Graham

Awkward enthusiast. I guess I’m smarter than a turtle.

[Graham @GrahamVL]

@matthewqueen84 @hedgequote What do you think is happening or about to happen?

[Graham @GrahamVL]

@TwistNH7 That might be the one that that could get me to change my allegiance. Losing Zima was a dark day in world history.

[Graham @GrahamVL]

@TwistNH7 @solsrvca Jokes aside, my current favorite is Seth, but all time is Foley.

[Graham @GrahamVL]

@TwistNH7 @solsrvca Seth Rollins is a fed? I knew there was a a reason he wasn’t my fave.

[Graham @GrahamVL]

@TwistNH7 Yeah, but not for this reason.

[Graham @GrahamVL]

@TwistNH7 @WrestlePurists He’s using both, when he got a new contract he insisted on Ron Killings so he can use his real name to promote at cons going forward. But R Truth is more recognizable still.

[Graham @GrahamVL]

@brianlfrye I think their guts are right, but more limited than current evidence allows. It's like reverse hallucinations, seeing infringement where it isn't (yet) in a higher volume than there is likely to be.

[Graham @GrahamVL]

@brianlfrye I think that's where the other factors come in, difficult to prove what will be created by AI and may be that the plaintiffs were both too early and lacking evidence Chhabria wanted to see.

[Graham @GrahamVL]

@brianlfrye Everything I know about copyright I learned from a courthouse dinner presentation by the lawyers from Spirit v Zeppelin, which is dangerous when talking to someone who actually knows enough to have his own doctrine of where the law should be going.

[Graham @GrahamVL]

@brianlfrye My understanding is that substantial similarity is a proxy for actual copying. Meta admits it used the licensed works to generate similar material, isn’t that actual copying that doesn’t usually come up? This is my armchair understanding.

[Graham @GrahamVL]

@brianlfrye If there was evidence of declining book sales with contemporaneous requests for AI to generate stories in the style of the specific author would you buy that as evidence? Expensive pipe dream to prove.

[Graham @GrahamVL]

@brianlfrye Do you see a difference between training on licensed works and not licensed? I think Chhabria is frustrated by bad arguments and going further than he should in that frustration, but I still think there’s some there there.

[Graham @GrahamVL]

@brianlfrye I think there’s an assumption that the AI is generating a similar work with a similar purpose. Without that, it doesn’t reach dilution, right? I think this is also the distinguishing issue for the “intermediate copying” cases. As a copyright outsider, I definitely don’t know.

[Graham @GrahamVL]

@brianlfrye I agree using licensed works to generate diluting work is likely some kind of market substitution and requiring licensing fees to train off copyrighted works fixes the issue. I don’t think that’s too far aground on 1A.

[Graham @GrahamVL]

@brianlfrye I’m mostly a 1A hobbyist and barely an observer in copyright, so I can’t say how that factor has been applied. I think there’s strength to the argument that AI dilutes the market for original works. Enabling the creation based on unlicensed works is fine on that end.

[Graham @GrahamVL]

@brianlfrye Specifically here, it’s about this specific opposition, not the issue.

[Graham @GrahamVL]

@brianlfrye My read is that these plaintiffs didn’t show dilution, not that they couldn’t the opinion is pretty negative on the litigation approach, not the legal standard.

[Graham @GrahamVL]

@OrinKerr @DougtheLawyer Orin, you’re invited to a lot of parties. You had an annual auction event where multiple people bid thousands of dollars to go out to bars with you for a single night. (Along with two other professors who were also invited to parties)

[Graham @GrahamVL]

Evidently the bar I’m in has a standing threat to kill its patrons.

[Graham @GrahamVL]

@TwistNH7 We’ve been loose acquaintances for 18 years now.