Google is the best search engine in the world. Still. In the Googleplex in California, there are three issues that will determine the future of search and will decide whether Google will still be the market leader at the end of the decade: First, the fight against spam in web search. Second, the improvement of local web search, especially on mobile devices. Third, the so-called searching without searching (SWS), in which users receive suggestions without performing an actual search.
In all three points, the Author Rank plays a crucial role. In all three points, Google lacks the necessary data. Facebook has this data, works closely with Microsoft’s Bing search engine and works in parallel on its own search engine. With Facebook Home, Mark Zuckerberg also wants to completely monopolize the mobile devices of the users, at the expense of Google’s integration into these devices.
All algorithms currently used by Google are also used in a similar form by other search engines. While Google still delivers the best search results, however, the lead is achieved only through good practice, not through genuine innovation.
The Author Rank is the only truly new approach to ranking search results according to relevance since the invention of PageRank. So far, the Author Rank has only been used to sort posts within social networks. A variant of the Author Rank is Facebook’s EdgeRank, that ranks the posts in Facebook’s news feed.
If it would be possible to transfer the Author Rank over to web search, SWS and local search, the result will have a similiar scope as Google’s market entry and its PageRank. This is Larry Page’s biggest nightmare, because only Facebook has the treasury of data required.
Brief explanation of terms: I use Author Rank as a general term for any algorithm that evaluates the relevance of content and documents on the basis of an author’s popularity. Agent Rank and EdgeRank are Author Rank algorithms of Google and Facebook.
Back in 2005, Google patented an algorithm named Agent Rank that determines a document’s relevance based on the reputation of its author. The reputation of an author is determined by the average feedback his publications produce. Documents that were written by popular authors, therefore, are more relevant than those of unknown or unpopular authors.
The problem at the time: There was no way to reliably identify the author of a document. How should Google verify whether a text on the web actually comes from the author stated therein, or if the authorship is only claimed to achieve a higher ranking?
Identity on the web
Today, identification is possible: More than a third of internet users worldwide are signed up with Facebook. When I publish a post on Facebook, Facebook can reliably identify me as its author.
Whether my Facebook friends actually get to see my post is determined by Facebook’s EdgeRank algorithm. For friends who commented/liked/clicked on my last post, my recent post is more relevant than for those who have ignored all my previous posts. For these friends, a post that I have written myself is more relevant than a post that I have merely shared or commented on.
But Facebook does not only know with which friends my posts are particularly popular. Facebook also has information about what topics I write about. And, of course, Facebook also creates long-term statistics on how popular my posts are on average.
Google has not forgotten about its Agent Rank algorithm: In 2011, Eric Schmidt talked about spammers on the web:
it would be useful if we had strong identity so we could weed them out. I’m not suggesting eliminating them, what I’m suggesting is if we knew their identity was accurate, we could rank them. Think of them like an identity rank. (src)
For this reason, Google has launched Google+. Website operators are supposed to sign up with Google+ and verify their authorship by inserting a code in all documents published online. Thereby their identity can be unambiguously determined by Google and Google can use its Agent Rank to rank the content.
In contrast to Facebook’s EdgeRank, Google’s Agent Rank is not primarily targeted ranking posts within social networks. Its purpose is to improve web search results by using the popularity of the respective authors for the relevance ranking of a web page.
How important the Agent Rank is, is shown by Google’s strategic realignment by Larry Page with a focus on Google+:
This is the path we’re headed down – a single unified, ‘beautiful’ product across everything. (src)
Author Rank in web search
In addition to verified authorship, yet another ingredient is necessary so that the Agent Rank/Author Rank can be applied to web search: A huge database of indications to determine the popularity of authors.
While Google is working hard on increasing the number of true Google+ users, Facebook already has the necessary data. But not enough: With its EdgeRank, Facebook also has an appropriate algorithm to use this data. The step to transfer the EdgeRank to web search is smaller than one would initially assume.
Facebook’s EdgeRank as Author Rank
Facebook’s EdgeRank determines for a user, which contributions of other authors are interesting for him. EdgeRank can be applied to web search if you turn it around: For which user groups and topics are an author’s contributions relevant? From reader-related EdgeRank to author-related Author Rank.
This makes it possible for any author to calculate a relevance value that indicates how relevant his contributions are on specific topics or for specific user groups. This general Author Rank can be used analogous to Google’s PageRank for a relevance evaluation: The higher the author’s Author Rank, the higher his contributions are ranked. Unlike the PageRank, the Author Rank from the beginning will be very hard to manipulate.
Why Author Rank is vital for Google
The Author Rank represents the social graph, i.e. the relationships of people. The social graph is therefore much more difficult to manipulate than the link graph, i.e. the links between web pages. However, Google’s PageRank and other important algorithms used by Google are based primarily on these links between web pages. The result is that it is often not the most relevant sites that are at the top of the search result lists, but low quality pages that are well linked. Dealing with these spam pages is the biggest remaining challenge in the development and improvement of web search engines. Almost all other major algorithms are now also used by Google’s competitors and are often even available as open source.
For this reason, the development of an Author Rank algorithm is one of the most important projects in Google’s development department. This includes the creation of Google+ to get the necessary data.
Facebook, Bing and Google
A scenario that is expected to bring cold sweat to Google’s CEO Larry Page’s forehead, is a use of Facebook’s EdgeRank data as an Author Rank algorithm in Microsoft’s Bing search engine. Facebook and Bing presented a first version of a joint search engine in January and announced the further deepening of cooperation:
“Almost two-and-a-half years ago, Bing and Facebook started down a path to make search more social. (…) Our two teams will continue to experiment and innovate towards our shared vision of giving people access to the wisdom of their friends combined with the information available on the web. ” (src)
Through an Author Rank algorithm jointly developed with Facebook, Bing could gain the lead.In the past, Bing was perceived primarily as a Google copycat, which almost, but not quite, comes close to the original. How rapidly market share in the search industry can change, Google itself has shown. Whether there will be closer cooperation between Facebook and Microsoft, however, is not certain. Mark Zuckerberg keeps all options open in this regard.
But also an independent further development of Facebook’s internal Graph Search is threatening to Google.
Facebook as a search engine?
Even if Facebook would turn its homepage into a large search field and rename itself Facesearch, it would probably bring very few Google users to change their standard web search engine from Google to Facebook. Such a Facebook search engine would probably suffer the fate of all other previous “Google killers”.
So why all the fuss? The assumption that a possible Facebook search engine (similar to Microsoft’s Bing) will be a Google copy is too short-sighted. The mobile use of the web is more than just a trend and Facebook is the most used mobile app . At the same time, Google is convinced that the search engine of the future is more of a suggestion engine. This provides users with search results that match their current context, without them actively having to carry out a search. Amit Singhal, Google’s senior vice president, explains:
“I call it searching without searching [or SWS] (…) That’s going to take an awful lot of innovation.” (src)
While normal web searches will still rather take place on the desktop in the future, searching without searching is predestined for mobile use. Even without having to deal with technical details, it is obvious that relations between authors and readers for such a suggestion engine are better than links between websites. Facebook has everything that the search engine of the future needs.
SWS is one of the two ways to generate significant growth in the search business in the future. Facebook clearly has a real advantage here.
The second opportunity for growth is local search. More and more users are searching using their smartphones. Usually they are looking for a facility that is located in the vicinity of their current location. Although many of these facilities do have a website, Google’s link-based search engine reaches its limits here. Better suited for the evaluation of relevance in this context are reviews of users. The problem: As long as the reputation of the authors of these reviews is not traceable, the reviews can be manipulated almost at will. Similarly to Google’s PageRank, which doesn’t simply count incoming links, but rates their quality, an Author Rank is needed here, which not only measures the number of reviews, but also evaluates the influence of the respective authors.
Currently, Google Maps is still the most used mobile app for local search. However, Facebook is not far behind:
Imagine what could happen if Facebook really concentrated on making its local search experience competitive and invested heavily in promoting it. (src)
That’s exactly what Facebook Home is all about:
With the announcement of Facebook Home, Facebook has pushed the accelerator on its local mobile strategy
The next version is likely to also contain Facebook’s (local) Graph Search or even an evolution of it. Facebook Home will also display content on the display in stand-by mode, an ideal place for SWS proposals.
On newer Android devices, in the delivery state, the Google search box is prominently placed at the top of the home screen. As Danny Sullivan notes, the Google search box disappears when you install Facebook Home:
I’ve had two people from Facebook confirm that if you add Facebook Home (…), the search bar goes away. (src)
That’s the stuff Larry Page’s nightmares are about.