« Change For The Better | Main | Where Have The Lions Of Literature Gone? »

Searching For Searches

Does anyone really know how search engines work? I know in theory they are supposed to scour the web in response to commands given by the person using the service and respond with the addresses of web sites that have the information pertinent to what was requested. But the practice seems to be something else altogether.

Take for example the other night. Just out of idle curiosity my wife wanted to find out the history of the word pretty. As her search command she had asked for "history of the word pretty" (without the quotation marks). What she got in return was anything but pretty.

Not one site in the top forty offered had anything vaguely to do with the history of the word. Instead the majority of them were simply sites where the word pretty was being used and no mind being paid to the original request. After trying a number of variations using history and pretty returned pretty much the same results; which in turn resulted in various comments being made on the parentage and history of Google other search engines were consulted.

When it became obvious that this was not a Goggle specific problem, and that all search engines seemed to be particularly obtuse when it came to try and finding out "something as fu*king simple as the history of a goddamned stupid, word for fu*ks sake" she gave up the search. I didn't blame her, because although sometimes Google can be your best friend, on other occasions search engines are only as good as what they are capable of doing.

The first thing we always have to remember when using anything to do with computers is they can't think outside of the parameters that have been defined for them. Search engines are designed to pick out keywords from something published on the Internet and match them to the search query. They have very few ways of knowing the actual context of the words in question; so will simply return the addresses of sites where they find the words requested.

At least this is what I've come to believe, and to be honest I can't come up with any other reasonable solution. I've been looking at how many people are directed to my site based on the fact that a word in the title or body of the text has matched their search request. That the post in question has absolutely nothing to do with the subject of their search is irrelevant to the search engine.

Of course to further confuse the poor person doing a search is the fact that some web sites are given more weight in a search then others. This doesn't mean it will show up if it doesn't contain the required keywords, but a site with a high ranking will be receive priority so long as it contains key words, even at the expense of one more relevant.

Google, and I'm sure other search engines do the same thing, have created their own arcane system for the ranking of sites, but I've yet to meet anyone who really understands how they do it. Perhaps it's based on the number of times the site in question has had a high number of keywords show up, the number and turnover of material, or heaven forefend, maybe even the quality of the material comes into play.

What is obvious is that no one seems to be keeping track of how pertinent the posts are to people's search requests. Of course that may be an impossible task to perform; just as it's impossible to teach the search engine the difference between pretty awful and pretty good.

I've heard of people who will spend hours coming up with keyword listings for a post to optimize their search engine results, as the ads put it, but I can's see why they bother. There doesn't seem to be anyway to influence how search engines are going to decide what to choose beyond a word in a post matching a search request. Now that's scientific.

Leap In The Dark

↑ Grab this Headline Animator

Google