The system of ranking a webpage using PageRank has been around for along time now and is starting to show its age. With spam on the increase due to the use of links as a quality measure, it has become necessary for search engines to look beyond the PageRank algorithm to find a better way of differentiating the good from the bad. Enter TrustRank.
TrustRank works on a simple principle. A trusted (good) page will rarely point to an untrusted (bad) page. So if we can find the most trusted websites, we can create a map of the web from them and propagate trust. By most trusted we mean sites that are authoritative, do not currently employ any spam techniques and are unlike to ever succumb to them. We gather together a set of these most trusted sites via human inspection and this is the starting point of our map of the web known as our seed set.
We know that a site in the seed set is trusted so we give it a maximum TrustRank value (mT). We also know that a site in the seed set is unlikely to link to a bad site. As we can not be absolutely certain, we use a decaying factor (D) when
passing TrustRank. Also we split the TrustRank across the outgoing links (3 outgoing means each is given 1/3). Therefore a link from a site in the seed set has a value of ;
(mT –D) / number of outgoing links.
This then fans out across the web and produces a weighting for all connected sites that decays the further we travel from the seed set.
There are of course flaws with this system. The initial seed set has to be small enough to be manageable (humans have to do the research on each site), but large enough to propagate through the complete web. To overcome this we include trusted hub sites into the seed set such as DMOZ.
So, where does this all fit in to SEO. Well since the Big Daddy update we have seen changes in linking. Authoritative links (.gov, .edu etc.) have a much greater effect on a site ranking than pre Big Daddy. This is due to them having a high TrustRank. This means the shift has gone from the more links the better to the higher the quality the better. We are seeing a single link having more benefit than 30 others. We are not saying you should stop relevant linking (reciprocal, one-way etc.) but if you can acquire links from local government or authorities, they will have a greater benefit.
For a more detailed look at TrustRank have a look at Zoltan Gyongyi, Hector Garcia-Molina and Jan Pedersen’s 2004 paper entitled “Combating Web Spam with TrustRankâ€.