It's not the whole story. The biggest change is actually the internal rules for how downtime was reported, it wasn't actually such a large change in the actual reliability then.

Hosting that on GitHub Pages is ballsy.

Heatmap of the past year: https://isgithubcooked.com/

It got worse around April/May this year, while that graph stops at February.

God that's sad. I almost feel bad for all the engineers there, though I'm sure they made good money and probably left

That's missing a some of the LLM era? Maybe the author is working on a logarithmic view?