Does Watermarking Actually Cut Down on E-book Piracy?

[Len]

Watermarking has been held up by some industry members as an effective way to deter e-book piracy that doesn’t require more restrictive DRM technology. I recently discussed the state of watermarking with Bill Rosenblatt, founder of Giant Steps Media Technology Strategies.

Do you have a sense of how many U.S. publishers are using watermarking in their e-books?

The biggest use of watermarking so far is the U.S. market is in STM. An increasing number of STM publishers are using it. The pioneer has been O’Reilly Media, which has been watermarking its PDF e-books for years. That’s especially interesting given O’Reilly’s longtime and very public stance against DRM. Springer uses watermarking. MIT Press recently announced that it’s giving up DRM and moving to watermarking for e-books sold on its own site. STM publishers dislike DRM because it interferes with the spread of scholarship, so they are turning to watermarking as a scheme that gives them some protection without hindering sharing. In all of these cases we’re talking about PDFs [rather than ePub formats].

Otherwise, watermarking is more prevalent in other countries, such as in parts of Europe—the Netherlands, Germany, Italy, and elsewhere. U.S. trade publishers tend not to use it; the big exception is J.K. Rowling’s Pottermore, which uses it on Harry Potter e-books in ePub format.

Generally I find that there’s very scattered awareness of watermarking among publishers. There are several independent suppliers of watermarking technology out there, and some publishers do it themselves. But watermarking is not a thing the way DRM has been a thing, and in fact some people confuse watermarking with DRM because it’s been referred to as “social DRM.” I consider the two technologies to be distinct.

Are U.S. publishers using e-book watermarking in addition to traditional DRM, or instead of it?

Watermarking and DRM can be used together, but the major U.S. e-book retailers aren’t supporting it. So in practice, it’s generally either one or the other, or neither.

The key to effective watermarking is to embed data in each downloaded e-book file that identifies the individual purchaser. It could be human-readable, such as a name or email address—as O’Reilly and Springer do—or it could be an obfuscated ID that only the retailer understands, as Pottermore does. That requires that the retailer do this embedding on the fly, just before it sends the file out to the user, which in turn requires integration with the retailer’s user database.

E-book retailers do the equivalent for e-books with DRM: they typically encrypt each file using a key that’s different for each user just before sending it out. It’s certainly possible to embed a watermark and then encrypt an e-book. The reason why major retailers don’t do this has to do with incentives. Retailers have incentive to use DRM because it helps them protect their “walled gardens,” but they don’t have similar incentive to implement watermarking.

Some white-label retail service providers, like Aerio and Firebrand, offer watermarking as an option, as do a couple of small independent distribution platforms. I do know of a few instances of watermarking combined with DRM, but they aren’t for e-books; they are used for things like sensitive corporate documents and high-priced training materials. But I wouldn’t expect to see Amazon or Apple or B&N support watermarking anytime soon. Maybe they’d consider it if the big trade publishers decided to drop DRM.

Do we know how prevalent e-book piracy is, and whether watermarking has any impact?

Research on the impact of piracy mitigation techniques such as watermarking is virtually nonexistent. The only decent study I know of is the one that Prof. Imke Reimers of Northeastern University published last year using data from Rosetta Books and Digimarc. That was a peer-reviewed study published in a prestigious academic journal, the Journal of Law and Economics. She presented her paper at [Rosenblatt’s] copyright and technology conference this past January. The study showed a 14% increase in e-book sales with certain antipiracy measures in place. But it didn’t cover watermarking; it covered searching for unauthorized copies of e-books on the Internet based on metadata, such as filenames and book titles.

I believe that a few publishers have done their own studies of antipiracy techniques and are keeping the results confidential. I think it would serve the industry if there were more publicly available, independent, peer-reviewed studies of this type, and I’ve tried to pitch industry bodies on doing them, to no avail so far. But I’m confident that watermarking does work to reduce piracy, for the simple reason that savvy publishers like O’Reilly continue to use it. From my experience with O’Reilly, including as an author, I can say for sure that O’Reilly wouldn’t use a technique like that unless it believed that it works.