Power Googling: Your most sophisticated google search?

[philipjterry]

I been experimenting with power googling... for example I just crafted this one to learn more about advanced google operators from reputable sources....

(advanced OR power) AND (~search OR operators) intitle:google filetype:pdf site:.edu

What's your experience been like with advanced google search features...?

[lucy24]

It's unnerving when you add a term or constrain the search in some other way, and the reported total number of hits gets bigger.

[philipjterry]

I always wonder why people who write or talk about this dwell on reducing the number of results as though it's better. My hypothesis is the more results = more competition = better natural selection from google (like in the mating sense) survival of the fittest!

[philipjterry]

Just twigged, it's the seo's who are trying to pin point there competition by reducing it down to the people who know what they are doing...

[tangor]

I expect a "power search" to be more restrictive. I want fewer results. When the results increase, it makes me wonder if the SE (any of them) have their ducks in a row.

[lucy24]

Exactly.

The search engine found 20,000 pages that mention widgets.

It found 300,000 pages that mention blue widgets.

Obviously I am making up these examples, but I really have made searches where adding a term increases the total number of putative hits by an order of magnitude. I say "putative" because, as we all know, we are never allowed to see beyond the first 1000, and therefore have only the search engine's word for it that they exist at all.

Hypothetical explanation: some terms are so common in searches that they function as stopwords-- that is, they are not counted at all-- unless the total number of hits drops below some number.

"it's": 25 billion hits (10^9, not 10^12)
"my": 4 billion hits
"it's my": 6 billion hits

I pulled those off g### Advanced Search, U.S. edition, using my ordinary browser from home.

[philipjterry]

I'm learning now to take each search case by case, almost like coming up with an individual strategy for each search which requires a couple of minutes preparation at most before jumping in. And this requires a bit of understanding of the topic and how far away the results are from what i expected. The bottom line is, how can i retrieve the most valuable insights? I don't want these rehashes of existing content matter that offer nothing new - you only have to search 'advanced search operators' and practically all the results are the same - repetitive defintions of each operator without a benefit or a story which makes meaning.

The widget is a funny one because I hear it a lot in the seo community - if you ask an seo to give you an example a big proportion will use the widget example with a color. And seo's are the ones who blog like maniacs, you dont see widget mfr's blogging their heart out. So that explains that one. Another meme gone wild!

[lucy24]

The widget is a funny one because I hear it a lot in the seo community

It becomes self-reinforcing: think up an example and it's the first thing that comes to mind. But you can get the same results with real-life searches. Uh, oh, here's where we spin-off to Foo: Your Most Improbable Search Result.

language: 4 billion
inuktitut: 4 million
inuktitut language (no quotes or plusses): 22 million

If you tell g### to start at no. 900, you're taken to pages 87, 63 and 68 respectively. Plug that into the right exponential formula and you can probably get the two sets of numbers to end up roughly proportional. But the question remains: where did those extra 18 million and/or 50 pages come from?

[philipjterry]

lucy24, are you a mathematician?! :)

I noticed you anchor your thought process off of the numbers a lot, which admittedly is not something I do. Until now that is! I tend to anchor my thoughts off of where is the value; that gold nugget that fills the gap in my understanding of a subject matter, that changes everything. Since I improved my searching techniques I been getting a lot more of these ahaaa moments...

A good point about the numbers though - I'm getting 26:28 million or 2 million stray pages... When used in conversation people are more likely to say "inuktitut language" than just inuktitut, doesn't it sound more natural? I wonder if this has revealed a non obvious variable in googles algo?

[lucy24]

Mathematician, nope, not me. That would be my son the math major, or possibly my father the crystallographer. People claim I look at things analytically, but this is not be how I see myself. Maybe those people should get together with the other people who mutter that I over-anthropomorphize robots ;)

Anyway! I went over and tried the same two-word search at g### Canada. 870,000,000, close to 5 million... and 28 million. Or g###.co.uk: 4 billion, 27 million and 28 million. Same for .au, .za and .nz. And for US (.com, that is), using a different browser.

I love doing parallel searches in google vanilla and google dot other countries. Don't you wonder about those three million references to language that are only of interest to the non-Canadian part of the English-speaking world?

I did find one in my art studio's logs where you could see that the searcher was absolutely determined to find the studio director, because the search came through as "firstname lastname city state {exact name of studio}". I had to wonder how many shorter searches they tried before dumping in everything they could possibly think of. (Except grouping quotation marks, which would have got them there a lot sooner.)

Even with that massive list, it's still got a quarter of a million hits-- but the page they were looking for is in the top ten. And either g### couldn't find any suitable paid links or my ad blocker is working really, really well, because it starts right at the top of the page. If you put quotation marks around "firstname last name" and "exact name of studio" the quarter-million drops to four.

Hm. Lessee if any of those other three pages can be coerced into giving us a link. They're all local.

[aakk9999]

Maybe - when you search for:
widgets - you get everything to do with widgets
blue widgets - you get everything to do with blue and also everything to do with widgets (so the result set grows)
And maybe, to filter down, you need to search for
+blue +widgets
or for "blue widgets" - where you would get everything to do with both, blue and widgets and this should make result set smaller?

[lucy24]

That would imply that the default operator is OR. Isn't it supposed to be AND?

Every word matters. Generally, all the words you put in the query will be used.

Mark that "generally".

A particular word might not appear on a page in your results if there is sufficient other evidence that the page is relevant. The evidence might come from language analysis that Google has done or many other sources. For example, the query [ overhead view of the bellagio pool ] will give you nice overhead pictures from pages that do not include the word 'overhead.'

And thereby hang several hundred Forums threads ;)

[phranque]

aakk9999:
google removed support for the + operator several months ago.

Operators and more search help:
http://support.google.com/websearch/bin/answer.py?hl=en&answer=136861 [support.google.com]