Totally!

I lean into "this is happening / already happened" because there's no point discussing the aesthetics of airplanes with an audience that hates flight

But my actual motive for being interested in language models is that "interactive, tunable libraries" is a dizzyingly attractive idea

I think implied license with fairly broad scope is plausible, and fair use would certainly have very broad scope here, too. If Warhol is key, infringement claim doesn’t pass the laugh test. In what market would ICE be substituting its work for the artist’s? Hard to imagine.

And here are shots of the 2025 thread (with alt text, ofc).

2021 X thread screenshot, thrilling conclusion (2/2)

Screenshots of the 2021 thread, for folks who can’t see it because of X’s stupid policy of hiding threads from non-users (1/2):

Yikes - thanks for the heads up! I’ll see if I can find a workaround. Love to have my own words locked up in a platform’s walled garden…

Bottom line, these critics are trying to make the project look "controversial" or "partisan," but all they're really proving is that their own partisanship isn't sufficiently catered to in the document. That should be reassuring to anyone who wants unbiased guidance.

In 2025, when the maximalist caucus resigned en masse from the project, I responded again, explaining why it might be in their interest to prevent the publication of a doc that will help judges make decisions based on legal doctrine rather than property rhetoric.
x.com/bc_butler/s...

In 2021, I responded to the idea that you can't restate a statute because, unlike the common law, it speaks for itself. This idea is manifestly silly, barely passes the laugh test. If the law's so clear, why are there 3 treatises and a 1000+ page Copyright Office Compendium?
x.com/bc_butler/s...

to do digital research, even digital research that is critical of AI, Big Tech, or any other bête noire you like. Anyway, that's a long-winded way of saying I was glad to see @CommonCrawl.bsky.social respond to the latest Atlantic piece: commoncrawl.org/blog/settin...

Since fair use copying is a necessary step in almost all digital research, when you embrace "I caught somebody copying!" as a newsworthy event, you inevitably end up smearing and intimidating resources that make it even conceivable for anyone*other than 'Big Tech'*

is, "Look Who We Caught Copying!" it was only a matter of time before this campaign resulted in an overheated "exposé" targeting an open resource like Common Crawl. In addition to just being crummy yellow journalism, this approach actively undermines its own stated goal of taming "Big Tech."

. @TheAtlantic.com has published a steady stream of this stuff, stoking moral panic by characterizing as rank piracy activities undertaken in good faith belief that they are fair use—a belief largely vindicated in the courts. Because the TL;DR of every article in this genre

if you ignore this subtlety, exploiting it to publish "revelations" and "scoops" that amount to "Look: technologists treat our art as data! It must be because they want to DESTROY ART for BIG TECH!" That's the modus operandi of too many journalists right now.

Both are right. Creative works are both art and data. Balanced copyright allows both of these important communities to thrive. It protects commercial creators from piracy, but enables research and innovation unfettered by rightsholder control. But you can get a lot of clicks