Order In the Plex Now

Archive for January, 2012

Is Too Much Plus a Minus for Google?

Thursday, January 12th, 2012

On Tuesday, Google announced something called Search, plus Your World (SPYW). It marked a startling transformation of the company’s flagship product, Google Search, into an amplifier of social content. Google’s critics—as well as some folks generally well intentioned towards Google—have complained that the social content it amplifies is primarily Google’s own product, Google+.

They have a point. With SPYW, the search experience deeply becomes intertwined with Google’s social networking product. You see it in the search box, where the Google+ identity becomes the way to identify a person whose name is in a query. You see it in the search results, where Google+ content is overwhelmingly displayed compared to other social material from Google’s competitors. You see it in a “People and Pages” list–suggestions for connections on Google+–that appears in the same column as Google’s ads.

In short, they say there’s too much Plus and not enough of Our World, which has oodles of content on other social networks.

Let’s take a step back. Is it a good idea for Google to integrate social information into search? The answer, at least in concept, is yes. Google’s mission is to provide all the world’s information. Social information is a big part of that, and Google has been aware for quite some time that its failure to successfully handle the “people” side of its products has been its greatest failure. Google+ was part of a larger initiative to fill this gap. The other part was to make all of Google’s products more social. Including search.

Google CEO Larry Page prepped us for this recently by saying that Google+ was only the first part of Google’s social ambitions—the next step is to “light up” all of Google. As someone who watched the evolution of Google’s social strategy over a year before the release of Google+, I can affirm that this has always been the plan.

But when it came to search, there was a big question: would lots of social results will actually improve search for Google’s users? Do people want their searches full of information about the people they know? Google is convinced that the answer that, too, is yes. According to search quality guru Amit Singhal, it has carefully measured the responses of its users to its existing social search product (one much more modest that SPYW), and found that people respond favorably to search results tagged with connections to people they know. In other words, they click more often on social links, and are “happy” with the results. (Google knows they are happy because they don’t immediately return to the search box to try the same query again.)

The canonical example is that of Singhal’s dog, Chikoo. Previously, if Singhal had typed in the dog’s name into the search box, he’d get nothing but stuff about the tropical fruit by that name. Now, because he is fond of posting pictures of his canine on Google+, he can get doggie content. And since he shares those pictures with people in his family, when his wife does a search for Chikoo, she sees those images, too, and (says Singhal) is delighted to see the family dog in the results. (When I search for Chikoo, I see nothing about the Singhal family pet. As is appropriate.)

Now think about the circumstances that would lead to such a search in the first place. Singhal is going to Google search to look for information about his own dog, the way you or I would use a search engine to look for information about Jessica Biel or Mitt Romney. On first blush this seems odd, to be sure. But Google is always striving for more information in its indexes. It stretched the boundaries its results with its Universal Search product, which brought in media other than web pages, and got some flak for it. But it was the right thing to do, and Google figured out how to do it well. And now we demand images, video, books and other media in search.

Indeed, Google sees SPYW as a similar advance to Universal Search. With the fervor of a recent convert, it believes that social information is a corpus that must be included in search. One day we may marvel that when we searched we didn’t have access to all our social content. Fair enough.

But to really satisfy the user, you need a critical mass of that information to make the searches truly relevant. While there may well be some pictures of your dog on Google+, the bulk of such photos (for those of us who don’t work at Google) are probably be on iPhoto, on Path, and of course on Facebook. But you won’t find them on SPYW. (In a blog item defending SPYW, Google engineer Matt Cutts notes that you can find results from sites like Quora, Twitter and FriendFeed. He even shows how a Flickr photo can appear. But none are as deeply integrated into the product as Google+. And Cutts does not mention Facebook, by far the world’s richest social corpus, in his post.)

Google defends the predominance of Google+ content in its new product by claiming that competitors won’t share. “We always want to provide the most relevant set of results,” Singhal says. “We’re open to working with others. But that information is not available to us. They won’t even let us crawl it.” (And to its credit, Google offers an opt-out function—a toggle switch that lets people see results without social content. Maybe it should have been an opt-in.)

Now that Google has released the product, Eric Schmidt has publicly called for negotiations with social sites to integrate their content in Google search results. But there are competing stories about how much Google really wants to make those deals. Sources close to late 2009 discussions between Google and Facebook tell me that Google had the opportunity to integrate Facebook information in its search results–on the same terms that such content now appears in Bing. But, those sources say, Google refused, on the grounds that it could not technically provide the privacy protections required. Those privacy protections involved restricting social information only to people who users want to share with—basically what Google has now provided for users of its own service. (Google’s head of communications and public policy Rachel Whetstone responds: ” In 2009, we were negotiating with Facebook over access to its data, as has been reported. To claim that the we couldn’t reach an agreement because Google wanted to make private data publicly available is simply untrue.”)

In addition, it seems that in the run up to the SPYW launch, Google didn’t approach social sites and ask them to deeply integrate their services into the new product. To understand why Google may not have been so aggressive in seeking launch partnership, look back at what happened last summer. Google had made a deal with Twitter to get access to the “firehose” of its tweets in real time. Google had devoted considerable resources and effort to launch a real time search product that depended on Twitter content. But last July, Twitter decided not to renew the contract. That basically killed Real Time Search, and Google had to take the embarrassing step of shutting it down. The minds of Mountain View do not want a repeat of this incident. So Google is reluctant to develop new products that put it at the mercy of other providers.

Understandable? Maybe. But it’s counterproductive.

In any case, competitors say that Google could do more without such partnerships. Alex Macgillivray, Twitter’s general counsel and an important early employee in Google’s legal team, says that there was plenty Google could have done to make sure that relevant Twitter content was treated more fairly compared to Google+ content. “In the Google I knew, you didn’t have to have a deal for your stuff to be considered relevant,” he says.

(I don’t want to let Google’s social competitors off the hook. All too often, they regard the personal information users share with them as their own. Obviously, users should be free to dictate where their information should go—including whether information is available to search engines.)

No matter what the reasons, the unbalanced delivery of social content in SPYW is unsatisfying. It’s like a music service starting up with a license with just one label, or a map service going live with under half the terrain depicted. If I’m searching for social content, I don’t want to have to figure out which company is naughty or nice. I just want to find the pictures shared with me, whether they were posted to Facebook or Google+.

Nonetheless, Google decided it was imperative to get its social content into search now. Larry Page is a notoriously impatient leader. Read Google’s just-released quarterly magazine—it’s all about how Google is built for speed and nimbleness. It also boasts that Google is more nimble by shifting from a consensus culture into a command-and-control structure where someone is empowered to give a fricking order to the troops.

Obviously Google’s CEO seems to have concluded that deeply integrating lots and lots of Google+ into search is not a problem. I think Google wants us to see its social efforts not as a direct competitor to Facebook and others, but a core value of the company. By that thinking, Google+ is Google. And search itself is part of Google+.

That’s a heavy concept, and one that’s potentially groundbreaking. If Google is able to leverage the knowledge it accumulates about you, it can deliver much better search results. (Simple example: Google knows you are a vegan. When you search for a restaurant, you won’t see BBQ results!) Google notes that in order to do this, it has to know who you are, and that’s why the Google+ identity is now integrated in search.

But there is a risk to proceeding on this path. The company has spent its entire corporate life protecting the integrity of its search product. When writing In the Plex, I learned that the secret behind Google’s somewhat bland design was that if Google looked like it was designed by a machine, users would implicitly understand that Google search itself was unpolluted by strong opinions. Google meticulously positioned its flagship product as a neutral judge of what was relevant to the user.

Search, in short, should appear to be like Caesar’s wife, above reproach. When using its algorithmic wizardry to deeply integrate social information into its search experience, it behooves Google to avoid even a whiff of bias. With SPYW, though, the odor is unmistakable. No matter how you cut it, the search engine now increases the value of participating in Google+. It may be Google’s right to do this. But it also may turn off a lot of users. And it also provides ammo for Google’s detractors, including those in Washington.

In fact, some people are saying that Google’s move may trigger an antitrust action, and there’s already talk that the FTC is on the case. But you don’t have to get into legal issues to see why Google’s new product as it appears now takes the company into dangerous territory.

Like I wrote — Google is full of Speed Freaks

Wednesday, January 11th, 2012

I know, I promised a bigger take on Google’s increasingly controversial Search plus Your World. I’m still assimilating it and gathering information, but it’ll come.

While I’m dragging my feet on that, a word about Google’s lavishly produced Think Quarterly magazine. The theme this quarter is speed. Readers of In the Plex know how important this is to Google, especially to Larry Page. I tell a story in the book where Page mentally measures a product’s speed in milliseconds.

One of the people I spoke to about Google and speed was early Googler Urs Hoelzle. In this issue of Think, Hoelzle outlines what he calls “The Google Gospel of Speed.” He outlines the company’s goals thus:

At Google, we don’t plan on stopping until the web is instant, so that when you click on a link the site loads immediately, and when you play a video it starts without delay.

I bet Google won’t stop even then.

In the Plex also talks at length about how Google tries to maintain startup nimbleness while growing and growing. There’s an interesting essay by Kristen Gil, Google’s VP of Business Operations, in Think Quarterly that describes this, with several bullet points that include Google’s recent shift to more of a command structure. (Something I first noted in my dispatch about Google+ and its origins in Wired.)

Gil also mentions Google’s internal management system called OKR’s but to get the full story on how Google has adopted Andy Grove’s system of Objectives and Key Results (brought to Google by John Doerr), you need to go to (you guessed it)…In the Plex.

Just sayin’.

Did Google Just Pop the Filter Bubble?

Tuesday, January 10th, 2012

Today Google launched Search plus Your World, a significant change to the flagship product that just about everybody uses every day. It’s a big deal, and Mike Issac has the deets and some analysis. I’ll post something myself on the broader issues later on.

For now, let me note one feature that may be lost in the larger news. Because some people may not want shared items from people on their social graph to intrude with their searches, Google offers a quick opt-out: a button that removes all social results from a search.

But it does more than that. Choosing that option blocks Google from using the history of your previous searches when it provides results. (Google still uses some personalization in choosing results, namely language and location. Otherwise, says Google’s search quality guru Amit Singhal, the results would probably be unsatisfying if not confusing. If you’re really motivated, though, you can plow through some menus and change those settings, too.)

According to the Google blog:

We’re also introducing a prominent new toggle on the upper right of the results page where you can see what your search results look like without personal content. With a single click, you can see an unpersonalized view of search results..

This seems to address the complaint known as The Filter Bubble, as popularized by Eli Pariser’s cogently-argued book of that name. Pariser contends that when Internet providers personalize their services, people wind up seeing only what those providers think they want to see—stuff in their comfort zone. Pariser engages in world-class hand-wringing at the prospect of people exposed only to things they already agree with or are familiar with. Google personalization is his bête noire.

I reached Pariser this morning. He had not heard the news, but I briefed him and he checked it out on Google’s blog. His first reaction (pending a more thorough dive into the feature) was positive. “It’s definitely a big step in terms of transparency and control. It’s kind of awesome to see them do this,” he told me. There are still issues he’s like to look into — does the opt-out also block information like what browser people use?– but at first blush, he says it seemed to really address the concerns he raised in his book.

In one fell swoop, Google might have popped the Filter Bubble.