Skip to main content

Like Apple’s Siri voice-activated question -and-answer software, Google Now also relies on being able to tap into large bodies of structured data – reams of organised information that can be polled for answers, rather than trying to trawl the web for documents that might contain the desired information.

There are two common complaints about the sort of radical personalisation that Google is attempting with its new Google Now mobile search service. One is that it doesn't work: the other is that it works too well.

For the sake of Google's bottom line – not to mention all the other Internet companies pushing up against the frontiers of the mobile Internet – the latter is by far the preferable error to make.

Google Now is an oxymoronic idea that might best be called "push search". If it works as advertised, your phone could one day automatically serve up answers to your most pressing questions – before you even think of asking them.

For advertisers, who live with the perennial conceit that they have the perfect product or service to satisfy your every unexpressed want, this is a powerful idea. Google says it has no plans yet to put advertising into the service, though this seems inevitable.

The "works too well" argument holds that people are not ready to accept the fruits of personalisation that technology has made possible. According to this view, a machine that "knows you" will simply be too creepy.

The "doesn't work" complaint, on the other hand, is strongly voiced by author and internet commentator Doc Searls. In his new book The Intention Economy, he argues that amassing piles of data about Internet users to "understand" them completely will ultimately prove futile.

My quick test of Google Now – which will officially be released for a number of handsets running Google's Android software in mid-July – suggests the technology has a long way to go.

This week, it offered baseball results for the San Francisco Giants, yet I am a fan of the rival Oakland A's; a schedule of bus times, but I commute by train and never take buses; and details of a women's clothes shop across the street.

Over time the technology will doubtless learn more about the user, but the gap between current performance and anticipating what I might want to know is a wide one. As anyone who looks at the supposedly "targeted" adverts that come to them over the Web can attest, online personalisation is a much stronger idea in theory than in practice.

Google certainly starts from a strong position, with a wealth of data about users drawn from its different services. By changing its privacy policies this year, it freed itself to amalgamate all of this in one place to power services such as Now.

Google+ is the latest addition to this: the social network might be a poor second-best to Facebook, but more than 250 million people have so far set up accounts. Even if they don't spend much time on the network, this basic profile data alone is a powerful supplement to Google's knowledge. Mobile devices add the new, all-important dimension of place – and with 1 million new Android devices being activated a day, Google is starting to dominate in terms of sheer numbers. Knowing where someone is at a certain time of day – and how they habitually spend their time – makes it possible to guess what they will do next.

Like Apple's Siri voice-activated question -and-answer software, Google Now also relies on being able to tap into large bodies of structured data – reams of organised information that can be polled for answers, rather than trying to trawl the web for documents that might contain the desired information.

Google's recent demonstration of the "knowledge graph" of organised information it has amassed was a reminder that information retrieval remains one of the company's core strengths. There is certainly a lot riding on this audacious experiment. The key to Google's prosperity has been its ability to divine a user's intent, thanks to the simple search box. That has produced one of the most effective forms of advertising ever invented.

In a world of mobile apps, however, the search box vanishes. In its place, Google has been left to infer intent, using all the technological tricks at its disposal. "Intelligent agents" that work on behalf of users and "push" technologies that bring just-in-time information were ambitious ideas from the early days of the Internet that did not live up to the hopes. Google is now as well placed as any to try again – and with the rise of mobile, it has an urgent need.

The recent history of rapidly changing privacy norms suggests that users will be open to the idea – but only if the benefits are great enough. What seems creepy or odd today can quickly become part of the normal context of life. It is now up to Google to prove that it really can move beyond the search box.

Interact with The Globe