Judge Bibas: "Wisdom does not always find me, so I try to embrace it when it does—even if it comes late, as it did here"

Judge Bibas, sitting by designation in the District of Delaware, revises his prior summary judgment opinion.

There is a long-running dispute between Thompson-Reuters and Ross. TR alleges that Ross used the Westlaw headnotes to train a competing AI product. In 2023, Judge Bibas largely ruled against TR. But on the eve of trial, the case was stayed. Today, Judge Bibas issued a revised opinion. In short, he changed his mind.

The opinion begins:

A smart man knows when he is right; a wise man knows when he is wrong. Wisdom does not always find me, so I try to embrace it when it does––even if it comes late, as it did here. I thus revise my 2023 summary judgment opinion and order in this case.

Felix Frankfurter  wrote in a dissent, "Wisdom too often never comes, and so one ought not to reject it merely because it comes late." However, Frankfurter almost certainly did not make this quote up. Sir Arthur Conan Doyle wrote in a Holmes short story, "but it is better to learn wisdom late than never to learn it at all." And I'm certain others said it before him. Judge Bibas channels that higher power.

Bibas explains his new thinking on headnotes:

A headnote is a short, key point of law chiseled out of a lengthy judicial opinion. The text of judicial opinions is not copyrightable. Banks v. Manchester, 128 U.S. 244, 253–54 (1888). And even if it were, Thomson Reuters would not get that copyright because it did not write the opinions. But a headnote can introduce creativity by distilling, synthesizing, or explaining part of an opinion, and thus be copyrightable. That is why I have changed my mind.

Kudos to Judge Bibas. People should always be open to revising their thinking. And even better, they should explain why they changed their minds. This opinion demonstrates that virtue.

Comments

[steve sturm]

Is the (new) decision correct?

[David Nieporent]

I'm not convinced, but too much of the underlying material is under seal to get a good grasp on it. (The issue isn't West's headnotes in the abstract, but actual, specific headnotes, and we don't know which those were.)

[Noscitur a sociis]

I’m not saying this is the least self aware posting Prof
Blackman has made. But top ten?

[Doug Heffernan]

I'd like to see a list of the top ten things that Blackman has changed his mind about. A substantive list with drastic changes in opinion. Not something like changing his mind about Trump being a great president to instead thinking that Trump is the greatest president. That is not really changing to a contrary opinion.

[Judge Magney]

When I was in law school, one of my professors assembled his own (looseleaf) casebooks, rather than requiring us to buy more expensive commercial textbooks and the inevitable supplements. Most of the text was, of course, edited copies of judicial opinions, often from West reporters. West Publishing threatened to sue him, for including material from their headnotes. Not a word of text, mind you. Just the bracketed numbers. So far as I heard, they never followed through.

[Devin Watkins]

That explains why the headnote is copyrightable. But is merely training the AI on the headnote a violation of copyright?

[David Nieporent]

That is indeed a question in many of these LLM lawsuits — and I'm dubious that it does violate copyright to do that — and Bibas seems to distinguish this situation with little more than a handwave. ("This isn't software code, so you don't need to use their headnotes for training.")

I'm also a bit dubious on his copyrightability holding in the first place. (Though as I noted, it's hard to know since the specific headnotes are sealed.) Simply saying "Only minimal creativity is required" doesn't explain how this passes the threshold, and why the merger doctrine doesn't work here.

[wvattorney13]

That's where I come down on this. No doubt, say Blackman and his article is copyrightable. But if I am training an AI on Blackman and/or Reason authors in general, why is how I trained the AI a violation of the copyright?

It would be like a kid doing a book report being in violation of the book's copyright because the material he used in the report came from something subject to copyright.

[David Nieporent]

It would be like a kid doing a book report being in violation of the book's copyright because the material he used in the report came from something subject to copyright.

Well, that's what fair use is.

Here, Bibas's reasoning is that the use isn't transformative in any way — it's not generative AI — so it's just infringement. I'm not convinced, but I'd need to see more about what Ross was actually doing.