Google is using authorship to rid us of the faceless web but our relationships and their social signals can have a much greater impact. Welcome to the trust network.
Tom Anthony wrote a very thought-provoking post over at SEOmoz showing how authorship and social signals could have a profound impact on search. Inbound links have decreased in influence over the past couple of years, in part due to abuses of black hat SEO, and Tom asks one key question in his post: “How much do we trust this link?”
The introduction of social signals and authorship means that we can now see not only where we are being linked from but by whom regardless of where they have posted so that we have a situation where “a person” not just a site is creating that link. Tom states that Google, “knowing this is an authored link, by a human who they have data about” can place far more trust in it. Using a combination of existing PageRank and AuthorRank would give an overall weighting.
This lead me to draw together a number of points I have been considering for a while and ask what if we went beyond just authorship?
Welcome to the trust network
With authorship Google is trying to address the faceless nature of the web but is simply knowing who created something enough? While we can infer that an item (and consequently any links from it) may be of higher quality because someone is willing to put their name to it this might not always be the case.
We, therefore, need a multi-faceted approach incorporating trust, interest and authorship to gain a true picture.
Trust is something I’ve been talking about for a while which is, I believe, one of the biggest reasons that Google launched Google+ with a real names policy.
The introduction of social signals on their own just ends up having the same problem as links with endless accounts “curating” content possibly with an ulterior motive. With the launch of a new social product, that also aimed to be an identity service, Google needed to establish a “trust network” as well as a social network in order to be taken seriously.
Who we are, what we do and how others view and relate to us is key for determining our trustworthiness when both creating or curating content and linking to others.
Current influence systems are badly flawed as they simply measure raw numbers. We need to go deeper and that can only be done by combining more signals: volume, yes, but also trust levels and sentiment analysis. Rather than purely looking at the number of shares, comments or likes on an item we have posted we need to take into account the “response sentiment” to our actual posting as well: is it relevant, is it appropriate or accurate, are we just trolling in order to falsely generate a response?
When looking at the Ripples feature on Google+ I suggested that analysis could be extended to the actual responses so that influence could be determined by sentiment and the analysis of multiple ripples starts to give us a picture of who is sharing our content on a regular basis.
Social sharing is now just as valid as traditional linking – it is, after all, just linking in a different form to a different audience.
One of the big movements in social which hasn’t quite come to fruition is the interest network. The big players are all dabbling on the periphery but so much more can be done to truly utilise our interests to create connections and ad hoc social graphs as opposed to the explicit graphs we create by following “people”.
When examining interest we will be trying to identify the key players and subject matter experts for each area thus implying an inherent sense of trust in these individuals.
I have mentioned in the past that influence should not be based solely on numbers but, for those who publish research papers, articles, even trusted blogs on a given topic, to have influence also determined by the critical response to their works – again sentiment is playing a big part here. An individual post could receive hundreds of shares or comments – good for purely crunching the numbers – but if those responses only serve to criticise and point out how wrong the author is then this is false influence.
Common interest contributes to a persons relevance and gives us a basis for trust. Google is already working on the relevance of our connections, and we can sort those we follow on Google+ by this criteria, although a lot more work is needed and most probably forthcoming.
Relevance is not static, however, as it will depend on our interests, location and even current platform – be it web, desktop application or mobile – so there are numerous factors which determine the relevance of other but this may change over time and based on specific circumstances.
When looking at Facebook’s acquisition of Instagram I suggested how the social giant might use the alternative graph information to facilitate per-platform targeting of advertisments. Similar principals could be employed to assess current relevance based on what we are doing, where we are doing it, the platform we are using and its capabilities: what is our connection speed or do we have a front facing camera for example.
Search Plus Your World was a huge move towards increasing the importance of the social signal: the links are important because they are relevant and our trusted connections say they are. Extending this relevance and trust will provide even greater value when using social results.
Keywords usage is also changing as the social web develops; simply using keywords is no longer enough as we must take the interest graph in to account and, you guessed it, the interest graph is also influenced by our trusted connections and their relevance.
On Google+ we can apply a weighting to each of our Circles so that the content appears in our streams with different priorities, imagine the impact when this starts being applied to the social signals we see elsewhere on the web and in search – the trust network will be well established.
If we do not get value from our search results we can currently use the “ask your friends” option to post a question to our Circles but true relevance would already be placing emphasis on the posts and shared items from those deemed most likely to provide an answer to the particular topic we are searching for. Perhaps we could have a “trust flag” for those we always want to see results from but the majority of the relevance and trust calculation will be automatic.
Bringing it together
Where our trust network (a combination of those we rank in active circles and who we interact with most frequently – an extension of the relevancy algorithm) intersects with our interest graph we have a meaningful indication of influence about potentially any given topic which can be built on the fly. Combined with traditionally trusted sources from outside our networks we have the potential for excellent results.
Images by thorinside and myself