During the last Google Jagger updates (2 and 1 ) some really weird SERP results were spotted.

Jani, a reader contacted me last week about backlink results that were also discussed on the SEM2 list in the last days…

questioning: do non-live links count
meaning: do we get a backlink count from Google even if there's no href in the code?

People currently wonder if web mentions in normal text, especially in high authority sites, might increasingly get measured/counted by search algorithm.

For example, the name of a company or person is mentioned in a whole bunch of places, and that company or person is strongly associated with a particular website or web page. Perhaps that site or page comes up in the rankings on a particular query because of this.

So this addresses the potential for an implied link to a few sites or pages that are likely to be associated with that mention.

Aaron Wall / Seobook.com put's it straight:

I believe they have counted mentions in text files, but why would Google go through all the hoopla of pushing NoFollow if they were going to count unlinked mentions in HTML documents?

I don't think Google would want to infer trust on an unlinked mention when they explicitly created a new link attribute to deny trust.

Andrew Goodman mentioned, that this behavious might be based on the trust – i.e. Trust Rank of the site mentioning.

I could imagine something like that: based on Trust Rank (site authority), subject, user preferences, contextual aspects of the text – YES: google might count it… but I fail to find any decent rule on this… the problem as Jani put it straight:

What's interesting, is that if google goes this way, it needs, or it probably will, examine the page's context to determine the keyword relevancy.

So for example, if you got a site, say 'marketrprof.com'. And people would be writing on forums like "Beware of marketrprof.com", "marketrprof.com are f***ing spammers" and "I absolutely hate marketrprof.com" this could prove fatal for the site.

However, if people would be writing "marketrprof.com is such a good site", "I love marketrprof.com, I use it daily" and so on, that could eventually build up your site for a good ranking.

Text parsing technology is as far as to judge about positive/negative adjectives… I'm sure… so Google could to it this way IMHO.

Observations so far:

A) weird

Well – a weird sample is indeed this page "hobbly.com" … it has backlinks showing backlinks from Apache, KDE …

First I thought hobbly does mirrors for Apache/KDE, which it doesn't seem to…

B) 302 redirect parsing

a lot of links – e.g. from the hotscripts.com directory link out to their resources via 302 redirects… these redirects were used for hijacking some time ago, but generally have the function to tell a browser or search engine a new url to go to after the initial request..

People did this to implement statistic click-thru trackers usually.

These Hotscript redirect links count as backlinks now, as I found in at least one sample that my reader Jani sent me…

Any more observations?

BTW: I agree with Todd about hunting algorithms is painful personally I currently enjoy finding new Google features that will help understand future search better… just like Google Travel or Google Movies