31 October 2008

Watching Google watching the world

Filed under: books, collaboration, Google, infallibility, mobile data — David Wood @ 6:07 pm

If there was a prize for the best presentation at this week’s Informa “Handsets USA” forum in San Diego, it would have to go to Sumit Agarwal, Product Manager for Mobile from Google. Although there were several other very good talks there, Sumit’s was in a class of its own.

In the first place, Sumit had the chutzpah to run his slides directly on a mobile device – an iPhone – with a camera relaying the contents of the mobile screen to the video projector. Second, the presentation included a number of real-time demos – which worked well, and even the ways in which they failed to work perfectly became a source of more insight for the audience (I’ll come back to this point later). The demos were spread among a number of different mobile devices: an Android G1, the iPhone, and a BlackBerry Bold. (Sumit rather cheekily said that the main reason he carried the Bold was for circumstances in which the G1 and the iPhone run out of battery power.)

One reason the talk oozed authority was because Sumit could dig into actual statistics, collected on Google’s servers.

For example, the presentation included a graph showing the rate of Google search inquiries from mobile phones on different (anonymised) North American network operators. In September 2007, one of the lines started showing an astonishing rhythm, with rapid fluctuations in which the rate of mobile search inquiries jumped up sevenfold – before dropping down again a few days later. The pattern kept repeating, on a weekly basis. Google investigated, and found that the network operator in question had started an experiment with “free data weekends”: data usage would be free of charge on Saturday and Sunday. As Sumit pointed out:

  • The sharp usage spikes showed the latent demand of mobile users for carrying out search enquiries – a demand that was previously being inhibited by fear of high data charges;
  • Even more interesting, this line on the graph, whilst continuing to fluctuate drastically at weekends, also showed a gradual overall upwards curve, finishing up with data usage signifcantly higher than the national average, even away from weekends;
  • The takeaway message here is that “users get hooked on mobile data”: once they discover how valuable it can be to them, they use it more and more – provided (and this is the kicker) the user experience is good enough.

Another interesting statistic involved the requests received by Google’s servers for new “map tiles” to provide to Google maps applications. Sumit said that, every weekend, the demand from mobile devices for map tiles reaches the same level as the demand from fixed devices. Again, this is evidence of strong user interest for mobile services.

As regards the types of textual search queries received: Google classifies all incoming search queries, into categories such as sports, entertainment, news, and so on. Sumit showed spider graphs for the breakdown of search queries into categories. The surprising thing is that the spider graph for mobile-originated search enquiries had a very similar general shape to that for search enquiries from fixed devices. In other words, people seem to want to search for the same sorts of things – in the same proportion of times – regardless of whether they are using fixed devices or mobile ones.

It is by monitoring changes in server traffic that Google can determine the impacts of various changes in their applications – and decide where to prioritise their next efforts. For example, when the My Location feature was added to Google’s Mobile Maps application, it had a “stunning impact” on the usage of mobile maps. Apparently (though this could not be known in advance), many users are fascinated to track how their location updates in near real-time on map displays on their mobile devices. And this leads to greater usage of the Google Maps product.

Interspersed among the demos and the statistics, Sumit described elements of Google’s underlying philosophy for success with mobile services:

  • “Ignore the limitations of today”: don’t allow your thinking to be constrained by the shortcomings of present-day devices and networks;
  • “Navigate to where the puck will be”: have the confidence to prepare services that will flourish once the devices and networks improve;
  • “Arm users with the data to make decisions”: instead of limiting what users are allowed to do on their devices, provide them with information about what various applications and services will do, and leave it to the users to decide whether they will install and use individual applications;
  • “Dare to delight” the user, rather than always seeking to ensure order and predictability at all times;
  • “Accept downside”, when experiments occasionally go wrong.

As an example of this last point, there was an amusing moment during one of the (many) demos in the presentation, when two music-playing applications each played music at the same time. Sumit had just finished demoing the remarkable TuneWiki, which allows users to collaborate in supplying, sharing, and correcting lyrics to songs, for a Karaoke-like mobile experience without users having to endure the pain of incorrect lyrics. He next showed an application that searched on YouTube for videos matching a particular piece of music. But TuneWiki continued to play music through the phone speakers whilst the second application was also playing music. Result: audio overlap. Sumit commented that an alternative design philosophy by Google might have ensured that no such audio overlap could occur. But such a constraint would surely impede the wider flow of innovation in mobile applications.

And there was a piece of advice for application developers: “emphasise simplicity”. Sumit demoed the “AroundMe” application by TweakerSoft, as an illustration of how a single simple idea, well executed, can result in large numbers of downloads. (Sumit commented: “this app was written by a single developer … who has probably quintupled his annual income by doing this”.)

Google clearly have a lot going for them. Part of their success is no doubt down to the technical brilliance of their systems. The “emphasise simplicity” message has helped a great deal too. Perhaps their greatest asset is how they have been able to leverage all the statistics their enormous server farms have collected – not just statistics about links between websites, but also statistics about changes in user activity. By watching the world so closely, and by organising and analysing the information they find in it, Google are perhaps in a unique position to identify and improve new mobile services.

Just as Google has benefited from watching the world, the rest of the industry can benefit from watching Google. Happily, there’s already a great deal of information available about how Google operates. Anyone concerned about whether Google might eat their lunch can become considerably wiser from taking the time to read some of the fascinating books that have been written about both the successes and (yes) the failures of this company:

I finished reading the Stross book a couple of weeks ago. I found it an engrossing easy-to-read account of many up-to-date developments at Google. It confirms that Google remains an utterly intriguing company:

  • For example, one thought-provoking discussion was the one near the beginning of the book, about Google, Facebook, and open vs. closed;
  • I also valued the recurring theme of “algorithm-driven search” vs. “human-improved search”.

It was particularly interesting to read what Stross had to say about some of Google’s failures – eg Google Answers and Google Video (and arguably even YouTube), as a balance to its better-known string of successes. It’s a reminder that no company is infallible.

Throughout most of the first ten years of Symbian’s history, commentators kept suggesting that it was only a matter of time before the mightiest software company of that era – Microsoft – would sweep past Symbian in the mobile phone operating system space (and, indeed, would succeed – perhaps at the third attempt – in every area they targeted). Nowadays, commentators often suggest the same thing about Google’s Android solution.

Let’s wait and see. And in any case, I personally prefer to explore the collaboration route rather than the head-on compete route. Just as Microsoft’s services increasingly run well on Symbian phones, Google’s services can likewise flourish there.

Leave a Comment »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Blog at WordPress.com.

%d bloggers like this: