About
We read the status pages so you don't have to.
Every 15 minutes, a script on our server reads the official public status page of each service below and translates it into the only question anyone is asking.
How it works
We don't test whether anything works. We read what each company publishes about itself, in the machine-readable form most status pages offer, and pass it on, translated. If their page says operational, we say NOT FUCKED. If it says major outage, we say FUCKED. Where their page sums itself up in a sentence, the card quotes it word for word. Where it only publishes a feed of incidents (AWS, Google Cloud, Azure, Datadog US1 Fed, Microsoft 365 and Slack), the card says what the feed shows, in our words, and lists the incidents it names. Either way, each card links to their page so you can check our work.
The big answer on the board is the worst state of any service: one fucked service and the internet is FUCKED. If we can't read more than half of the pages, the answer is FUCKERY UNKNOWN.
The board's history and the Hall of Fuckery start from the day we started watching. They're made from our own checks, not the companies' records.
The translation
| Their status page says | We say |
|---|---|
| Operational | NOT FUCKED |
| Degraded performance | PARTIALLY FUCKED |
| Partial outage | PARTIALLY FUCKED |
| Major outage | FUCKED |
| Maintenance | SCHEDULED FUCKERY |
| Status unavailable | FUCKERY UNKNOWN |
Whose pages we read
- Cloudflare
- GitHub
- npm
- Docker
- DigitalOcean
- Linode
- Vercel
- Netlify
- Datadog US1 Fed
- AWS
- Google Cloud
- Azure
- Discord
- Microsoft 365
- Slack
- Zoom
- Dropbox
- Jira
- OpenAI
- Claude
Not every company uses the same kind of status page, so we read some a little differently. AWS: an event in one region is PARTIALLY FUCKED, and a disruption in three or more regions, or a global one, is FUCKED. Google Cloud: an incident they rate high severity, or call an outage, is FUCKED; if every open incident is only informational, that's NOT FUCKED; any other open incident is PARTIALLY FUCKED. Azure: their feed only lists incidents with wide impact, so any open one is PARTIALLY FUCKED. Microsoft 365: their public page is only updated when an issue is big enough to keep admins out of Service health, so it misses smaller problems; service degradation there is PARTIALLY FUCKED. Datadog US1 Fed: we take the worst status of any of its components.
Questions
Where do these statuses come from?
From each company's own official public status page. We read what they report and translate it. We don't independently test whether a service works.
How often does this update?
Our script checks every 15 minutes, and the board picks up each new check on its own without a reload. If our checker hasn't reported in over 45 minutes, the board says so. That's our own fuckery, and we'll own it.
Why has one service been PARTIALLY FUCKED for days?
Big vendors report hundreds of components, and one of them is usually off somewhere. A problem in one region or one product keeps the label up until they mark it resolved. It doesn't mean the whole service is broken. The card lists what their page says is affected.
Their page says operational, so why does the card list incidents?
Some status pages keep reporting overall operational while they work on smaller incidents. We go with their overall status, and still show you the incidents they listed.
What does FUCKERY UNKNOWN mean?
We couldn't read their status page. Their page didn't load in time, or answered in a form we didn't understand. It doesn't mean they're broken or fine. It means we don't have a readable report.
Are you affiliated with any of these companies?
No. We're not affiliated with, endorsed by or connected to any of them. We use their names only to say whose public status page we're reading. Their statuses are theirs. The swearing is ours.
Does this cover the whole internet?
No. We watch about twenty services that a lot of other things depend on. The internet declined to fit in a grid.
Can I use the data?
The raw files are public: status.json (right now) and history.json (the last 90 days). Please don't fetch them more than once a minute. They only change every 15 minutes anyway.