It wasn't a request routing outage; apps running on Fly.io didn't stop running. It was a deployments outage. For reasons passing understanding (I am reliably informed I'm wrong to complain about this), our website is the same Elixir app as our dashboard, and the dashboard got redeployed at one point. Our website being down is not the same as the whole service being down, though I guess there's a truth-in-advertising poetry to it being down when deployments are busted.
A lot of apps did stop running - https://community.fly.io/t/fly-io-site-is-currently-inaccess...
The entire API was also unusable, not just deployments.
Sorry, you're right: pretty much any time I'm saying deployments are blocked, I'm really saying the API was down.
I'm not sure if your explanation is comforting or disconcerting.
Why not both? Tell me what's comforting and I'll tell you why you shouldn't be comforted; tell me why you're disconcerted and I'll tell you maybe something else. All we can do is be straight about things.