During a long-winded IM (they all are) I found myself explaining the appeal of Twitter to a friend who didn’t quite understand it. Being my friend, by definition, means he says and does absurd and hilarious things all the time, of which I’m absolutely confident I’m only hearing about 10-20% of it. Why miss out? So I underhandedly convinced him to use twitter and learned something about the limitations of search engines all at the same time.

Knowing he was competitive, I appealed to his sense of one-upsmanship and discussed a competition whereby he would have to post some minimum number of tweets anonymously, and see if I could figure out what his username was. The guidelines were that he had to post as himself, about things that occurred during his day, and about things which would be identifiable enough to locate him based on how well I knew him. The last part was the one I was most worried about (as I said, he is super competitive).

I had over a month to try to find my friends’ twitter account and I failed, but for an interesting reason. Search engines such as Google are designed to help you find *one* web page given the search terms you input. They are nearly useless in finding a collection of related pages (e.g. tweets) if you have a long list of possible content. For example, I know which teams he is a fan of, where he usually eats, where he lives, and some other personal particulars. If I search for Taco Bell, Michigan Basketball, falafel trucks, using these terms individually will give me too many responses to sift through, and collectively, they will not identify a single page – remember, each tweet is a separate page.

What would have been useful, but doesn’t exist, is a search engine wherein the input is a collection of search terms and the output is a collection of related pages, which would be useful if you were actually trying to discover new sites to read. Hypothetically, let’s say there are some fixed number of sites that I check on a regular basis, either in the browser or through an RSS reader. If there were another site out there which would be right up my alley, given my interests, there is no way I would be able to find it. Google Reader has a feature which may use some variation of this, where it may suggest newsfeeds you might like, although I can’t say it’s ever worked for me. It’s almost like we need a search engine version of the Netflix recommendation algorithm – Given a set of things I like or input, what other things are likely to be enjoyable to me?

Anyhow, his tweets are funny, I lost $1, and learned something about search engines. I will say his entries, collectively, give him away, there is no way I would have been able to find him.