Negative SEO blog : SEOassassin

TrustRank has probably caused more heartache to more webmasters than any other single ranking parameter. It is the primary cause of the “sandbox” effect, or rather it’s lack is. It’s the reason you see old domains ranking for terms mentioned in passing, and why sub-domains can have such power.

For those who like incredibly complicated and abtruse mathematics, please see the TrustRank white paper. Even if you can’t handle the maths, read it anyway - it takes a little bit of sifting, but there’s good information in there.

TrustRank (TR) is an extension of the older PageRank concept. Where PageRank was just a popularity contest, TR tries to make judgements. A site that is trusted by other sites, and trusts other sites, should be trusted. Confused yet? I’ll attempt to explain…

TR is similar to PageRank, but takes chronology into account. A link you created yesterday has accumulated little trust of its own, because you might choose to change it quickly (if you were selling the link as adspace, for instance), so has only a little value, but a link you’ve had since ‘97 (and not changed) is at the very least not annoying your users, or you would have had requests to change it. Similarly, a link that has pointed to one of your pages for less than a week, say, need not say much about your sites quality, whereas one that has been stable for 3 years probably does.

Any individual link does not exist in a vacuum though - it’s part of the flow of the WWW. If a site has a history of creating stable links to “quality” sites, then its links are valuable. If not, then… not. There is the qualifier for “quality” in there - which raises another question : what is quality?

Links that remain stable are quality. Constantly rotating links are bad. Links from hub/authority sites are good. Links from my-spam-directory.com are bad. Links TO hub/authority sites are good. Links TO my-spam-directory.com are bad.

That’s not a perfect list, but it’s not a bad start, I think. So, new sites that only have incoming links from a known link network (say, an SEO firm’s standard “new site treatment” network, or known link pimps), or to / from other new sites CANNOT rank well against established sites for competitive terms in a short timeframe. If their existing linking pattern holds stable, and new, quality inbound links are added, they can quickly acquire some trust and start ranking.

I’ve seen a domain go from nowhere to page 1 in Google for a competitive (not ultra-competitive, but hard enough) travel phrase in a week. That was based on a sitewide link from the parent site, which ranks highly for several related terms. Is that a good thing though?

In the above example, should the new site have been able to displace older, perhaps more deserving sites simply because it had a monster site looking out for it? What about the current rash of old domains like Amazon and eBay taking over SERPs where they aren’t welcome? What about WikiPedia getting in as an “authority” for just about every topic under the sun, when there is precious little control or accountability on it to be even marginally correct?

These sites are ranking for terms they may only mention in passing, or in severe cases, don’t mention at all! To give an example, check this Google search for single white wardrobe. Only a single result of the top ten actually refers to the searched phrase directly, and look what else is intruding. No less than 3 versions of Amazon (.com, .co.uk and .ca), the IMDB and MSN Movies, all returning results for the film Single White Female. WikiPedia chips in a bit later with a result for “The Lion, the Witch and the Wardrobe”. Close, but no cigar.

How come these sites can beat seemingly almost every site on Earth that’s actually about single white wardrobes? Trust. Those sites have bucketloads of trust, due to their age, or due to the large numbers of quality links they have gathered. That makes them a first choice when SE’s are trying return results for a query, regardless of how good a fit their content is as an answer to that query.

This is why, in the title of this post I characterised trust as a reservoir : it’s huge pool of generalised “relevance”, which is considered to apply to EVERYTHING in the site, at all times, and once you’ve got it, you’ll always have it. It applies to content created in 1996, it applies to subdomains of the root domain (a potential spam problem for SEs, especially where a trusted root domain allows users to create and use subdomains), and it applies to phrases you didn’t intend to, and occasionally don’t WANT to rank for.

Trust flows into a site and accretes over time, but it doesn’t *really* flow - once a site has acquired it, that’s pretty much it. They can start selling off text links, and slapping up auto-gen spam, pretty much whatever you like. It might eventually start to suffer (although that’s debatable - I can’t recall an example right now), but with all the extra money washing around, it’s likely that a solution could be found. Given the generally self reinforcing nature of the WWW, I think it’s unlikely that anything but the most desperate abuse over a fairly long period could really drain the reservoir.

I think the SEs need to address this, as it’s hurting their ability to provide relevant results. Trust should be important but it should be harder to earn, and it should be much harder to keep. A site that acquires a load of links over several months, and then stops or drops to a trickle of low quality links, should probably see an erosion of it’s trust, as the “river” flows and takes some of the trust away. If you don’t move forward, you move back.

The WWW is always in motion, flowing, so why do the SEs rely on a metric that can only increase? It’s not benefiting anyone, and it’s not clear what they are going to do about it, or even that they can. Search needs to come down off it’s high horse, and go with the flow.