New DRM Changes Text of eBooks to Catch Pirates

Meh.

Buy three ebooks from different accounts, render them to pain text, diff them and take out the differences.

Why three? The assumption is 2 books would have the correct value and 1 book would have the copyrighted value.

Though I don't see this as DRM. I've done this before on a much smaller scale to track things as they enter the wild. It's just security through obscurity, the digital equivalent of a tracking collar without any of the benefits, like guaranteed uniqueness.

It's also crazy overkill. Here's a much simpler way to do it: At the end of every period randomly decide to add an extra space. Hardly noticeable. Spread (again random) between 1024 and 2048 of these over the entire document. Count them up... now you have a pretty decent identifier without having to do any sort of special logic to the words themselves. To the average person it's not noticeable, it doesn't require messing with the original document, and it can be made unique enough to be a digital fingerprint. Of course this house of cards falls apart when you know the secret.

I have worked with e-book authors in creating systems to mitigate piracy of their books.

Even if you know who gave out the book that the pirates were able to get, it doesn't make them automatically guilty. In one case, the brother of a customer was the culprit... the author was actually surprised because from the looks of the evidence, it was someone who he trusted that "leaked" the book.

Sure, the person could have lied and blamed his brother, but the author trusted him enough to believe the story and accepted the apology from the brother.

Anything involving DRM starts to make things get hairy. I can understand the want of the authors to protect their contents that they worked hard on producing. But I also understand how the real customers don't like DRM. In the end, DRM was ditched and a subscription model was created. Both the author and the users were happy with that resolution.

I can see some benefits of this but I think the idea of a script automatically going through and making changes is quite horrible. In one example they removed a new paragraph break. That's an unacceptable modification to me.

I think that the best way to do something like this (if you want to do it at all) would be if the original author chose 30 small acceptable changes themselves. The presence of these changes (of lack thereof) would then allow a 30 bit ID number to be embedded in the document allowing each copy to be identified separately. They could also have a built in checksum to allow the original ID to be recovered even if some of the changes were found. By increasing this to 100 possible changes (50 active changes in each copy on average) then you could have a very large amount of error correction data and the original ID could still be recovered when a large number were found and changed (this would prevent pirates buying 2/3 copies to find the changes as 2 copies would only highlight half the changes on average if there is an equal chance of the change and original showing).

Ultimately I do not think this will work however. A pirate can use a pre-paid credit card to buy the e-book. In this way they cannot be tracked. So I honestly don't think this is worth the effort.

Just my thoughts on the matter.

That's not DRM. If anything, that's a textual version of "digital watermarking" http://en.wikipedia.org/wiki/Digital_watermarking

It's old news - it's been done to find internal document leakers for years - I believe even in typewriter days.

Thing is, it only works on a very small scale, when the people under surveillance are unaware, and are unwilling to collaborate. When these assumptions don't hold, it is trivial to generate a version which has all the information yet does not correspond to any original watermarking (or worse, replicates some watermark that is not part of the set used to generate the version)

This DRM actually changes the content by rewriting the text, not unlike Microsoft Word's grammer checker/corrector tool. I would be very surprised if publishers and authors would go for this. Clumsily changed sentences could ruin a good book.

I have always assumed that publishers embed an ID somewhere in the Mobi or EPub files (they are just HTML, you could insert all sorts of things that would never be displayed) since it is trivial to remove the DRM from them.

What if someone steals a copy of my file, then uploads it to TPB? Now do I get sued for millions?

Am I presumed guilty, since it's clear there's "evidence" to show that I was the source of the now widely spread copy? Do I now have to prove to a judge that I actually didn't upload that file to TPB?

Guilty until proven innocent?

I had hoped from the headline that they would be doing something subtle, like replacing some capital "A"s with Greek "Alpha"s, or replacing a single-character "ü" with separate "u" and "combining umlaut". But visible changes that mess with, not just the text formatting, but the actual words of the text, are unacceptable.

I have to say I started to have more faith in ebooks when I realized how easy it is to strip DRM from Kindle purchases. I've made more purchases since then and feel better about doing so. I like the convenience of ebooks, the way they make my bookshelves look antiquated, but I always had in my mind the "1984" scenario where they take away your purchases. So I strip the DRM of all my purchases and keep a local copy, even though I only ever read them with the Kindle app. When they take away my ability to do that, I'll stop purchasing.

I speak German and had a look at the example document [1], and it's hard to tell how big the changes are - this document shows the original and the changed version and asks how annoying the changes are, from a scale of 1 to 6, presumably to know how much their algorithm can change before the rights-holders complain.

Often, the words are replaced by synonyms ("nicht denkbar", not thinkable gets replaced by "undenkbar", unthinkable) or switches the ordering of words ("zwischen Blei Gold und Silber", between lead, gold an silver becomes "zwischen Gold, Silber und Blei", between gold, silver and lead), just the connection of words is changed ("Schmiergeldübergabe", transfer of bribes, becomes "Schmiergeld-übergabe") or the linebreaks are slightly changed, for example one word is moved to the next line.

At this stage, it's very hard to tell how much the final mechanism is going to change the original text, but I can't see much changing of very minor elements like punctuation.

It's still a stupid idea since many of these above elements are important to the "vibe" and meaning of a text - the way an author orders elements, or the specific words used by authors can be interpreted differently.

[1] http://www.scribd.com/doc/148113352/Sidim-Changes

Why on earth do the organizations that 'fight' piracy keep suffering from the delusion that it's just random everyday people who share a copy with a friend that are the source of pirated material on the internet?

This is more of treating the symptom(s) and not the problem(s).

Regarding DRM in general, it exists to prevent piracy, and piracy exists because content is easier to obtain, consume, and share through these 'illegal' methods.

If prices came down, accessibility went up, and shareability is embraced, piracy will fade into the background, and the remaining people doing it will be unjustified in their actions.

If content creators and distributions embraced a system that benefitted the consumers everyone would be better off, but they are years (if not decades) behind the times.

> the DRM shuffles some words around, inserts synonyms, changes the paragraph format or the punctuation. For example, the word “unsympathetic” could be changed to “not sympathetic,” and so forth.

This is horrible! I would prefer my book to be pirated than to have an algo. mix words up.

Besides, if you wanted to do this, why couldn't you use non-printing characters instead, like the soft hyphen U+00AD or zero width space U+200B?

And another comment...

I've always wondered if companies like Apple, Google, etc do text watermarking of company-wide confidential emails. Everytime I see an e-mail in the full text posted to Techcrunch, I imagine someone just got busted...

I remember seeing a startup doing something similar to catch employes leaking internal docs. Can't find it now, though.

DRM where I can still email my favorite books to friends. I support this.