The cover art alone is almost enough to convince me. @jdebbiel.bsky.social did a fantastic job here.

There's nothing wrong with not liking retellings, but you can pry them from my cold, dead hands. Modifying stories for the audience in front of you is how all stories used to be told. It'd be a shame to forget that and ignore wonderful works like Lavinia.

no

Sadly, you missed a chance to go through the hazing ritual of a first edition dragon book. Never has a book been more aptly named.

Not a defense of LLMs, but the surprise of them was how much of what previously appeared to require "intelligence" (whatever that means) was reproducible by this dumb process that's barely more sophisticated than a Markov chain.

I'm surprised I've never seen this considered. A big enough publisher would benefit from smaller, regular installments. Not least because the payments pool would generate interest.

Camera thickness is constrained by physics. They're already pushing the boundaries. They can either make the phones thicker (worse sales), make the camera worse (*significantly* worse sales), or have a protruding module that people will compensate for with a case if it annoys them.

Perhaps they misunderstood the distinction between "usually locked" and "permanently locked". The doors only open a couple times a year for regularly scheduled shipments. They're locked otherwise.

The part I find most galling is that this was purely a case on piracy. The part where they trained on digitized secondhand books was declared fair use, as was training itself.

It's worth emphasizing that this lawsuit has determined that *training* is totally legal. Anthropic is paying the fine only for the books it pirated. The other strategy of buying secondhand books and digitizing them for training was considered fair use.

A weirdly large number of people are convinced disease and poverty are just moral failings in disguise. They often see public health measures that don't address the perceived moral failings as either eliminating the natural consequences or missing the root cause.

Oops, I should have read closer

The contest was OpenAI, not Meta, but there's a straightforward motivation: marketing. They want a veneer of scientific respectability for their products and tossing a few token prizes at a well-known competition site is institutionally easy. They didn't even consider second order effects.