I always got the impression that downdetector worked by logging the number of times they get a hit for a particular service and using that as a heuristic to determine if something is down. If so, that's brilliant.

It's brilliant until the information is bad.

When Facebook's properties all went down in October, people were saying that AT&T and other cell phone carriers were also down - because they couldn't connect to FB/Insta/etc. There were even some media reports that cited Downdetector, seeming without understanding that they are basically crowdsourced and sometimes the crowd is wrong.

I think it's a bit simpler for AWS- there's a big red "I have a problem with AWS" button on that page. You click it, tell it what your problem is, and it logs a report. Unless that's what you were driving at and I missed it, it's early. Too early for AWS to be down :(

Some 3600 people have hit that button in the last ~15 minutes.