except when you are paying for their enterprise edition - and they make no distinction between paying customers (quite a bit actually once you add in actions costs + their other addons, it's quite a bit of money) and their free tier full of AI slop PRs on rando projects.
As someone who pays github, I'm super annoyed at the downtime and still I'm getting charged for actions running that all fail from their lack of concern for their enterprise customers.
Having some independent pools of capacity for paying vs. free customers isn't rocket science, it's basic competence (or care). I'm quite sure their are competent people remaining at GitHub (although I hear most of their original folks have left, and the 2nd tier of folks who joined from microsoft seem lackluster at best). I chalk this up to lack of care, or hubris, or both.
Oh I agree that the outages are unacceptable regardless of their “social media origin”.
I also don’t buy their excuse that it’s due to increased loads due to AI. If that were the case then we’d see more patterns to the outages. But instead it affects all GitHub systems but at different times. It feels more like a problem with their orchestration layer than capacity issues.