hit a thing that I'm looking for. And bear in mind that's two short sentences; if they hide enough of these things in there, they'll have plenty of leftover bits to track you down. columbo's idea is much more solid. It directly identifies the changes, and while it's not even all that challenging to put enough bits into these books to identify which 3 books the pirate copy comes from[1], you are making much greater strides towards defeating the DRM. And if you learn enough from the comparisons to begin reverse-engineering the algorithm, you may be able to entirely defeat it. ("Aha, they always treat "fire" and "flames" as one bit, I'll just use the same one all the time." String enough of those together and you win.) By the way, to any who might question whether students would do this... I used to work on a learning content management system primarily used for highly randomized physics problems. It is absolutely staggering what students will do to cheat. They'll happily generate Excel spreadsheets fully exploring 48 permutations of a problem, each of which may be missing a random 2 out of 5 variables to be solved for, and share it far and wide. Though whether the utility of "not being caught pirating" rises to the level of "hope to score with members of the desirable sex by helping them cheat through a mandatory class" (the only possible motivation I can imagine for some of what we saw) is dubious. [1]: In fact I'd consider it an obvious elaboration to also fingerprint other information into the books, such as destination University, and so on. Basically you're encoding a bit string into the books; you can do all the usual things you do with such bit strings, like parity, encoding various bits of information, etc. It's just steganography being put to a new use, and we know that can encode arbitrary information with arbitrary characteristics.