The day Google broke the Internet

By Andrew Clifford

The recent problems with Google are a wake up call to keep the Internet as a tool of freedom, and not let it become a tool of control.

The entire Internet broke for about 30 minutes on Saturday 31 January, or so it seemed to many people.

My father was one of them. He telephoned to say that he could not get on to any website, that they had all been labelled harmful. He joked that it was a takeover by a totalitarian regime, who had banned all his favourite web sites.

What really happened was rather less sinister. Google maintains a list of badware sites that are known to contain scams, viruses and other malware, and warns users before they click from their search results onto these sites. It is a very useful feature. But for a short time on Saturday, it blocked all websites.

Apparently it was a simple human error. Something as simple as adding a "/" to a list of badware sites.

The problem is that Google has become fundamental to the Internet. Not only does it have an 80% share of Internet search, people now use it instead of the address bar. Instead of typing http://news.bbc.co.uk, people type "bbc news" into Google. So this was not just a problem with search. It was, to many people, a complete breakdown in the Internet's addressing system.

From our technical perspective, it is easy to make excuses for this sort of problem. We have all seen, or even caused, big IT interruptions. We know it was only a temporary application problem, and not a fundamental problem with the Internet.

But this technical perspective blinds us to something more serious. From its inception, the internet was designed to be resilient, to cope with the loss of servers and the loss of network connections. The addressing system is similarly resilient, based on multiple name servers without a single point of central control.

All this engineering was undone, albeit only temporarily, because somebody added a "/" to a list of badware sites.

This is a wake up call. This simple human error shows how fragile the Internet could become. The vast majority of Internet search, addressing, email, chat, mapping, and so on, are now controlled by a small number of large players. It may not have been the act of a totalitarian regime, but Saturday's incident shows how this concentration of control could be used to take over the Internet.

I have nothing at all against Google, but the Internet is too important to have this sort of failure and to run this sort of risk. I do not believe the answer is more control and regulation, but more of what the Internet is built on: standards, and multiple, independent implementations.

We, who have at least some knowledge of the Internet, have a serious responsibility. We must use and promote standards, especially standards that encourage multiple, independent but interoperable implementations. We must avoid single points of failure and single points of control, however convenient and benign they are. Only this will keep the Internet working, and keep the Internet free.

Subscription