Before diving in to any of the details regarding the actual meaning of "final approval" of the Settlement on 20 July, it's fairly important to revisit a couple of context items.
First, and perhaps most important, one must remember that any settlement is a least-worst-outcome compromise. By definition, a difficult-to-value harm has already occurred, and reducing that to certainty — a number of dollars, a clear promise to do/not do something in the future, compensating all of the professionals who facilitated reaching the settlement — is by its very nature an ephemeral agreement that outsiders can find fault with. The very existence of statutory damages in copyright actions (presuming a number of prerequisites have been met) represents recognition that a hypothetical infringer's accounting might not be sufficiently clear to determine "lost profits," let alone any other aspect of "valuing" a particular copyright infringement claim in the moment. Now multiply, or even exponentiate, that by multiple parties, let alone class actions, and one can reach a pretty clear conclusion: No matter what or how a copyright-infringement claim is resolved, via settlement all the way through trial, outsiders can always criticize it as either inadequate or excessive (sometimes both simultaneously!), and even the parties will remain dissatisfied. If there's one aspect of conflict resolution to which the aphorism "It's easier to ask forgiveness than get permission" does not apply, it's "control of individual expression."1 "Because it would be kewl" doesn't transfer well when an author's mortgage payment is on the line… or, more to the point, the author's ability to pay a professional or agent to review a prospective authorized use of their material in advance of making that blockbuster.
Second, and somewhat less abstractly, one needs to remember that just as the map is not the territory, a settlement — and especially a class-action settlement — does not define the interests of the parties at the outset. Instead, those interests relate to what is recognized and explicated by the parties. In this context, consider whose interests actually control (and that is a relatively simple example, because that author didn't work with a lot of coauthors), and it's incredibly easy to imagine that changing even between the infringement and the expiration of the statute of limitations. Authors are mortal, too… and their heirs and executors usually ill-qualified to judge or serve the interests of either the author or the ouevre. The interests of the individual authors — and other copyright holders — are at most an intersection, let alone a union or congruence, even before considering temporal-change issues. This should, but never does, mean that one should not map "interests served by a class settlement" onto "interests of an individual copyright holder… or infringer," let alone onto "Progress of… useful Arts" related to a specific instance of infringement of a specific work.2
Third, and last for the moment, any judicial remedy reflects the compromise of not allowing the perfect to be the enemy of the good (or good enough, or satisfactory, or less-bad-than-the-alternatives) when it applies to organizational conduct. The Bartz v. Anthropic matter makes it even murkier because there are multiple organizations, both formal and de facto, on both sides of the v.… including objectively "bad actors" on both sides. Consider, for the moment, a hypothetical supervisor (if that's not an overstatement of power/influence/responsibility) at Anthropic whose only role was to recommend against using this convenient database of material as a training corpus, but who was ignored/overruled, and how that supervisor's reality interacts with this settlement.3 What is having "Anthropic" on one's work history with a job title indicating this going to do in five years when applying for a job at Disney?
Put together, the real point here is "don't overgeneralize" — a generalization that is not itself reflexively too broad.
- Techbros generally haven't figured this out yet, because they conflate "identical means of storage and manipulation of information" with "the means of storage and manipulation of information determines the nature of the information." Those who have figured it out warp and/or deny it due to lack of relevant context and conflicts of interest. Consider, for example, the "context" that would be applied by someone whose only experience is abstract manipulation of a large dataset (Newtonian mechanics) to both the expression "That which we call a rose | By any other name would smell as sweet" and its connotations in context and generally (quantum mechanics regarding a single molecule, let alone its constituent parts… or presence and function in a cellular organelle).
More to the point, as a group techbros have a semiinstitutionalized tendency to overgeneralization, epitomized by shrieking of "Information wants to be free!" at communication by metaphor merely because the literal bounds of the metaphor can be recorded… and imputing motivation and conduct of some prominent bad-actor owners of copyright interests to all of copyright. CP Snow was far too optimistic when he divided knowledge-communities only binarially — there's a lot of fuzziness, and it comes from more than two sources.
- Put another way, the interests of whoever-owns-Simon-&-Schuster-today are not the same as those who wrote Star Trek™ novels, and contributed to Star Trek™ collections and nonfictional works, that appear in the data set of the Bartz matter. And there are more than a few.
- As foreshadowing of an issue that I will discuss down the road — I'm not even going to identify the potential conflict because that would give it away and activate the conflict — consider the publisher's claim in the Bartz matter based on a contractual agreement where the relevant contractual clause was itself unlawful, however customary and however "agreed to" by the actual author(s). Consider this the creepy music playing just before the teenagers agree to spend a night in the old haunted house down the road, notwithstanding their later choices (and that's an unduly optimistic view concerning authors!). Foreshadowing: Your mark of quality literature.