Bing Director Dings Google's Real-time, Universal Search

 
 
By Clint Boulton  |  Posted 2010-03-26 Print this article Print
 
 
 
 
 
 
 

It's easy to be the startup rabbit-punching the giant from below when the giant is so much taller it can't even see you when it looks down.

At least that's the way I see it when Microsoft executives shepherding Bing, which is pretty much a startup, on the rare occasion criticize giant Google in public. If it's possible to be politic in criticism, Bing executives have certainly mastered the art.

Two weeks ago, I spoke to Bing Director Stefan Weitz. Yesterday I caught a keynote speech by Yusuf Mehdi, senior vice president of the online audience group for Bing, who was in New York City for Search Engine Strategies March 25.

Mehdi unveiled several new search features coming from Bing this summer, including a deal to put Foursquare on Bing Maps.

Both Weitz and Mehdi were respectful of Google, offering what seems now like versions of a script with alterations in diction. Mehdi told the attentive SES crowd yesterday:

"It's more of a dialogue with the consumer," Mehdi said. "We are about understanding user intent, and in mapping the intent into tasks and into actions."

Weitz told me two weeks ago:

"We aim to refine intent so that a human can understand what you're asking," Weitz added. "Google organizes the world's information and that's great. We are moving beyond organizing information and what comprises knowledge and that's more than links or multimedia on a Web page."

The rub is this: Google mastered the art of helping users navigate the Web and returning links that bring users more information. But Bing is about helping users get to the information they seek much more efficiently -- in other words, in fewer clicks.

In the past, we would search for concert tickets and look for links to ticket purveyors. Today, we want those links from our search engines, but we also want options to purchase tickets and maybe even book flights, taxis or hotels to accommodate us -- all from that Web page where we found the ticket offers

So Bing is trying to narrow the gap between discovery and e-commerce transaction. That goes for goods and services to advanced research and other areas. That's good work if you can get it, and Bing has been gaining share. Ah, but it's still gunning for Google, who reigns in this search jungle.

Google tried to put more stuff on its search engine results Web pages with Universal Search, but Weitz said it doesn't really work for today's users. He told me:

Universal search is still a multimedia and text-based approach, where [Google] is still trying to cram as much stuff onto the page as possible. The problem we have, that every engine has, is this notion of how do we derive intent or divine the intent of the query. When you have a 2.4-word query on average coming in, divining intent is really hard so universal search was a faceted approach, but it wasn't really faceted. It was, 'we'll put as much as we could into the page in the hope that one of those things would be the thing you're looking for.'"

Our approach is that when you punch in an ambiguous query, in many cases we'll fire off that left-hand rail, which has those most common refinements people are doing post that initial query. Our goal is less about trying to guess your intent on that first query and more about trying to help the user refine the intent where either engine or even a human could answer it intelligently."

And Weitz really doesn't like Google's real-time results, which are directly jammed into the SERPs.

I think that is noisy in many cases. The two things that we think a lot about when it comes to real time is the ability to tap into the zeitgeist of what's popping quickly, so you're actually able to see if there is an earthquake in Chile. I like the ability for Twitter and Gowalla and Foursquare with check-ins and comments capabilities to detect the spikes in popularity without having to rely on search queries and backlinks and anchor text, which are going to lag significantly from real-time info. That has opportunities for us to promote results higher in the SERP or fire new answers.

Two weeks later, Bing pulled the triggered on its Foursquare deal, surfacing check-ins on Bing Maps. But Weitz had more to say on real-time. He added:

Real time is a facet of social and there is a distinction to be made there. We're looking at the social space, where people are turning to when they don't know what words to use in a query. We've done research that talks about how people were able to find an answer faster using their social graph than they could using keyword search because they didn't know what words to use.

Bing Twitter is a great service because it does what I think needs to be done for that real-time stream of data, which is to take out what people are actually talking about with respect to those simple keywords. You could type in SXSW right now and see the most common pages and stories people are pointing to when they reference the SXSW hashtag. That is one area where real-time/slash social Web is interesting because you can find the nuggets in that massive 50 tweet a day stream that using any client that exists today is not possible.

One area where Bing and Google engineers are likely to agree on is the disambiguity challenge. Weitz told me:

For example, take the query "Gossip Girl." Clearly we have the notion that, based on volume, it's pointing to the TV show. But even that itself is a really ambiguous query. Are you looking for the cast? Are you looking for the showtimes. Are you looking for clips? are you looking for sound bites? Whatever it might be. That's the disambiguation challenge. How do we actually move search away from this notion of a query and hope, or a query and click, to more of place where the query "Gossip Girl" has these common intents once these folks actually query for that particular entity and these intents cluster around these domains. It's about allowing people to easily or naturally refine that intent to the point where you and I can understand what they're asking.

The trick, of course, is empowering computer applications such as Bing and Google to understand what humans want, right down to the most fringe-like nuances.

When you consider the hit-and-miss frustration of many complex searches today, the task seems Herculean. There's so much high math. Couldn't we just wire a human brain into the Matrix to make machines think more like humans?

Oh wait, people already tried that. In a trio of science fiction movies.

 
 
 
 
del.icio.us | digg.com
 
 
 
 
 
 

Submit a Comment

Loading Comments...
 
 
Manage your Newsletters: Login   Register My Newsletters























 
 
 
 
 
 
 
 
 
 
 
Rocket Fuel