I know we're all a fan of Googlecounting results. But this is weird: When I search for a term in Firefox's Google box, the result count is exactly 200% what it should be.
1. Search for "Wikipedia is great" in the FireFox search box. 2. Mostly (but not always), Google reports 29,600 matches. 3. Repeat the search just by pressing Google's "Search" button. 4. Google reports 14,800 matches.
Strangely, the difference in count seems to correspond to a difference in layout. The 14,800 count groups related matches by indenting. (The second match is [[All your base are belong to us]], indented under [[WIkipedia:Why Wikipedia is so great]].) The 29,600 match shows no indenting.
FYI, here's the URL I'm using. http://www.google.com.au/search?q=%22wikipedia+is+great%22&ie=utf-8&...
Constantly pasting that into FireFox, sometimes I get 29,600, sometimes I get 14,800. Pasting into IE7 I always get 14,800. Anyone have an explanation?
Steve
On 2/8/07, Steve Bennett stevagewp@gmail.com wrote:
I know we're all a fan of Googlecounting results. But this is weird: When I search for a term in Firefox's Google box, the result count is exactly 200% what it should be.
- Search for "Wikipedia is great" in the FireFox search box.
- Mostly (but not always), Google reports 29,600 matches.
- Repeat the search just by pressing Google's "Search" button.
- Google reports 14,800 matches.
Strangely, the difference in count seems to correspond to a difference in layout. The 14,800 count groups related matches by indenting. (The second match is [[All your base are belong to us]], indented under [[WIkipedia:Why Wikipedia is so great]].) The 29,600 match shows no indenting.
FYI, here's the URL I'm using. http://www.google.com.au/search?q=%22wikipedia+is+great%22&ie=utf-8&...
Constantly pasting that into FireFox, sometimes I get 29,600, sometimes I get 14,800. Pasting into IE7 I always get 14,800. Anyone have an explanation?
Steve
Posing the question to a guy I know who works at Google received this response:
The count of results is just an approximation; it's not the actual count.
So I assume that he means it's figured on the fly, and is subject to random variation? Just proves you can't put any real stock in the number returned.
-anon
On 2/9/07, David Ashby humble.fool@gmail.com wrote:
Posing the question to a guy I know who works at Google received this response:
The count of results is just an approximation; it's not the actual count.
So I assume that he means it's figured on the fly, and is subject to random variation? Just proves you can't put any real stock in the number returned.
Well, that doesn't explain why the number randomly oscillates between precisely two numbers, with a different layout of the results on the screen. I actually suspect that what's happening is the "discard similar results" option is generating the smaller number with indenting, and that option is being set or not set randomly. Using a less popular query and specifically choosing to discard similar results gives a stable number.
Whether the bug is in how firefox interacts with google, or how google interacts with firefox is unclear.
I would also be curious to know if Google has any kind of confidence parameters on those estimates. Can one say for example that if query Q returns "approximately" 1,000,000 results and P returns "approximately" 1,500,000 results - then are there definitely more real results for P than Q?
Steve
Steve
Take a look at whether or not your google results are "localized" when you use the toolbar. For example (I'm based in NZ) and quite often when I use the toolbar to search I get local results only (i.e. all from xxx.co.nz)
However, when I hit search on the page itself, using the same keywords, it does a full "international" search and the number of results increases dramatically.
Just an idea...
Glen
On 2/9/07, Steve Bennett stevagewp@gmail.com wrote:
On 2/9/07, David Ashby humble.fool@gmail.com wrote:
Posing the question to a guy I know who works at Google received this
response:
The count of results is just an approximation; it's not the actual
count.
So I assume that he means it's figured on the fly, and is subject to random variation? Just proves you can't put any real stock in the number returned.
Well, that doesn't explain why the number randomly oscillates between precisely two numbers, with a different layout of the results on the screen. I actually suspect that what's happening is the "discard similar results" option is generating the smaller number with indenting, and that option is being set or not set randomly. Using a less popular query and specifically choosing to discard similar results gives a stable number.
Whether the bug is in how firefox interacts with google, or how google interacts with firefox is unclear.
I would also be curious to know if Google has any kind of confidence parameters on those estimates. Can one say for example that if query Q returns "approximately" 1,000,000 results and P returns "approximately" 1,500,000 results - then are there definitely more real results for P than Q?
Steve
Steve
WikiEN-l mailing list WikiEN-l@lists.wikimedia.org To unsubscribe from this mailing list, visit: http://lists.wikimedia.org/mailman/listinfo/wikien-l
Steve Bennett wrote:
I know we're all a fan of Googlecounting results. But this is weird: When I search for a term in Firefox's Google box, the result count is exactly 200% what it should be.
I had a similar problem a little while ago. Word counts against Google using Firefox were off by some significant factor (in my case I think it was *ten*) from results returned by lynx or other browsers.
It was driving me nuts. I eventually discovered I could make the problem go away by... clearing my cookies. Go figure.
Steve Bennett wrote:
I know we're all a fan of Googlecounting results. But this is weird: When I search for a term in Firefox's Google box, the result count is exactly 200% what it should be.
- Search for "Wikipedia is great" in the FireFox search box.
- Mostly (but not always), Google reports 29,600 matches.
- Repeat the search just by pressing Google's "Search" button.
- Google reports 14,800 matches.
It's because every search request is routed to a randomly-selected Google server farm. Some of those farms may be running on the newest index, while some may still be waiting for the latest index update. This effect is not unique to the search query you entered; lots of people keep noticing it and wondering about it. I'm pretty sure there is an entry about it in one of Google's many FAQs, but I couldn't find it.
Timwi
Timwi wrote:
Steve Bennett wrote:
I know we're all a fan of Googlecounting results. But this is weird: When I search for a term in Firefox's Google box, the result count is exactly 200% what it should be.
- Search for "Wikipedia is great" in the FireFox search box.
- Mostly (but not always), Google reports 29,600 matches.
- Repeat the search just by pressing Google's "Search" button.
- Google reports 14,800 matches.
It's because every search request is routed to a randomly-selected Google server farm. Some of those farms may be running on the newest index, while some may still be waiting for the latest index update. This effect is not unique to the search query you entered; lots of people keep noticing it and wondering about it. I'm pretty sure there is an entry about it in one of Google's many FAQs, but I couldn't find it.
Turns out the proper term is "Google Dance", and googling for that gives us an interesting blog entry. http://dance.efactory.de/
On 2/9/07, Timwi timwi@gmx.net wrote:
It's because every search request is routed to a randomly-selected Google server farm. Some of those farms may be running on the newest index, while some may still be waiting for the latest index update. This effect is not unique to the search query you entered; lots of people keep noticing it and wondering about it. I'm pretty sure there is an entry about it in one of Google's many FAQs, but I couldn't find it.
Turns out the proper term is "Google Dance", and googling for that gives us an interesting blog entry. http://dance.efactory.de/
Ah, randomly selected servers - that explains it perfectly. But I wonder why I don't get the effect with IE. Maybe IE caches the server's IP or something.
Steve
Probably because when you go to http://www.google.com/ it goes to one server but when you use the firefox search box (http://firefox.google.com/ or whatever) it goes to another one. -Yonatan
On 2/9/07, Steve Bennett stevagewp@gmail.com wrote:
On 2/9/07, Timwi timwi@gmx.net wrote:
It's because every search request is routed to a randomly-selected Google server farm. Some of those farms may be running on the newest index, while some may still be waiting for the latest index update.
This
effect is not unique to the search query you entered; lots of people keep noticing it and wondering about it. I'm pretty sure there is an entry about it in one of Google's many FAQs, but I couldn't find it.
Turns out the proper term is "Google Dance", and googling for that gives us an interesting blog entry. http://dance.efactory.de/
Ah, randomly selected servers - that explains it perfectly. But I wonder why I don't get the effect with IE. Maybe IE caches the server's IP or something.
Steve
WikiEN-l mailing list WikiEN-l@lists.wikimedia.org To unsubscribe from this mailing list, visit: http://lists.wikimedia.org/mailman/listinfo/wikien-l