So TrustRank really does exist

[wige]

As I am sure many of you remember, a few years ago Google trademarked the term "TrustRank". There was a lot of discussion at the time regarding how TrustRank would be used, if and when it might be implemented, and what effect TrustRank might have on rankings. These discussions seem to have quickly died down, however, as very little information has come out regarding the use of anything "TrustRank" related from Google.

Anyway, after the flurry of posts regarding the recent update of toolbar pagerank data over the past week, I decided to take a closer look at the data that the toolbar actually sends to Google, and what information Google is actually collecting. So I set about collecting data, and as I was reading through the logs, I saw a toolbar request that follows the pagerank request from the toolbar whenever a toolbar user visits a new domain:
http://sb.google.com/safebrowsing/update?client=navclient-auto-ffox&appver=2.0.0.8&version=goog-white-domain:1:23,goog-white-url:1:371,goog-black-url:1:15788,goog-black-enchash:1:37185

More interesting, I think, was the response:
Server: TrustRank Frontend

I have not had much time to investigate this yet, and I haven't really picked the data that is sent and returned apart, but I think it is interesting that TrustRank is actually being used. I haven't seen any feedback from this tool - I don't see any indicator in the toolbar, for example, telling me what the TrustRank of the page is, or if there is a problem. But then again, I might not be visiting sites that are bad enough. It is possible that this function has not been fully implemented yet.

Do any of you have further information about the use or implementation of TrustRank? Theories and hearsay will be accepted, provided they are marked as such, and not too paranoid...

[Ace]

Yes you are right now pages with 0 Page Ranks are ranking at top position from last 4 - 5 months and this is due to this trust rank.

[wige]

Interesting... further investigation shows two things:
1) The outgoing connection is triggered on an interval and not related to the page you are viewing. After a set interval, a new list of known-bad domains is downloaded. User information is not sent to Google regarding the page you are visiting.
2) This is not part of the toolbar. The download of TrustRank data is a feature built into the Firefox web browser as a phishing filter.

Still, interesting.

[erikko]

may i ask how can i attain Trustrank or how can i see it?

[wige]

I found this using the Firefox web browser and a header analysis tool. If the site is considered trusted, no data is returned (on the live version). The default method simply stores a list of known bad domains on your computer. When you visit such a site, FF will display a fairly obvious message.

[ctabuk]

Wige - I thought Yahoo registered Trustrank and Google asked if they could use the system?

It appears I was wrong
Google Trademarks: TrustRank & The Neighborhood Wide Web [SearchEngineWatch] Oh humm

[crankydave]

Not exactly David.

Google has an intent to use patent on the "word(s)" Trustrank. Yahoo filed a patent based on the whitepaper. Two totally different things. Some of the particular links are on my blog...

CrankyDave » Blog Archive » TrustRank and Yahoo

It's certainly no secret that Google intends to use something they will call "TrustRank". Exactly how it is implemented, calculated, and what it may or may not do has yet to be seen as indicated by the intent to use patent.

Google has filed and been granted another extension for their intent to use patent, an indication that have not yet used the word(s) in commerce.

Dave

[wige]

From what I can figure out (mostly intuition and guesses based on observation), what comes off the "TrustRank Server" is sites that are blacklisted through user submission (there is a report form in Firefox that sends it's report directly to Google) and through the crawler. One specific thing the crawler looks for is malware that is pushed to users. Supposedly this can also spot malware contained in iframes.

[crankydave]

Interesting wige. I wonder if this could be used, at least in part, for their "this site may harm your computer" links and information.

As an example, 3rd site for this search...

week 9 nfl injury reports - Google Search

Dave

[wige]

Oddly enough, the trustrank for that page comes back ok. I wonder if it was one of the sites affected by the ad server attack a few weeks ago. Anyway. It is interesting that the SERP shows a problem and the TR server does not.

[wige]

I did another test, going to a phishing site (in this case an e-bay spoof site first detected and reported an hour ago). The response from the TrustRank front end was:

phishy:1:1

This triggered the normal bells and whistles from the browser. I accessed a few others that were still working and looked for them in Google using site: and inurl: operators, but got no listings on any of them. Trying various permutations gave me other sites on those domains, but no results for the spoofing sites themselves.

[crankydave]

Interesting again wige.

On the surface, it appears to delivering the likelyhood (ranking?) of site being "bad" rather than a particular level of "good".

I'm using "bad" and "good" loosely.

Dave

[wige]

Pretty much. Good sites come back with an empty response, and phishing sites come back with the quoted data. The more I look at this, though, the more it seems the data is user generated and pulled from additional sources, rather than determined algorithmically. Most of the sites are gone before Google would even have a chance to index them - after about three days on the phish list, they are mostly offline.