Google: fu or F.U.?

If you read the Google help documentation for Google, it says that it will search for all of the words in the query. For instance, Googling “white house” will bring back a list of sites that mention the White House, along with other sites that mention a white house. We don’t expect that Google will return sites that just mention “white” or “house”, but not both.

Why then, when I am searching for Miscrosoft documentation on a specific subject, does Google see fit to display entries that DO NOT CONTAIN ALL OF THE SEARCH TERMS?

For instance, I was trying to find out how to work with the MFC class CFileDialog, so I entered “CFileDialog” into Google, but I got back too many replies that were not what I was looking for, so I changed it to “MFC CFileDialog”. That was better, but I wanted to find a tutorial, so I added “tutorial” to the end of the search terms. None of the first three entries have the word “tutorial” on the page.

I have found the same thing to be true of Bing as well. It seems like adding more terms to narrow down what the search engine returns can actually increase the number of “matching” sites!

Didn’t some search engine once have an option to refine your search starting with the currently active search result and winnowing the chaff from the wheat that way?

I checked, first entry has tutorial 13 times.

My search

The word order makes a difference. You might get better results with “tutorial” as the first word.

Bing does not seem to have any problems either–the results show lots of tutorials. It might help if the OP gave an actual example of a search that did not work.

It can also include words that were used in links to the page. So if someone elsewhere on the web wrote “Here’s a good tutorial”, then Google will associate the word “tutorial” with that site.

I read somewhere (can’t find the cite, sorry) that Google had changed its policy circa 2009, and that the search engine no longer feels obligated to include all your words. When I do the


mfc cfiledialog tutorial

search, I get results without “tutorial” but with the word “example” highlighted in bold; so somehow Google figured out that an example and a tutorial are similar concepts.

You can use the + prefix:


+mfc +cfiledialog +tutorial

and then you get strictly what you asked for.

This is what I typed in Google’s input:

mfc CFileDialog tutorial

No quotes. The first three pages that Google displays do NOT contain the word “tutorial”.

Ah, “example”, I’d forgotten about that. Before I searched for “mfc CFileDialog tutorial”, I had searched for “mfc CFileDialog example” and got the same first three hits, all with “example” highlighted in each occurrance of “for example,”.

If Google changed their policy about considering all of the search terms, they should have also changed their documentation, which, as of May of 2014, they had not.

Also, google will also search for synonyms to the terms you enter, even expanding an acronym into its various meanings (e.g. psu returns results about 2 different universities, power supplies, and playstation save files).

If you don’t want to wade through tons of irrelevant search results, you might want to read up on google’s advanced search operators. Specifically the following:
[ul]
[li] wrapping a word in quotes (e.g. “manual”) will exclude that word from synonym searching [/li][li] wrapping a phrase in quotes (e.g. “user manual”) will search for those words (and possibly synonyms of them) in that exact order, with nothing between them. [/li][li]putting - or NOT in front of a phrase will eliminate results containing that phrase (e.g. -sale when you’re only interested in facts or news about an item)[/li][li] putting + or AND in front of an term means that term must be included in search results (e.g. epson +lawsuit if you’re interested in people suing Epson)[/li][li]putting intext: in front of a term means that term must appear somewhere within the text of a page, not just in metadata, links to the page, or it’s title. (e.g. espy software intext:scam will return articles about espy software being a scam, but will prevent pages (i.e. espy’s homepage) that are simply described as a scam [/li][li] you can put site: before a web address to limit your search to that website (e.g. submarine site:boards.straightdope.com would show all submarine-related posts in these forums)[/li][/ul]

The examples above should not be taken to imply that espy software is or is not a scam (or even that there’s a product/company named espy software), that Epson is or has sued or been sued, or that there are any posts about submarines on these forums. Also, if one of those happens to be true it should not be inferred that any others are also true.

Nevertheless, there are in fact oodles of threads about submarines on this message board. The google search:


site:boards.straightdope.com submarines

will find you pages of them.

Running the same query, I get “tutorial” results about halfway down the first page. Regardless, Heracles and dstarfire explained the issue. The search is trying to be helpful by including synonyms.

Usually Google auto completes what I am searching for. How does it know?

As of 2011 this no longer works. Google retired, or rather changed the semantics of, the + operator so that it searches Google+ pages. This was never formally announced, though it is reflected in their official documentation, and did not go unnoticed in the tech press and on Google support forums:
[ul]
[li]Punctuation, symbols & operators in search (Google Support) [/li][quote]
Symbol: + What you can use it for: Search for Google+ pages
[/quote]

[li]Why was the Plus Sign (+) removed as a Search Operator? (Google product forums) [/li][quote]
The + operator was retired when Google+ was launched, because + was needed as a searchable character rather than an operator.
[/quote]

[li]Google Kills Its Other Plus, and How to Bring It Back (Wired) [/li][quote]
On Wednesday, Google retired a longer-standing “plus”: the + operator, a standard bit of syntax used to force words and phrases to appear in search results… Google wouldn’t disclose exactly why they phased it out, though it seems obvious that they’re paving the way for Google+ profile searches.
[/quote]

[/ul]
Nowadays to indicate that a search term must appear in the result pages, you need to wrap it in double quotation marks, “like this”.

This is interesting, and at least provides that cite for the 2009 change I was missing upthread.

But have you tried using + today, in 2015, as mentioned above ? It certainly works for me. They must have changed their mind since 2011 (and not told anyone!).

Guess what the sixth search result is now for "mfc “CFileDialog” “tutorial” …

You made two mistakes.

1–The period is supposed to go inside quotation marks.

2–You gave an example of regular quotation marks.

OffByOne very likely did, in fact, given an example of the search that didn’t work- remember, what they type in and what you type in can give wildly different results; this is bubbling in action, folks.

Even with quotes Google usually ignores many of my search terms.

It’s ridiculous: enter 3 search terms and get too many results. Try to fine tune by adding a 4th: even more results since it now has more choices of which terms to drop.

The stuff that made Google great in the early days are mostly gone.

My example was to show just that. I was perhaps too subtle about it.

Just wait, they are getting ready to use “intelligent” computers to avoid giving search results that Google (or rather their computers) have declared not true.

Bolding mine. I believe in bold you’ve misspelled “profitable”.