I once went through security with a 500-1000 business-card-sized magnets stacked together in my carry-on.

They definitely wanted a closer look at those.

similar energy: “By seeing my client commit this crime, did the witnesses create this crime? I ask you this.” youtube.com/shorts/Ds9el...

gimmick idea: statistics lessons in 19c cartoon format, but carefully avoiding racist caricatures.

Explicitly leaning into this is one of the things I like about the Mad Max world.

And even when something is OA with the publisher and/or on personal web, putting it on arXiv is another publicity / dissemination channel. Anecdotally, I seemed to get higher early citation counts for stuff on arXiv, even if I didn’t post it there until it passed peer review somewhere.

Previous character in this setting was Bugs, an extremely ADHD private eye and former corpo data analyst who lived in a tiny apartment above a combination bodega / Radio Shack.

the character is named Deadlift.

Need someone dead? Need something lifted? Call Deadlift.

while insisting they are following the plain or literal interpretation 🤦

🙋🏻 yep. Such experiments might be measuring _something_, possibly even something interesting about the patterns in large corpora and models trained on them, but the bar to showing that thing is actually the social phenomenon of interest is very high.

Also have a short lecture talking about this — how are we preparing our students to avoid making such errors? cs533.ekstrandom.net/f22/week10/#...

Our forthcoming NeurIPS position paper, led by @aolteanu.bsky.social, makes this argument (along with several related ones) in more depth. Rigorous AI/ML work should flow from explicit and rigorous premises, not just have a final evaluation that checks some rigor boxes. arxiv.org/abs/2506.14652