Say that someone alerts Amazon that a certain book that it is selling contains a libel — or discloses information that invades someone's privacy, or some such. And say the matter is clear, and Amazon has no doubt that the book is indeed libelous. Under traditional American libel law (which is quite constitutional under the First Amendment), Amazon would at that point be legally actionable for Amazon to continue distributing the book.

Bricks-and-mortar bookstores have in fact long been potentially liable for distributing libelous books once they learned that they were libelous, though in practice such lawsuits have been rare. It's also possible that they could even be enjoined from selling the book, once there's a final finding by a court that the book is libelous; they could certainly be sued for damages. It's possible that 47 U.S.C. § 230 would shield online bookstores from this liability rule, though it has long been applied to bricks-and-mortar bookstores. But let's assume that Congress modifies § 230 to track traditional libel law rules. Or let's say that this arises in England, or some other country that doesn't have § 230.

My question: Should it be proper for Amazon to at that point delete the libelous books from users' Kindles? Let's say that Amazon tried to add a provision to its user license agreement that would give it the contractual right to do so. Should consumers be very troubled by that? Or should they react the same way they react to the fact that if a Web site operator is alerted that some of his writings are libelous, he must change them or face the risk of liability? (That's just normal libel law.)

I would be troubled, partly because I wouldn't be at all certain that the judgments about what is and is not a libel will be made accurately. That's always a risk with any libel law system, but I would think it's even more egregious if it can affect existing copies and not just the distribution of future copies. I also would worry that the same would end up applying to books that contain alleged "hate speech," or blasphemy, or insults to political officials, or whatever else some country that has jurisdiction over Amazon might insist on. Again, the ability to change existing copies in customers' hands strikes me as materially more dangerous than just the ability to demand that people who post things online change their postings. But would I be correct?

Incidentally, what if instead of deleting the material, Amazon changed the file to include an explanation that this and that passages are false, with details about why they are false? (Of course, that would not be helpful in invasion of privacy cases, and many people who are libeled would prefer that the libels of them be entirely deleted, rather than just supplemented with correcting information.) Should we also be troubled about that, either on its own terms or because of the possibility that it would lead to broader editing? Or should we welcome it as a means of providing more information to readers, as well as mitigating in some measure the harm to the wrongly defamed?