Curious to see that markdown table performed any differently than pipe-delimited, considering markdown tables are pipe-delimited also. How reproducible are these results? Also it would be nice to see accuracy vs input token usage compared in a 2d plot as there is a cost/performance tradeoff.

I appreciate the prompt and courteous reply. There's no real resolution here, I've already blocked my PTO with my employer and can't change. Just do better. Stop gouging plane ticket prices. You Wednesday flights were overbooked because your Thursday flights were over-priced. Do better.

The challenge is trying to figure out how this is relevant or useful content. Maybe don't scrape random websites to generate content and clicks.

Screenshot for anyone curious...

Yeah, crappy luck for the poor thing.

I can explain over drinks but likely the phenomenon is due to "eventual consistency" design decisions in how they built the thing.

I haven't studied up on how blue sky works, but being decentralized it might not be as "realtime" as other social apps. I could be wrong though.

futurism.com/elon-musk-ci...

Thanks for the clickbait title, a paywalled article, and nothing of substance... (Why am I following Washington Post??)