Why does Google return less results in firefox?

&client=firefox halves the number of results?

[anandus]

Hi,

For some reason Google returns half the amount of results, when the client is identiefied as Firefox?
If I do a search for site:www.somesite.com in Firefox, it adds "&client=firefox".

When I remove this, I get more (often double) the amount of results.
The same goes when I add this parameter in IE for a search.

Anybody any clue why this might be happening?

[rubenski]

Tried this in IE and I see the same thing happening. Adding the client variable to the query string influences the number of results a search yields sometims by a factor 2+. Strange stuff. Any clues why this could happen?

[edited by: rubenski at 10:03 pm (utc) on Oct. 6, 2004]

[peterdaly]

Does it only show pages it think firefox can properly render?

That's just a wild guess theory.

[sabai]

That's just a wild guess theory.

Pretty wild - firefox is better at rendering pages correctly than anything else out there.

[moltar]

It seems like as long as &client= is present, it returns different set of results. For example you can say:

&client=1

and it will return a different set. But it is always the same, no matter what the value of

client

is

[rubenski]

...and it doesn't matter if you use Firefox or IE. It happens in both browsers. Although IE doesn't seem to get the client variable by default. I entered it manually. (IE 6.0, Windows XP pro)

[anandus]

Maybe it has to do with serverload on the webservers?

If it's identified as a browser - and thus require more resources than an auytomated query (linke bandwidth for images and such, it'll show less?

Okay, true it's a bit far fetched.

Or maybe it only shwos W3C-compliant sites? And leaves out badly programmed sites?

Come to think of it, this latter idea might be more on track...

[Teknorat]

Neither of those sound like a logical thing to do. Anyone know for sure?

[raptorix]

Sounds like spyware in IE, i had it a time ago, some malicious spyware tool mixed up google results with strange urls

[anandus]

Well I don't think it's spyware, as everybody can see this from happening, independent of the computer you using.

Maybe it's the way google uses the parameters? One parameter too many? Misinterpretation?

[webdude]

I have been seeing this for the past 6 months. I always thought that the client affected the returns. Not sure why, but it is there.

[jamesa]

That only appears in the query string when you initiate the search using the Firefox search bar (as opposed to the search field on the page).

Why? I dunno.

[Chndru]

I use &sourceid=firefox and it spits out the same number of results on both Firefox and IE.

[anandus]

@ Jamesa:
It doens't only do it with the searchbox in firefox.

If you search in IE, and paste this behind the url, you get the same results.
paste this:
&client=banana

[valeyard]

The obvious theory is that "client=" is a flag Google uses for something other than a browser type.

At a quick look, adding "client=1" seems to return approximately the same number of results as using Safesearch.

Guess: perhaps "client=whatever" refers to queries coming from third parties using Google as a back-end? Google is filtering the results to avoid offending their user base.

[anandus]

that seemed to be a smart idea, but I tested this and I got different results for Safesearch, client=, both, and none, so I don't think that's it.

But maybe you're thinking in the right dierection..

[g1smd]

Here on Moz 1.6 I have a query where I use &num=100&filter=0 to show unfiltered results with 100 results to the page.

Use the same query in IE 5.5 and you cannot see unfiltered results. It just doesn't work.

[jamesa]

@ Jamesa:
It doens't only do it with the searchbox in firefox.

If you search in IE, and paste this behind the url, you get the same results.
paste this:
&client=banana

Yea, I know. What I meant is that Google isn't browser sniffing. If you search using Firefox and enter the search terms in Google's search box on the page itself, the results will be the same as any other browser. But if you do a search using Firefox's built in search bar, then the client=xx is added. So this would be Firefox modifying the URL, not Google.

Why the number of results is different, though, I don't know. The adult filter was the best theory yet but that was disproven.

[moltar]

I found this on a website dated 2003/10/27:

The toolbar buttons provided by Google also provide a `client' input with `googlet' value which seems to do nothing (I tried both ways). I left it in, guessing that they might be collecting statistics of toolbar button usage, and the Jargon version probably should be included.

Now we at least know it was used by toolbar before.

[moltar]

On another website I found info that it is still used to Toolbar internally. It is used to find website's PR. The following request is passed:

http://www.google.com/search?[b]client=navclient-auto[/b]&ch=[checksum here] &features=Rank&q=info:http://www.example.com

[moltar]

And finally... &client is also used for AdSense for search. Your user id is identified there. E.g.:

&client=pub-XXXXXXXXXXXXXXXXXXXXX

[moltar]

And finally another answer, that makes sense the most of them, is from Google's Search Results Protocols [elastic.org] page:

/search?q=<REQ>&num=<NUM>&start=<START>&output=<FORMAT>&client=<PARTNER>

<PARTNER> is a string which serves to identify the partner to Google. If you do not know the <PARTNER> string that you are supposed to use in conjunction with the Google search results protocol, please ask your sales contact at Google for it.

[anandus]

Today Firefox 1.0 came out, and when I go to Google, I see this:
[google.be...]

Maybe it has something to do with this?

[borisbaloney]

Today Firefox 1.0 came out, and when I go to Google, I see this:

You mean this?
[google.com...]