The Google Book Search Settlement, which was amended November 13, is a big document and a big deal: $125 million and a dramatic shift in copyright law. Having spent the last year studying the settlement and writing about it, I'd like to explain why it matters so much. On Google's home turf—search—what it does is almost wholly good. When search engines work right, they empower users to seek out whatever they want to learn. That's the exact opposite of broadcasting, in which a few big speakers choose what everyone else hears. In a world where everyone can self-publish, search engines turn what would otherwise be deafening cacophony into the best party ever, where every guest can instantly join the conversation that most interests them. That's good for freedom, and good for democracy. Building better search engines is a moral imperative.

That's why Google ought to prevail if the lawsuit that resulted in the settlement were ever to go to trial. Courts have treated online copying as a fair use when the purpose is to create a Web search engine. In the same way, scanning books to make a book search engine should also be a fair use. The settlement, though, goes far beyond just scanning and indexing. Selling full books is the kind of commercial transaction at the heart of copyright. It would be obviously infringing unless the copyright owner consented. Using a class action to obtain that consent—en masse—was a stroke of genius by the settlement's architects. It turned the lawsuit inside out.

My immediate reaction, as I blogged the afternoon the settlement was announced in October 2008, was “This is a Google-only deal.” The settlement would let Google sell out-of-print books unless the copyright owners objected. A competitor, however, would need to get individual permission first or be sued into oblivion. That's hard enough in general, and for orphan books it's impossible. There's no one to ask. The class action opens a door for Google, but leaves it closed for everyone else. That fact has always been the absolutely critical feature of the settlement. It makes the settlement exciting, because it means that millions of out-of-print and orphan books would become much more widely available. It also makes the settlement dangerous, because millions of copyrights would be collectively press-ganged into Google's service.

Either way, it makes the settlement important. Google cofounder Sergey Brin compared the project to the Library of Alexandria. That's thinking too small. The switch to digital is, like the switch to movable type, a paradigm shift in the technology of text. It will have similarly large repercussions, and the Google settlement is making it happen well ahead of schedule. But delete the shift to an opt-out system from the deal and what remains is a big but fairly routine settlement of a copyright lawsuit.

The original settlement has opened our eyes to some exciting new possibilities and would do remarkable things to improve access to books. Some kind of digital access for out-of-print and orphan books along these lines would be a huge step forward for literary culture, closing the digital divide, and the future of information. But the fact that the settlement gets some things very right shouldn't isolate it from searching review of the things it doesn't. And the only way to make those distinctions is to have an informed, extensive, public discussion. Broadly speaking, there are three sets of reasons to worry about the settlement.

The first is whether having a single, central source for all these millions of books is good for society. Will it drive libraries to eliminate their print collections and become dependent on Google's institutional subscription, only to see its price rise uncontrollably? Will the FBI force Google to turn over its lists of who's been reading the Qur'an? If these kinds of broad-reaching policy decisions were being made by Congress, the legislative process would in theory take everyone's interests into account. But in a settlement negotiated by a handful of lawyers, the danger is always that the “public interest” means whatever they say it does.

A second is that Google-only exclusivity is an antitrust problem. No one else will be selling access to orphan works; it would be illegal for them to. That's a government-conferred franchise, and unlike the “monopoly” of copyright itself, it would be concentrated in Google's hands. It was gratifying to see the Department of Justice agree both with my belief that the settlement would create worrisome barriers to entry and with my proposed fix: open up the settlement to any competitor on the same terms Google would receive.

A third reason for concern is inherent in the class action procedure itself. Class actions can be a powerful tool for justice, offering compensation for mass harms and deterring future wrongdoers. But because class action settlements impose a judgment on all of the class members, there's a serious risk that their own lawyers will sell them up the river for a song and make millions of dollars in legal fees for doing so. Guarding against such abuses requires looking closely at how the settlement treats different classes of copyright owners, whether class members are receiving sufficient notice of its terms, and how the Book Rights Registry will handle the money it receives from Google.

All three of these concerns flow from that same central feature that makes the settlement what it is: the literally unprecedented way the settlement uses a class action to create an opt-out system for selling books. Here, the plaintiffs aren't just giving up the right to sue Google for scanning their books; they're also being shanghaied into a complicated commercial deal that includes a controversial allocation of electronic book rights and requires them to give up the right in the future to sue Google for plenty of things it hasn't even contemplated doing yet.

A comprehensive digital books database would be a very good thing, but expediency is no substitute for the rule of law. Even if this settlement is “fair, reasonable, and adequate” to copyright owners, as the law requires, it would still be important to make sure that it doesn't create a bad precedent for future class actions. What if Union Carbide had been able to settle the Bhopal lawsuits on terms that allowed it to keep on pumping out toxic methyl isocyanate in the future, provided it split the revenue with people living nearby?

Settlement 1.1

I had hoped that Settlement 2.0 would fix these problems, but it's turned out to be more like Settlement 1.1: one big feature cut and a bunch of small bug fixes. The feature cut is that most foreign books—those from everywhere besides the U.S., the U.K., Canada, and Australia—are out of the settlement entirely. The new class excludes continental Europe, Japan, and New Zealand, home to the squeakiest wheels the first time around. They wanted out of the settlement, and they got it.

The smaller changes are mostly welcome news. Gone is the frightful provision by which revenue earned by copyright owners who didn't claim their books would be diverted to other copyright owners who did. In its place, there's a fully funded independent fiduciary to search for orphan owners. Other changes address competitive concerns. Copyright owners have more flexibility to negotiate a revenue split other than 63/37; Google has more flexibility to offer discounts to readers on its own dime; the price-setting algorithm has been defanged; and the most-favored-nations clause in Google's favor is gone. Google's recent promises to allow Creative Commons licenses and to let others set up their own branded storefronts to sell Google-hosted books are now officially in the settlement itself. The definitions have also gotten a clean-up job: comic books, children's books, and sheet music are now more clearly not covered.

Still, the dark heart of the deal remains: Google will still have effectively exclusive access to unclaimed books. That's still a problem of concentrated power, it's still an antitrust issue, and it still raises grave concerns about using a class action in this way. The next few months will be quite interesting; the parties are optimistically calling for the notice, objections, and fairness hearing to be wrapped up by mid-February.

A fuller response to the amended settlement will come in a white paper being drafted by the Public Interest Book Search Initiative. The paper should be completed by the first week in December.