When Google sneezes, the internet gets flu

Algorithms don't help solve the net's biggest problems

We can remember it for you wholesale

The second issue is so unmentionable it's rarely raised in polite company. Again, it's an area in which we find Google not to be the unique company facing the problem - but uniquely ill-equipped to deal with it.

It's the question of how much we are prepared to disclose to an anonymous, friendly looking computer system.

Earlier this year, AOL Research quite deliberately and not without some pride, released the search queries of more than half a million users. It wasn't long before the anonymised queries were matched to their authors. AOL then expressed its horror, and dismissed the staff who it had encouraged to disclose the information (a fine example in its own right of corporate responsibility).

But the cat was out of the bag. The story confirmed that people today choose to disclose information to Google that they wouldn't tell their husband or wife - and a search engine never forgets this intimate knowledge. They would almost certainly not disclose it if they thought it would be released - or available for casual perusal by cops (which it is).

Similarly, bloggers who blurt away only to be "discovered" are often shocked to learn their writing was visible. Did they think they had some super-selective invisibility cloak?

Technology evangelists of a utopian bent - people who believe this great detritus of disclosure now being collected by information systems such as Google will prove to be of great importance to us - argue that the future is safe. We'll adapt to the machines, they say. But history tells us that the opposite is true: computer systems that fail to be trusted, fail to be used.

Google's own experience corroborates this.

Usenet is a medium not too dissimilar to many blogs today, where people wrote informally. The half-life of a Usenet posting was several weeks - it depended on the popularity of the newsgroup - but in most cases archives were not maintained, or easily accessible. And Usenet continued in rude good health long after its demise was predicted. Then, Google took over the creaking archive from Deja, and made searching the entire historical record trivially easy. People simply stopped using it. Usenet died the day Google turned it into a database.

Technical assurances are little use when the trust is lost - or was never there in the first place.

In another example, few countries think a nationally-accessible medical records database is a good idea, and the revolt in the UK against the proposal to create one has led to an opt-out campaign and some (cosmetic) concessions from the government.

People expect records to be computerised - but they don't want them to be confidential on a need-to-know basis - and not available to any browsing employer, insurance company, bureaucrat, or cop, which they effectively would be given the current state of security.

The saga briefly resurrected HR.4731, Rep Markey's bill to "Eliminate Warehousing of Consumer Internet Data". Then Markey went back to lobbying on behalf of Google for "net neutrality" - and the bill remained stalled.

Google's response to this issue has been a public relations offensive and some muscular lobbying - to prevent more bills like Markey's.

"We are reasonably satisfied...that this sort of thing would not happen at Google, although you can never say never," was Google CEO Eric Schmidt's response to the AOL scandal.

Obsessively secretive, and determined to hoard every piece of data it mines from us, Google appears ill-equipped to restore confidence in the relationship between surfers and the systems we use.

The company initially fended off attention by pointing to its inate goodness. These days it points to its own cleverness. Neither virtue nor engineering talent can solve either problem, however.

Regulation looms in both areas. ®

Sponsored: How to determine if cloud backup is right for your servers