It's (a) they're under massively increased load because everyone's vibing up new projects these days, (b) they've been in a weird frankenstein "on azure but also we have our own control plane" state for years and they're pushing to no longer have that be the case.

I don't think vibecoding at Github has much to do with it.

▲

altern8 5 hours ago | parent | next [-]

Ah, yes. A lot more repos, commits, and most importantly huge PRs.

That makes sense. Thank you!

▲

gilrain 5 hours ago | parent [-]

No, it doesn’t. Their competition is not similarly unstable, despite existing in the same world of LLMs. Think critically.

▲

datsci_est_2015 5 hours ago | parent | next [-]

Devil’s advocate, Pareto heuristic would let us speculate that 80% of LLM traffic would be aimed directly at the largest provider, i.e. GitHub.

▲

abejfehr 5 hours ago | parent | next [-]

I think it’s much more than 80%, it’s probably the default recommendation and folks who aren’t technical would just accept it. Probably closer to 95% or more

▲

necovek an hour ago | parent | prev | next [-]

Isn't the relative increase more of interest? If someone was only owning 10% of the market, and they've only gotten 8% (percentage points) of the 20%-not-GH LLM-related increase, they'd still be seeing a very similar spike compared to their baseline as GitHub.

▲

gilrain 5 hours ago | parent | prev [-]

Your speculation is that their competitors would naturally not see a commensurate increase in instability while “only” handling 20% of the same crisis?

I don’t buy the excuse. I want to hitch my wagon to those “mysteriously lucky” competitors. (And have. And haven’t had similar issues to Github, since.)

	▲	datsci_est_2015 2 hours ago \| parent \| next [-]
		Competitors would be long tail, so a different mode of traffic entirely. Maybe they get spikes that are more easily whack-a-moled than the constant hammering that GitHub receives. Tough to say as this is all speculative, though.
	▲	porridgeraisin 2 hours ago \| parent \| prev [-]
		It's probably a threshold thing isn't it? You wouldn't get 20% of the effect at 20% of the traffic. There's a step function in there somewhere.

▲

vitally3643 2 hours ago | parent | prev [-]

Their competition doesn't have nearly the same scale of traffic because they don't have nearly the same scale of users or network effects.

Think critically.

▲

ModernMech 4 hours ago | parent | prev [-]

I started using an agent (Codex) on my repo and it went from a a few dozen clones to thousands (3383 this week). I dunno what the agents are doing to clone the repo so many times -- I'm not running 3000 agents or prompts, maybe 10 or so this week. But if this is typical, a 1000x increase in usage across the board can't be good on the system.

	▲	12_throw_away 2 hours ago \| parent [-]
		> I dunno what the agents are doing to clone the repo so many times agentic "ai" is going great