Affiliate Algorithim and Localrank Implications

Affiliate filtering and site to site linking with localrank (Hilltop)

[zx_viper_xz]

I have been doing my reading on the hilltop algorithim however there are still some bits which I do not understand with respect to affiliate filtering and it's implications.

I understand the classing of affiliate sites in respect to Domain(s), IP classes etc. however the straight forward(ish) reciprocal linking seems to be a point of interest.

Consider this example:

Website 1 on IP 201.****.****.****
Website 2 on IP 202.****.xxx.xxx

Both Website 1 and 2 have link pages.

Website 1's Link page links to the Homepage of Website 2 and Website 2's Link page links to Website 1's Home page.

Since Site 1 does not link (Page to Page) to Site 2 they are not really affiliates. (Please confirm this someone).

They have different domains, are on different IP blocks etc.

Now this is where it gets iffy, from what we've all been taught a pagerank is assessed individually (per page). However with this whole affiliate thing, it appears that if expert site links to pagex.html on a site then that page gets the expert link (Localrank), however if pagex.html then links to pagey.html, Pagey will not get any of pagex's localrank (seeing that pagex is an affiliate of pagey).

So this brings up alot of questions. 1, is Localrank based domain by domain (in which case a link from the homepage from site1 to site2 may be canceled out completly by one small link back to site2 anywhere in that domain).

<snip>

James.

[edited by: Marcia at 1:40 pm (utc) on Mar. 25, 2004]
[edit reason] URLs not necessary. [/edit]

[Marcia]

Welcome to WebmasterWorld, James.

We can go right to the Hilltop paper itself to see what the criteria are for defining a non-affiliated expert page

When Experts Agree: Using Non-Affiliated Experts to Rank Popular Topics [www10.org]

From 1.2(i):

We define an expert page as a page that is about a certain topic and has links to many non-affiliated pages on that topic. Two pages are non-affiliated conceptually if they are authored by authors from non-affiliated organizations.

Then down in 2.1 - Detecting Host Affiliation

It goes on to refine the definition of how the judgment is made, which seems to be what you're asking about.

[Gory]

There is no evidence that hilltop algorithim has been applied to Google. I don't think this relate to Google.

[ILuvSrchEngines]

>There is no evidence that hilltop algorithim has been applied to Google.

Why do you say that? Where is your evidence that there is no evidence?

[allanp73]

I think the current serps really show that hilltop is in effect. Sites apparently filtered out are really just annihilated by hilltop. The problem with Hilltop is that it assumes that the best sites would be authorities, when the usual case is that the authorities link to the best sites. So Google filters out the very sites people want to find. Brillant! ;)

[bird]

>There is no evidence that hilltop algorithim has been applied to Google.

Why do you say that? Where is your evidence that there is no evidence?

That question doesn't make sense.
The absense of positive evidence is negative evidence in itself. Of course, as soon as you supply positive evidence, that absense vanishes. Do you have any data to contribute that would provide positive evidence?

[mfishy]

<<Where is your evidence that there is no evidence? >>

Imagine a lawyer saying this during trial. :)

[jchance]

There is no evidence that hilltop algorithim has been applied to Google

I'm not sure what your looking at, but I see PLENTY of evidence.

First and foremost, directories are doing great. This would make total sense in relation to hilltop because they could both be considered an expert page, and most directories have many directories (experts) pointing back at them too.

Second, for non competitive searches the results look like the pre-florida style results (in my opinion pre-hilltop). Again, this would make sense because hilltop can, by design, only return results for highly competitive general topics.

Thats 2 pieces of evidence right there and if you dig deeper you can find even more.

[allanp73]

So what do we do with this new algo? I noticed that sites that cross link if on separate IPs are not affected by local rank to answer the original question. I can give example, if necessary.

[zx_viper_xz]

So by the look of it they only disreguard affiliated sites and reciprocal linking of non-affiliated (host / domain / ip) does not come into the Hilltop Algroithim does it?

[allanp73]

zx_viper_xz,

I believe that Hilltop is only applied to select keyphrases. This is part of the reason why people felt that the filter was targeting money terms, because hilltop algo removes sites that lack enough local rank.
The local rank part seems to mainly look at where the links come from. Another element that seems to be important is links out to quality themed sites.

For me the algo has been devastating. Out of 100+ sites I work on only a small handful are ranking well. The rest are unfindable.

[zx_viper_xz]

I feel the big reason why some parts are unfindable are the 2 localrank links minimum (I recon google stole this ideology from the pubs) before you even appear on the results. I too have noticed sites drop from 1 to 800 and feel that this 2 link min is the key.

[allanp73]

zx_viper_xz,
"...feel that this 2 link min is the key."
I don't think you are correct my sites generally have 10-20 links in from multiple ip. Though you might be on to something with the minimum link issue. I will look into this more.

[Gory]

>First and foremost, directories are doing great.

Well, from where I stand I don’t see that Google is giving any extra boost to "authority" sites, as a matter of fact I see plenty of non-authority sites that are doing great on highly competitive searches.