Do you have a question? Post it now! No Registration Necessary. Now with pictures!
- Posted on
- google page count
June 9, 2004, 7:03 pm
rate this thread
re-elected ;) ?
searching <bush president> yields 7 290 000 pages
searching <bush -president> yields 7 350 000 pages
Let us now do a basic check: searching <bush> should yield 14 640 000
pages (the sum of the two: pages with the word bush AND the word
president, plus pages with the word bush WITHOUT the word president).
Alas! it yields 32 000 000 pages. More than twice as much !
Using page counts in Google could be very useful... if they were
My question is: How does Google calculate its page counts?
Re: google page count
You are over simplifying the word search process. You should read
about word associations in Artificial Intelligence and Intelligent
Enter bush president - I get 8 250 000 references
bush -president : 8 410 000, one link is
http://myflorida.com/b_eog/owa/b_eog_www.html.main_page which is about
Bush's brother. But the word president appears on that page -
"MOURNING FAMILIES TOUCHED BY PRESIDENT"
Enter: bush -president
site:http://myflorida.com/b_eog/owa/b_eog_www.html.main_page and see
no links are returned
Enter: bush site: http://myflorida.com/b_eog/owa/b_eog_www.html.main_page
and see 3 links
Enter: bush -frogs site:
http://myflorida.com/b_eog/owa/b_eog_www.html.main_page - 3 links (non
are governor's office, presviously listed as bush -president)
Enter: http://myflorida.com/b_eog/owa/b_eog_www.html.main_page see
link of Gov Bush, and no links for
It is obvious these results dont correspond.
What google has done is collected a lot of data which always gives the
appearance of a good result. But if 10 references match, google can
list maybe 1 of them. In other words google benefits from large
resources and a large Web. Those resources cost millions, or billions.
Google has then made the absurd connection that the number of links to
a website matters.
The optimisation modules in google are not very good. If the google
methods were worked on a smaller population of sites which needed
better optimising, the searches would fail.
The only way to build good searches is by associating words and
phrasing - that is how people talk. This subject has been researched
for decades and no one has cracked it.
President GW Bush
President George Bush
American President - Bush
Head of State Bush
G Bush Jr
Former Gov Bush
This all means that same thing to a person, but an algorithm cannot
make the connection.
If I say: Bush is popular, or G Bush is popular etc then people would
know the sentences are the same.
- » Yet another evil trick to work-around Google getting smarter
- — Next thread in » Search Engines