@johncarlosbaez
> a year that was 18 × 20 = 360 years long

"years" should be "days"

@robpike @sundogplanets
"The Tragedy of the Commons"

@licho
Indeed, there are many modern chocolate worshippers!

@johncarlosbaez

@johncarlosbaez
Well, yeah!

@johncarlosbaez
> From the ancient Maya texts, particularly the Popol Vuh, we learn that cacao was considered a divine gift from the gods. It was believed that the cacao tree was created by the gods to provide sustenance for humanity. The mythology surrounding cacao emphasizes its sacred nature, as it was often consumed in rituals and ceremonies designed to honor the deities. The Maya viewed cacao as a connector between the earthly and spiritual realms, making its cultivation and consumption a vital aspect of their culture.
> Chocolate played a multifaceted role in Maya society, influencing everything from social status to religious practices. It was primarily consumed as a frothy, bitter drink called "xocolatl," often flavored with spices, honey, or chili.

https://www.mexicohistorico.com/paginas/The-Maya-Use-of-Chocolate--A-Sacred-Drink.html

Making chocolate drinks with *chili* I recalled, but not honey. For the Mayans that is.

@ColinTheMathmo
P.S.:
> while ChatGPT and other "AI"s can't reason

They certainly don't reason as humans do. But just like old fashioned expert systems, they have *a* form of reasoning -- just a very strange one by human standards, and one that is strongly based on language.

You'll recall that arbitrary syntax *can* be Turing equivalent (semi-Thue systems), so they *might* be able to compute anything a Turing machine can do by such means. They don't, but it's a reminder not to underestimate that kind of thing.

https://en.wikipedia.org/wiki/Semi-Thue_system

At this point I also need to caveat that humans are not Turing machines: they can produce incorrect answers, unlike Turing machines, and your whole point is that LLMs also produce incorrect answers, so they are *not* Turing equivalent, although it makes them a little more similar to humans.

@ColinTheMathmo
Great example of why these things should absolutely not be blindly trusted.

That said, recommended best practice is to herd these critters carefully, by having the initial prompt include things like "If you don't know, say so, and if you think you know, provide explicit references and quotes" and so on.

BTW I'm glad you provided the answer, because I sure don't recall the series clearly after all these decades, even though I did read it more than once in my youth, both before and after he merged all his stories into one story universe.

@noplasticshower
If it's not too much trouble, that would be interesting and amusing.

@aka_pugs @lauren @SteveBellovin

@aka_pugs
Definitely a nice picture. Thanks.

@noplasticshower
Do you remember the subject of your talk? I'm just curious about the motivation, although it sounds like fun.

@aka_pugs @lauren @SteveBellovin