At SES San Francisco in the summer of 2010, I was on a panel where we discussed the topic of real-time search, representing the marketing and publishing point of view. My other co-panelists
were all from the top real-time engines: Dylan Casey from Google, Tobias Pegg from One Riot, and Paul Yiu from Bing. They all made a statement that came off as something of a whisper in a world where news travels around the globe in a matter of minutes; but that statement would become the first confirmation of the largely growing influence of social
relevance on traditional Web search. The statement was simple, and was echoed by all engine representatives on the panel: "Yes, we do look at tweets with links differently than other data
[as well as links in other status networks]."
At the time, these three engines were using the Twitter fire hose to amplify their own results with a stream of tweets that was propagated quickly, and turning the stream into a search signal. Perhaps most notably, One Riot was the first engine to use aggregate network effects in the actual ranking of results, a signal that both Bing and Google would later adopt as well. In other words, rather than showing you 100 duplicate retweets of popular content (as Twitter search did at the time), they rolled those retweets into an aggregate signal, and ranked results by Twitter popularity. One might also make the case that Google was doing the same thing with massively duplicated content (think press releases or other syndicated content that is highly replicated), but this was never confirmed as overtly as One Riot's algorithm.
Fast-forward to today, and we now have a sort of space race going on with network information sharing with the likes of Twitter, Google+, LinkedIn top sites, and other networks that use network effects to rank and also push content. Getting back to my opening premise on ranking documents in search, it is this aggregate network sharing effect of links and keywords that is having the most impact on search. Though the story of content and social signals has long been touted by search professionals, never in my experience in this industry have I heard so many direct statements from the engines (in print and in conversation), that content and social are indeed their main areas of focus -- and these are the areas recommended to marketers who wish to take their natural (and even paid) search programs to the next level.
Distinguishing the share graph from the link and the social graph
So while the social graph is having an impact, it is the intersection of socially shared links and keywords, and the network echo effect that cause search engines to begin to look at it the social graph in a different way. In making the distinction between the social graph and the core differences as it relates to search, the network effects appear to be more on a share graph, rather than a social graph, though it is fair to say that the share graph is a subset of the social graph, though still on par with the link graph in terms of potential for future impact on search results. And of course, it is just one of many other signals that the major engines use to rank pages, but it is appearing more likely every day that this will be more than just a signal among many others, elevating it to a true cornerstone of the search results page as we know it.
Here are a few attributes that make specific links and keywords resonate more in a share graph scenario:
- Authority of the network sharer
- Theme of the user, or related concepts of the network sharer
- Depth of the network sharer's network, and the authority of those in the network
- Trustworthiness of the network sharer (in relation to the likelihood that they produce spam or low-quality content
- Velocity of keyword usage across a network, and the spreading of that keyword or phrase over time though shared networks
- Velocity of link being shared, over time, based on time frequency, volume, and level that it cascades though various networks
Again, the distinction here is that search engines view status updates with links differently. While this effect is gaining traction and slowly trickling into the search results, we know it is a fact. Google+ brings this intersection square onto the marketer's doorstep. They see the network effects of sharing across live networks as a parallel to the link graph, and that it complements it in different ways, though it could be argued that the share graph is simply a faster method of compiling real-time links. But one thing is for sure: if marketers value their natural results and want to take their search programs to the next level, then they'd better get active in content production and social propagation, in a big way.