For anyone who's been away from civilization for the past few days, the big news is that Google have launched a beta version of "Google Suggest", which is like autocomplete for searching, based on things for which other users have searched. The UK version can be found at http://www.google.co.uk/webhp?complete=1&hl=en .

Now, I'm not sure exactly how useful this is, but it's damned good fun!  I've already wasted hours keying in a few letters to see what results come back. My geek curiosity takes hold - what's the top result for each letter of the alphabet?  How many letters of my name do I need to key in before I appear at the top of the options?  What about Jocelyn?  "J", "O", "C"... "Jock Itch"!?  Oh, she'll love that...

Assuming they're ranked entirely based on frequency of previous searches, then it's an interesting insight into search term popularity - a more dynamic version of the Google Zeitgeist pages. But I have my suspicions that there is a bit of manipulation and/or censorship going on here. "S", "E"... "Sears"?!  Are you kidding me?  "X". What?  It doesn't appear at all?  Nobody searches for sex on the internet any more?  Hmmm. So then began the thoroughly infantile (yet amusing) task of trying to find all the "rude" words I could in the Google Suggestions. I remember doing something similar with a dictionary when I was in junior school..

Having returned to my grown-up persona, I started to consider how incredibly useful this technology could be when applied to other web applications - when searching for customer accounts, users, locations, or the many other things we need to allow users to easily find. Negating the need for a postback would be a Godsend for the users, and this is certainly something I'll be looking into. Meanwhile, Eric Bachtal has this excellent early commentary on how the magic happens:

http://ewbi.blogs.com/develops/2004/12/google_suggest.html