Including a little dive into the physics behind it

Shortly after Uber COO Andrew Macdonald said that it was “getting harder to justify” spending money on AI as it was “very hard to draw a line” from that spend to useful consumer features (after its CTO said Uber burned its entire annual token budget in four months), Axios’ Madison Mills reported that one company had accidentally spent $500 million in the space of a month on Anthropic’s models after failing to set spend limits. A few days later, Mills would report that other companies were now looking for ways to reduce their AI spend.

There is something about numbers that makes people’s brains stop working. This is common in medicine, where reflexively treating a lab abnormality without thinking an iota about the patient or even about the ground truth — is this number here “real” or is it a blood collection/lab analysis/data entry error? — is a phase most doctors go through and some never leave. Call it video game brain: confusing the hardcoded information of an RPG stat or a FPS health bar with more malleable values we get from physical measurements.

So I was uh, downloading some linux isos, like usual. It was going slowly, so I opened up the Trackers tab in qBittorrent and saw the following:

There are some excellent gems: it's worth reading the whole post! A few things that really stood out to me:

From an information theory point of view: We want to communicate concepts, but we have to translate them into words. This can lose a lot of information. Examples are another channel of communication, that can identify errors in translation

💯

A really interesting article by Simon Tatham (the PuTTY guy) about his preferences for collaborating on software.