The problem is the source of the reports and display of the reporting.
I'd trust Down Detector a lot more if it was filled with Hacker News community -- people who are able to understand that there's "DNS" and "Routing".. and that your phone can have internet access at home while your home PC does not.
I personally hate Down Detector's graphing because it can make it 'look' like there's an issue when there isn't really... Facebook with 500,000 reports looked as down as Google with 1,000 reports... For equally sized / used entities, I would not trust that "Google" is down with 1,000 reports. I had a coworker ask me what was going on with the internet because "everything is down.. Facebook, google, gmail, microsoft!" (when seeing the Down Detector home page)
DD should normalize the graphs against the service history in some way. A service shouldn't spike because it had 30 reports / hour for a day, then suddenly has 100... when it has a history of being out with 100,000+ reports. The 100 reports are probably mis-reporting, but you can't tell until you dig into each service, one by one, with separate page loads.
I'd trust Down Detector a lot more if it was filled with Hacker News community -- people who are able to understand that there's "DNS" and "Routing".. and that your phone can have internet access at home while your home PC does not.
I personally hate Down Detector's graphing because it can make it 'look' like there's an issue when there isn't really... Facebook with 500,000 reports looked as down as Google with 1,000 reports... For equally sized / used entities, I would not trust that "Google" is down with 1,000 reports. I had a coworker ask me what was going on with the internet because "everything is down.. Facebook, google, gmail, microsoft!" (when seeing the Down Detector home page)
DD should normalize the graphs against the service history in some way. A service shouldn't spike because it had 30 reports / hour for a day, then suddenly has 100... when it has a history of being out with 100,000+ reports. The 100 reports are probably mis-reporting, but you can't tell until you dig into each service, one by one, with separate page loads.