Iterations: Graph Search, In Theory And In Practice

Editor’s Note: Semil Shah is a contributor to TechCrunch. You can follow him on Twitter at @semil.

The biggest news in consumer technology this week was created by Facebook. The social network’s new product – “Graph Search” – was the subject of an international press event where analysts initially speculated on all new things, from a Facebook partnership on the scale of Spotify or even their own phone. There were a lot of posts about Graph Search. Facebook is an important, powerful company, and that level of coverage is warranted. I’ve finally read through most of the analysis that made its way through my Twitter feed, and found myself surprised by the range of opinions shared on the matter.

Graph Search is probably not a “Google Killer,” as it indexes, restructures, and surfaces data within its own walled gardens (which includes data pumped into Open Graph). Facebook’s search product opens up a new window to search what many refer to as the “dark web,” a web that is not accessible to Google’s crawlers and algorithms.

Smart writers weighed in with great insights. John Battelle, who has covered Google and search for years, argues Graph Search gives Facebook a new layer of interaction to increase engagement; Steve Cheney artfully argued Facebook’s mining of “Likes” created distorted signals to begin with and puts Graph Search at risk of irrelevance; Pando’s Hamish McKenzie cast Facebook as a network of connections over a network of people, which presumes Facebook must continue to acquire user data, either directly through native activity on the site or indirectly through Open Graph permissions or acquisitions like Instagram; and Xconomy’s Wade Roush views Graph Search as Step One of Many in Facebook’s attempt to produce more relevant results for everything, based on the belief social filters and recommendations are going to best what algorithms can deliver.

I haven’t paid as close attention to Facebook as others, but my initial reaction to the unveiling of Graph Search was that it was less about what individual consumers could do (though that’s cool), and more about how companies, brands, and other institutions could further segment the Facebook audience in order to hyper-target their messages, advertisements, and attention. This is where Open Graph, in theory, could continue to funnel data into Facebook’s data centers and, over time, build all sorts of audiences with a few keyword search terms.

Where this logic breaks down for me, however, is that I rarely send data to Facebook. I use the service daily, and I like it a lot. But, I don’t connect many other services to it that send data back to Facebook. I also don’t go to Facebook to search, so it’s not an underlying behavior for me on their properties. This creates two issues: (1) I’m not trained to search on Facebook; and (2) I’m not giving Facebook more details about me or my graph in order to produce more relevant results against a search. I know it’s dangerous to extrapolate from just my own behavior, but I also sense people are growing more and more reluctant to sign into new services with their Facebook permissions, or to embed their Facebook account into Apple’s iOS operating system.

All this said, I wouldn’t bet against Facebook. I don’t know exactly how, but I feel they’ll figure it out and continue to extract key data, whether directly or through other means, perhaps even acquisitions as its revenue streams mature.

Yet, reading all these great analyses and reflecting on this week’s news, I find myself thinking about other great sources for information where the underlying behaviors for search are already being served in more natural ways. On Pinterest, users discover new things through streams of images, but users also can search directly for things they’re interested in, bypassing Google entirely. On Quora, most visitors to the site lurk (e.g. “browse” and “search”) and do not actively participate, many of them ironically ending up on the site after entering a Google search.

It remains to be seen whether Facebook is able to reinvent search altogether, or whether Facebook successfully opens an entirely new search channel which could monetize as well as Google’s, or whether other sites like Pinterest and Quora, for instance, already have a head start on capturing the underlying behaviors embedded into their properties and flipping them into a search business model.

What I’ve learned from this week, then, is that Facebook and newer companies have a great opportunity to allow their users to search within their walled-gardens and Wall Street, already primed on Google’s model, loves how online search monetizes. There will likely be many different types of search, especially as the “dark web” continues to grow outside the sight of Google. Graph Search is an important step in this direction, but given the competition and the pace of growth of particular startups in the market right now, it remains to be seen if Facebook will actually create the next killer search product. In theory, Graph Search makes a ton of sense; in practice, however, Graph Search most likely has a very, very, very long way to go.