Google Contextual Search and Microsoft Paperclip

Marissa Mayer, former head of the Google search team and now head of the Google’s mobile/social ambitions, has made no secret of her goals. Lately she’s been hyping “contextual discovery” … or search without search.

The idea is simple and compelling: you happen to be at the office, you often go out for a coffee at 10AM, and your phone mentions a new coffee shop around the corner. Or you’re in a new city halfway across the world, and your phone finds a coffee shop to wake up in … without being asked.

The challenge lies in execution: how to be helpful without being annoying. Basically, how to avoid being Cliff Clavin in a pocket.

Microsoft faced a similar challenge a decade ago. After all, 90% of the features in Word and Excel are used extremely occasionally – 10% of the features meet 90% of our needs. So there’s a discovery problem … if you’re Microsoft and want to justify new versions with increasingly more options and features.

Microsoft’s answer was Office Assistant, building on the foundations of Microsoft Bob. The visual representation of Office Assistant was a paperclip … hence Clippy … hence Microsoft Paperclip. Clippy would “notice” that you were doing something (like writing a letter) and offer advice or assistance. Unfortunately, Clippy was intrusive and annoying, and its assistance was wrong, stupid, or just plain obvious – a user experience disaster.

Here’s the question: how will Google avoid the Clippy fate?

Getting the right information at the right time is wonderful, excellent, and good. But there’s more opportunity to fail than to succeed:

Right fact Wrong fact
Right time Party time! Woohoo! Annoyance & anger
Wrong time Annoyance & anger Annoyance & anger

At first glance, it looks bad, but not too bad. After all, it’s a 25% change of hitting pay dirt right? But actually, it’s much worse than that.

How does Google know what is the right fact … one you’ll be interested in? And not just interested in generally, but interested in now? Can a smart contextual search app trace the route you’ve been following and the speed you’ve been maintaining and make some judgements about whether you’re being hurried and purposeful (and therefore not too interested in random desiderata this morning, but maybe very interested in traffic data) … or leisurely and rambling (and therefore maybe touring, and perhaps interested in historical facts that add color to your experience).

These are not trivial problems. And they’re just the beginning.

How, for example, will Google send the notifications? Will they use the same channel/interface as text messages? Will the user need to set a preference that “now I’m interested in various facts?”

I don’t have the answers … but there are a lot of questions to answer before a truly useful and non-annoying tool can be successfully launched.

 


Want weekly updates? Of course you do …



 

                     
%d bloggers like this: