atherton33

"Mythos-like" just means "hyped via hearsay". It's being used correctly here.

Yes, the strategy depends on lack of effort from other senders, even trivial effort

The latter. Using chatgpt to write their chat messages usually. Emoji, arbitrary bold and italics, bullets, etc.

Tangent to the tangent!

I've started using it professionally because it signals "I wrote this by hand, not AI, so you can safely pay attention to it."

Even though in the past I never would have done it.

In work chats full of AI generated slop, it stands out.

E911 would get bssid of the access point.

I agree with you about what's described here.

There is engineering when this is done seriously, though.

Build a test set and design metrics for it. Do rigorous measurement on any change of the system, including the model, inference parameters, context, prompt text, etc. Use real statistical tests and adjust for multiple comparisons as appropriate. Have monitoring that your assumptions during initial prompt design continue to be valid in the future, and alert on unexpected changes.

I'm surprised to see none of that advice in the article.

My first job we had office rooms shared by 2-4 that were usually pretty quiet with a generally closed door policy.

The VP Eng would always say "I always try to remember it costs the company over a hundred dollars for me open one of these doors."

I learned so much from that boss.

I think they're saying something more subtle.

In your basketball analogy, it's more like they have a model that predicts basketball performance, and they're saying that model should predict performance equally well across groups, not that the groups should themselves perform equally well.

I don't know the detail here, but in many of the discussions I've seen the operators themselves are based in the UK, and that changes the calculus.