Those damn humans! Google has revealed the cause of an unprecedented, 58-minute malfunction yesterday that saw every site on the web flagged harmful. 

For around an hour early Sunday morning, the whole internet was deemed bad. Any search result typed into Google prompted the warning: “This site may harm your computer”.

If you tried to click through to a site regardless, the search engine threw up an interstitial screen to try and further dissuade your visit. A server error message met anyone who tried to click on a link that promised an explanation.

Employee malfunction
On its official blog, Google says it uses both automated and manual systems to maintain a list of websites that that may host malware.

An employee mistakenly checked a “/” symbol while typing the address of a suspect site, Google says in its official blog. And the system was designed in such an apparently dubious way that the typo caused every site on the net to be labelled potentially harmful.

The episode was an embarrassment for Google, whose search engine has previously enjoyed a nearly flawless reputation.

It also drew a heated response from the nonprofit StopBadware.org, which says Google tried to pin the blame on it in initial explanatory posts. The company said in an online statement:

“Google has posted an update on their official blog that erroneously states that Google gets its list of URLs from us. This is not accurate. Google generates its own list of badware URLs, and no data that we generate is supposed to affect the warnings in Google’s search listings.”

Nevertheless, interstitial warnings on Google search results directed users to StopBadware.org during the malfunction. Millions of impressions soon swamped the non-profit’s servers, taking it offline.

Few New Zealanders would have noticed the glitch, which happened at 4.30am Sunday our time. But on the US east coast it was 10.30am Saturday morning and the net was soon alive with posts about the multiplying errors – annoyingly for Google, often through it rival Twitter.

Google says it fixed the error in rolling updates, meaning most users would have been affected for around 40 minutes.