Why Does The Google Client Parameter Give Less Results?

[inlogicalbearer]

When you use Firefox search box, Google add the "&client=firefox-a" to the URL and it display a lot less results. For the word "the" you got 2,5 billions results and if you erase the parameter you jump to 5,9 billions !!!
It does the same if any string is after the &client=. Why ?

[Mel]

I am running Firefox 1.0PR and do not see those results at all.
I am seeing:

Results 1 - 10 of about 5,840,000,000 for "the"

with a search url of:
http://www.google.com/search?sourcei...-8&q=%22the%22

Have you by any chance changed the user agent in your browser?

[inlogicalbearer]

I got also 1.0 pr Mel, but when using the right search box it add those parameters to my url
http://www.google.com/search?q=the&c...en-US:official and I got less results ? Weird !

[dannysullivan]

Great find! It doesn't look like a Firefox thing. Instead, it seems like ANY client parameter will make the count decrease. I've blogged more about this here: Client Parameter Makes Google Report Fewer Matches

[rustybrick]

Just to clarify, I believe this was first found at WebmasterWorld at this thread http://www.webmasterworld.com/forum3/26116.htm

Just making sure credit goes to those who found it.

I found it very neat as well and blogged on it as well.

[inlogicalbearer]

Some other search sites using Google got thoses &client parameter. If you go to La Toile du Québec and select the Google query you will see &client=netgraphe in the URL. Netgraphe is the name of the company who own www.Toile.com
http://google.canoe.com/toile/index....gorie=&ad=w5n2
&client=netgraphe&ie=latin1&oe=latin1
&output=xml_no_dtd
&ip=&userid=netgraphe
&start=0&num=20&lr=&query=car&x=31&y=7

I'm suspecting Google give different "view" to outsiders request to his database. Go look any site with the AdSense Google search box. You will have the same number of results using it or using the &client parameter. It's also like the Google API, you will have differents results, and less, using it if you compare to normal use of Google.

[dannysullivan]

Yeah, Barry -- saw your post after I did mind, followed through to the WW thread and very good. Gary had also pointed out to me some issues of a9 showing the same thing. In short, client seems to equal Google partner, not browser client -- and I wonder if Firefox simply made a mistake in using the wrong parameter. I added a postscript explaining this more to my blog post.

[rustybrick]

Yea, funny how these things come about. Hype, Hype, Hype and then, oh it was just a parameter used for this case. So firefox is to blame for misuse of the parameter. Interesting find, nonetheless.

[Dodger]

Originally Posted by rustybrick

Yea, funny how these things come about. Hype, Hype, Hype and then, oh it was just a parameter used for this case. So firefox is to blame for misuse of the parameter. Interesting find, nonetheless.

Misuse? Why do you say that Barry?

In order for it to be a misuse of this parameter, they would had to have that parameter set to "firefox-a" for some other purpose wouldn't they? I cannot think of anything they have that would be using that parameter right off-hand.

[rustybrick]

Dodger,

I think its misused because they have no reason to include such a parameter. Doing so limits the results of the search from the default search parameters. Maybe it was an oversight? Maybe I don't fully understand it?

[Dodger]

Okay, sounds good.

What interested me the most about this though, is the Client ID itself -- firefox-a. I could understand just firefox, if it were being misused as a browser client ID.

The -a on the end of it is what has me thinking that this was actually assigned to Mozilla at one time or another by Google. They did not pull this out of the air, in other words.

[Dodger]

The folks at MozillaZine (actually a whole lot of 'em) saw my blog entry about this and are working on a fix for it -- I think it is more like a work-around. They are not happy with that little search box returning half as many results.

They do not know where or how the client ID is getting injected into the URL. Keep mentioning a google.src file, but they cannot see the client-id inside of it.

[Dodger]

Boy that was quick. No sooner did I come over here and post and checked back there -- somebody had an answer.

The google.src was written (or provided) by someone at Google (according to the poster). Removing the following line of code from your google.src file will stop the search box from injecting the client parameter in the url:

<input name="sourceid" value="mozilla-search">

The google.src file is located in the searchplugins directory inside of the application directory.

.

[rustybrick]

Nice work Dodger!

[inlogicalbearer]

Not working for me Dodger. Or half way. I don't have that particular line in mine and I try to remove this

update="http://www.google.com/mozilla/google.src"
updateCheckDays=1
queryEncoding="utf-8"
queryCharset="utf-8"

... and not work

But when I remove thoses ...

<inputnext name="start" factor="10">
<inputprev name="start" factor="10">
<input name="ie" value="utf-8">
<input name="oe" value="utf-8">

...It stop writing thoses last parameters in my URL

[Dodger]

I don't have it in mine either. Here is the code in that file on mine (note the email address on the first line too)

# Mozilla/Google plug-in by amitp+mozilla@google.com

<search 
   name="Google"
   description="Google Search"
   method="GET"
   action="http://www.google.com/search"
   update="http://www.google.com/mozilla/google.src"
   updateCheckDays=1   
   queryEncoding="utf-8"
   queryCharset="utf-8"
>

<input name="q" user>
<inputnext name="start" factor="10">
<inputprev name="start" factor="10">
<input name="ie" value="utf-8">
<input name="oe" value="utf-8">

<interpret 
    browserResultType="result" 
    charset = "UTF-8"
    resultListStart="<!--a-->" 
    resultListEnd="<!--z-->" 
    resultItemStart="<!--m-->" 
    resultItemEnd="<!--n-->"
>
</search>

I would not remove any of the other lines inlogicalbearer. This file is like a template for the URL string that gets sent to Google.

There are different builds in Firefox, and I am wondering if that might have something to do with it. I am thinking that the string for the client-id is listed somewhere else, maybe hard coded into a binary file.

when searching for 'xml editor' with parameter hl='en' our website, www.xmlblueprint.com, ranks on page 3 (position 24), but with parameter hl='nl' (language: Dutch) the website ranks on page 2 (position 13). Mind, i'm not searching for Dutch websites and our website is in English.
It seems Google dislikes European websites?

[Marcia]

Just making sure credit goes to those who found it.

No doubt many have found it, but I'd like to give credit and a word of thanks to the person who found it and then posted about it for us here at SEW.

Thanks for posting about it and sharing with us, inlogicalbearer.