Liz Gannes: What’s the status of social search at Bing?
Stefan Weitz: We’ve been blending social signals for 18 months now, even just to do things like detecting possible spikes when we see lots of tweets coming in on a certain topic. And we have a separate SERP — a separate page — where you can see social results.
The first thing is, we are taking this pretty slow, and there’s a pretty good reason for that. People don’t understand how amazingly complex it is to make sense of any social signal. So we are being very conservative about where we fire social results.
That’s the first thing; the second thing is there’s more than likes and shares. It’s more about augmenting this mechanical product — the algorithmic search engine — with people. So we shipped things like understanding the cities where you live, friends’ opinions on stock quotes — a bunch of things besides just firing off social search.
Do you think it makes sense for search engines to pay to access social data?
I’m not on the business side, but I think for search to work properly, you have to understand that if a missing component has to be included, you have to [make a deal for] it.
Has social search positively impacted the Bing experience? Are there measurable impacts of social users being more satisfied with their results?
For sure — the biggest thing we see is when you look on the search page and see the faces [of your friends], the click-through rate goes up substantially. It goes back to basic neuroscience: We pay attention to people. The core user experience has gotten a ton better, and it’s very early. We’ve taken a while to do this, but it’s complex.
What in particular is complex?
Figuring out what does a “Like” mean, what does a share mean. Originally we were going to fire off “Stefan likes this result” even if there’s a comment. But what if I say in the comment, “This article’s totally wrong.” On one hand I have the “Like,” on the other hand I have the lexical comment. Or I might be retweeting it from someone else, or I might have just thought it was funny. Trying to understand that very atomic action is hard.
Bing’s Stefan Weitz says search is better with faces.
We’ve found it’s important to look at the whole person and understand “Stefan likes to share on computer science, and he has an interest in spatial dynamics.” On Twitter search we will identify experts on a certain topic. That’s something we can do but we don’t do that on any scale yet.
Why aren’t you doing more to capitalize on the goodwill from people who dislike Google’s Search Plus Your World? Shouldn’t you be mounting a “switch to Bing” campaign?
We are doing some ads this week (There was also a Bing-is-great blog post today). They [Google] are doing a nice job on their own of handling this problem.
But they are learning just like we are. They did what we didn’t want to do, which was make the user experience peppered with this stuff, with +1s everywhere, the Google+ content in the top corner. I think [Google] realized we were ahead and they overextended. But I know a ton of guys there and they’re smart and they’re reacting to what has been said.
What would happen if Microsoft had its own significant social network? How would that change your relationship to other social networking sites? Would you be tempted to give preference to your own on-network content?
Well, we do have Windows Live, which has half a billion accounts — though not a lot of social activity because we have linked to 25 or 40 other social network profiles for years.
I remember the discussion a few years ago that, even though we had a very robust social product, there were 60+ social networks across the planet. We thought, it’s naive to assume a single social network will rule them all or to make people come to ours. So we have the guys running around doing partnerships with 60 different networks.
Us partnering is the only way we’re going to make a big difference here. We have to use the whole web to actualize our vision of helping people do stuff, not just find stuff. And everyone wins, which is nice.
Can you explain what you get through these deals? What information is accessible through data feeds that isn’t through regular crawling?
Just from a technical standpoint, crawling is expensive. We could certainly hit a site a thousand times a minute, but it’s not efficient. Feeds just generally are more efficient. And also crawling doesn’t necessarily have a structured data set.
What about getting access to analyze each user’s social graph, something Google has said is very important?
Certainly having a social graph is a good thing for Facebook, which has an amazing amount of data. There’s also people I follow on Twitter, which is a public record. But different friends are valuable for different things — one single network can’t rule them all.
When are you going to press your social advantage in Bing, seeing as you have both Facebook and Twitter deals and Google doesn’t?
You’re going to see the culmination of a lot of our learnings in the not too distant future. All those lessons will be applied into something that I think is pretty interesting. How we think about social is always evolving, and the next turn of the crank is more differentiated than we’ve seen in the past.
