I have spent months on projects that benefit a small subset of a service that's a small dependency of another service that's ultimately only used in emergency/outage situations.

It was absolutely essential for the company to have these systems in place, but I was under no illusion that I'd actually see them used during my time in the team because disasters of the necessary magnitude are rare.

So seeing the user journey and understanding the importance did nothing for my feeling on disconnection from what I'm working on.

So I emphasize with the original poster a lot on this.

Yes, absolutely, because all "we" would be adding to the conversation is speculation. No new information can be gained here.

Question: Does this take into account memory bandwidth and caches between cores? Because getting them wrong can easily make parallel programs slower than sequential ones.

Even the first number he gives ("688,478 Excess Non-Covid Natural Cause Deaths (primarily from the Covid Vaccine)") is complete crackpot fiction and runs counter to the actual current evidence. In fact the very likely reason for this is miscounted/misattributed COVID deaths.

See https://pubmed.ncbi.nlm.nih.gov/38300867/

Let me state this again, I do not trust them as a source. I can't find any information regarding any Safire and llnl collaboration as well. Especially nothing from the llnl side.

So as long as no reputable independent team is a able to verify their claims, I'll remain extremely sceptical. So far all we have are wild claims and fancy videos all from a single source and that just won't cut it to convince me.

I hope it'll beat btrfs eventually though.

I don't have access to copilot so I can't compare but I'd wager it works a lot better for Google internally because the training data is customized to Google.