Why is the new google patent still not working

[baron13]

In the new google patent google told us that within the localrank only links from different C classes are counted. The rest of the links, for example: links from the same domain or from the same C class will be filtered out for the localrank. Additional it is more important to get links from websites which are within the same SERPS for the keyword I will get listed high.

Currently websites which have bought lots of high pagerank links are still in the top places....! Why does google not filter them out? Usually 500 links from the same high pagerank domain have to be counted as ONE simple link....but google doesn't!

[Gorilla]

Google is under no obligation to implement something, even if they take out a patent for it.

In fact, it is common practice among large companies to take out patents on dead end technology that they never intend to use in order to confuse competitors and send them down dead alleys.

You have been warned! ;)

[ciml]

A business on the scale of Google will normally have a number of patents. They might give insight into some of the research within the company and perhaps some peoples' ideas within Google for future development, but we shouldn't expect any aspect of a particular patent to be applied.

Google have made some interventions to prevent PageRank flow from certain sites, but it would be very surprising if PR would only take into account on-theme links, or links from outwith the domain and IP. I'd go as far as to say that PageRank could not work under these circumstances.

We might look for more subtle tweaking, or a theming/localrank weight applied to results separately from PageRank. That's only if Google choose to apply these things at all.

[valeyard]

As I understand it (which might be "badly") localrank is about IP blocks, not domains. You do not need to implement localrank to filter multiple links from the same domain.

Personally I think localrank is a ridiculous idea. So many people nowadays use the same hosting services that it would randomly penalise certain sites that just happen to use the same IP block as other sites in their field.

Spammers on the other hand would simply take out accounts on multiple providers.

So if localrank were to be implemented it would hit random innocent sites and not spammers. This would mean a reduction in the value of SERPS to users.

edit: corrected brain failure. localrank only has an effect when there are multiple links from the same class C block

[ciml]

> Spammers on the other hand would simply take out accounts on multiple providers.

Exactly!

University departments, newspaper sections and webmasters of sites in a geographical area using the same local hosting company don't generally ask for domains to be put on different IP ranges, and don't generally spend time locating various geographically disparate hosting providers.

A server IP says much less about a Web site than a postal code says about the reading habbits of a house-dweller...