Is Google too smart?

[tahnan]

OK, Google just creeped me out. Someone posted a rather incoherent puzzle to a board I read, ending with the question "What is the prices how the four items."

I wanted to know if he'd just taken it from somewhere else, bad grammar and all, so I googled "what is the prices how".

Go ahead, follow that link. See if you can figure out why it turned up anything, and why that phrase doesn't seem to be anywhere in the search results.

The answer is that those pages contain the phrase What is the cost?.

And I'm just a little nervous that Google did that, for reasons I can't quite identify.

Comments

[flynngrrl.livejournal.com]

Wow. I'm impressed and scared... computers are making it easier for stupid people to catch up to us fabulously smart folk! And our big brains are slowing us down when we run!

[tahnan.livejournal.com]

Well, it does make it harder for me to Google for sentences that I think may not be grammatical--something I do occasionally as part of my research--if Google's going to correct my grammar.

I think I'm also worried that it'll just correct things badly. I don't want hits about Vincent Cost if I Google for Vincent Price.

[chanaleh.livejournal.com]

Actually, I'm not sure that's exactly why you're getting those results. I clicked on the "Cached" copy of the first few returns (to see what text it would highlight), and for each of them it gives no highlighted terms but says instead "These terms only appear in links pointing to this page: what is the prices how".

Which is a known factor in Google's calculations for page returns. So it's not that Google is semantically reprocessing your request, per se; it's that enough other people have used that exact phrase in ways that link it to pages containing the phrase "What is the cost", that the interconnections are unavoidably present. (Okay, now, but does that constitute semantic knowledge or doesn't it?)

I was going to suggest going next to Advanced Search and selecting "Occurrences: in text of page" which should suppress those sorts of returns -- except it doesn't seem to actually suppress them very well, since that's still all I'm getting. Oy gevalt.

Re: On the other hand...

[chanaleh.livejournal.com]

If you Google for the phrase "What are the prices of the four items" (http://www.google.com/search?hl=en&lr=lang_en&ie=ISO-8859-1&q=%22What+are+the+prices+of+the+four+items%22), you get two nice neat hits for your puzzle. :-P

Re: On the other hand...

[tahnan.livejournal.com]

Oh, I knew I could find the puzzle with a better search. I was wondering whether it came complete with typos from another board.

As for the other: well, bollocks. If the term "what is the prices how" pointed to those pages--and that's a heck of a lot of pages for people to link to with so incoherent a phrase--then there would be some hits for the phrase itself. Certainly the reason that Microsoft used to turn up on "more evil than Satan" was that people linked to Microsoft with that text...but after the first hit or two, you started getting the pages that used that text. How can hundreds of pages link to a wide variety of pages with the words "what is the prices how" without any of those pages showing up on Google?

[(Anonymous)]

Actually, all those pages contain the phrase "what is the cost how" (regardless of punctuation). Likely, google knows that the word "prices" is similar to the word "cost" and it did a replace on what you were searching for. If you google "what is the cost" or "what is the prices", you'll get over 175,000 hits vs the 271 hits you got from your search.

So don't be too afraid. Google is smart, but not that smart.

Evan
(a friend of thedan)

[luckylefty.livejournal.com]

But there must be something else odd going on. Googling for "what is the prices how" gave me 273 hits, all of which I checked contained
"what is the cost how" rather than "what is the prices how". But
googling for "what is the cost how" gave me 445 hits. How is Google deciding which of the "what is the cost how" hits should register as "what is the prices how" hits?

I like the google feature where a search sometimes puts up a "did you mean to search for..." link. But I don't like the "feature" where google decides on its own that I really meant to search for something other than what I said I was searching for, with no way to turn it off. Sometimes I want to search for a certain specific string of words, and Google would be more useful if it let me do that somehow!

[furycom.livejournal.com]

Would it be a problem if it returned exact matches first, then offered matches with synonyms? Could that be what happened here?