Search is so crucial to the web that one without the other becomes largely meaningless. In the very early days of the web the only way of having your website found was to print out leaflets with its web address on it and go around the High Street irrationally handing them out to all and sundry you met there.
By the same token search, when executed incorrectly, does nothing more than take the High Street leaflet handout experience and translate it into a digital format. How? Well, search technology is incredibly complex. In order to work correctly it not only has to have a clear way of understanding the mass of data it has indexed and tell what it is, but it also needs to be able to understand what you have typed in the search query box and give you the best possible answer.
In practice the first of these two requirements is supported by the way a search engine creates, runs and maintains its proprietary databases. The second is answered by the way the search algorithm understands not just what a user has typed in the search query box but also the intention behind the search query. This is, actually, much more important. As end users, few of us really know how to search for what we need. Our inability to correctly phrase our search queries is compounded by the fact that much data is imprecisely documented and that makes search tricky.
Google works around all this by not just looking at the search query we have typed but also looking at search query histories to see what online behavior was exhibited by tens of thousands of other users who typed a similar search query. It also looks at the end user’s personal search history looking for hints of search outcomes based on similar searches in the past. It then looks at its Index and delivers not just what it thinks precisely matches the search query but also at what might be suitable based on its interpretation of the end user’s intent.
This has led to what is known as the serendipity in search algorithm, a function of search that is reinforced even more as Google transitions from Boolean search to semantic search. The discovery factor in search is crucial not just because it leads us to more interesting links that may increase our knowledge, enrich our search experience and broaden our perspective but also because that is exactly how search helps us discover the web in ways that make the online world feel like a wondrous place.
Facebook Search Sucks!
Facebook search has been problematic from the word go. For a start Facebook never envisioned that search would be the way its accumulation of data, gathered through its social graph, would ever need to navigated. That’s because it sees the end user who logs onto its platform connecting only with a small, relatively closed, circle of friends, many of whom he already knows well. As a result although we are in the digital domain we don’t interact beyond the environment of a virtual set of Friends and, when it comes to looking for anything, we are back to my High Street example of yore.
Facebook’s social graph never, for instance, supported disambiguation, a basic requirement of search that allows for the distinction between words like say, ‘Mark’ as a name, ‘mark’ as a symbol on the wall, ‘mark’ as a score out of ten and mark as a slang term used to denote someone about to get mugged. As a result Facebook search sucked!
Traditionally it did little more than allowed you to look for your friends or some pages or some possible friends with similar names and anything beyond that was supplied by Bing. Facebook’s reprogramming of its user interface with Timeline was designed to address some of these shortcomings and its launch of its Open Graph was an additional layer of data that had the potential to become much more navigable.
Facebook Launches Graph Search
Fast forward to now and we have the announcement of Facebook Graph Search. The way this currently works is that it mines information input by individuals and interlinks their relationships to display content based on who you know and who you interact with. On the face of it this sounds OK, after all I have said before that much of what we do online now is an attempt to turn back the clock 200 years and undo the facelessness brought about by the Industrial Revolution. We now want to know who gives us advice and who we can turn to for help. We tend to get our information from trusted sources and we tend to develop our own sources from the blogs, news sites and people we come across on the web and in the various social networks we inhabit.
So, really while we are trying to revert to digital equivalent of the community lifestyle our ancestors must have experienced we do so on a much greater scale and with much greater fluidity and control than has ever been possible. Our personal ‘communities’ these days, our digital tribes, are made up of individuals that inhabit different countries, speak different languages and come from different cultures and are in different time zones to our own. What we are trying to get back to, really, is not the controlling and tightly controlled pressure-cooker environment of the rustic community of yesteryear but rather its personalization where trust came from knowing who you were dealing with and were aware of their strengths, foibles and dislikes.
Semantic search of the kind used in Google and Bing takes into account personal connections based upon interactions and engagement, cross-references them with personal history and search query related searches (historically) and comes up with results that answer the search query intent as opposed to the search query itself. The personalized suggestions are also intended to reflect that.
Now, Facebook search does none of that. What it does do is it looks at friends and friends of friends and serves up suggestions based upon what they have posted or what they have chosen to make public or (even worse) what they have posted that now is public but which they think it is not. Because it does mine the social graph, Graph Search can turn up information and pictures from the past that perhaps were left better gathering digital dust.
Graph Search’s Blind Spots
This approach has a vulnerable spot or two. In order to work it requires that everyone on Facebook needs to openly share all their data. Indeed, as one post on Techcrunch noted “Facebook Graph Search Makes Privacy Seem Selfish” the other blindspot is that this approach completely by-passes the serendipity factor of discovery that makes search so powerful.
Search, basically, is the way we navigate the web. It is how we impose order upon the chaos of data. It is how we discover knowledge we did not even know existed and which we are grateful to have come across. This is necessary if we are to avoid the search bubble that Elis Pariser warned about in his book where, incidentally, he also mentions Facebook and the “closing off of results” due to data drawn from intensely personalized connections. From an ideological point of view alone this is hugely important. The web is the great equalizer, it is the largest, single repository of human knowledge outside our own brains. Locking it into walled gardens takes us back to the pre-social media days when information was safely hidden behind data silos.
But that is not all. From an entirely personal point of view search bubbles lock us up into ever tightening spirals where we see nothing but our own bias reflected back at us through our personal connections.
So, is Facebook’s Graph Search Useless?
Well, no, not quite. As Zuckerberg pointed out on his presentation video it is great for doing searches such as “friends of friends who are single in London” and getting back instant results. Every marketer knows that stalking tools have great marketing impact. The ability to drill down into personal connections and hyper-localize some results allows for a truly granular approach to what was, before, a fairly faceless crowd on Facebook.
The hyper-localized context of Facebook’s Graph Search is a direct competitor to Yelp and LinkedIn, two verticals that depend on personal information and the degree of connection to deliver results.
This should also indicate how you can get the most out of this Facebook feature. If you have a presence there you need to make sure that you have completely filled out all the fields requiring information and you should make sure your connections are still active so that you can come up in relative search results. This also means that now your sharing has to be more open and a lot less private than before if you are to get anywhere.
Bottom line, Facebook should not be off your marketing radar. While the changes it made to its Edgerank adversely affected the visibility of profiles and pages the introduction of its Graph Search should address this.
Will Google Search be Affected?
One big question everyone has of course is whether the introduction of Facebook’s Graph Search will affect Google. It’s worth noting here that results to search queries that point to the web are supplied by Bing. So if you were to type “what’s the best flat screen TV to buy?” in Facebook’s search and none of your personal connections could come up with an answer, Bing would step in and point you to the web, where good ol’ SEO (or rather the new SEO practices we have been looking at for the last 18 months) still work. So, no, it should not affect your focus on optimizing your web presence and Google is far from being affected at present.
It takes a massive psychological transformation to go from using Google to using Facebook to find anything beyond your Facebook friends and that has not happened yet.
What You Missed
Could Facebook Buy BING?
The Changing Landscape of Search Engine Marketing
Could Facebook Search Really Change the Balance of Power on the Web?
Could Facebook Finally Succeed in Search
Facebook vs Google+ The Fight for the Web Has Begun
EdgeRank Changes Hit Low-Budget Marketers
External Posts
Eli Pariser on Search Bubbles
Bing, search and serendipity
Bing, net neutrality and search