Personalisation vs. Serendipity
One of the advantages of the current Web search engines is the possibility of finding stuff that just wasn't on our radar. Of course there is also the disadvantage of wading through vast quantities of unwanted dross. The current method of ranking the results of search queries, based upon the population of user's click-throughs, is pretty good at returning the most popular results and suffices most of the time, if you can get the query term right. However, by personalising (or adapting) to the individual user's needs and wishes it would be possible to prioritise the results of the search or perhaps reduce the volume of information. However does the adaptation reduce the opportunity of finding something unexpected?
If something like the "I'm Feeling Lucky" button on Google.com returning a serendipitous list of results, rather than the one resultant Web page currently by Google, could allow the user to find a wider and weirder match to the search query. Is the chance find more important than getting to what you really really want fast and first time round? The opportunity of reducing the time take to find is more important as the user is in focused "search mode" rather than a relaxed "surf mode" where the user is more likely to follow the serendipitous links.
Although serendipity is fun, expands our knowledge and wastes our time, targeted searches can reduce the frustration of not finding the information we seek giving us time to follow those happy accidents.