Trivia: What percentage of google queries consist of concatenated words, like 2007kentuckyderby?

Answer: I don't know, but they show up periodically as keywords in google queries pointing to my site. This has to drive the query engineers nuts, right?

Basically, some people don't know how to use google, so they just type in everything they're looking for and concatenate it into one word for every query (yes, people do this, I see it in my logs, and Eszter Hargittai says they do). Were that the only problem, the query engineer just would have written a nice probabilistic word exploder to transform "kentuckyderby" into "kentucky derby" and gone home. They can't though, because this breaks a search for the userid "kentuckyderbyfan" which would get exploded into "kentucky derby fan", losing google's lovely specificity.

I figure I ought to help google out a bit and hook my blog code to catch a google referral and insert a leading div that tells the person how to use google, so if you check out the query for 2007kentuckyderby, you'll see this in action. Otherwise, google just has to hope that user will sign up for gmail and then google will be able to look at their query stream and go, "Oh right, you're one of those people."


Sometimes, usually due to field focus issues, I end up pasting a URL into Google's search box. I wish they'd put a clickable link to that URL on the results page instead of or in addition to a message saying there are no results. That would save me valuable seconds every time it happens. -- Don
You could do that with a quick GreaseMonkey script for firefox. Just look for the q textbox, when it has a url in it, print it below as an href. -- Patrick.
Don, I am sure they don't do that because of xss. Not that they couldn't thwart that, but it's one more piece to the machine.
I just noticed something. On match.com, they have "match" words. Kind of like keys or indexes for you to associate with your profile. However (and there is where it is slightly related) if you put in a two word "word", they concat it together into one word...indian food => indianfood. Now, I know this isn't exactly what you are talking about above, but it does show the developers have tried to remove ambiguity, making parsing easier...I guess the folks at match.com aren't as skilled as the people at google :) -- Nathan (p.s. the comment above is mine as well).