The Death and Rise of Search

O'Reilly Radar has an article by Tim O'Reilly called "Search Startups Are Dead, Long Live Search Startups". He notes that "In my talks on Web 2.0, I always end with the point that "a platform beats an application every time."

Hmmm.....I think there is actually more of a "helix" effect between platform and application. Platforms are critical until they become standardised - ie commoditised, at which point applications above that layer rise in importance - until they in turn become subsumed in the emerging next level platform structures....and so begins the Circle of Life again

To be fair, I think Tim O'Reilly may imply this when he says:

We're entering the platform phase of Web 2.0, in which first generation applications are going to turn into platforms, followed by a stage in which the leaders use that platform strength to outperform their application rivals, eventually closing them out of the market

Clearly, starting any search business to compete directly with the old platforms is pointless, it seems to me that the new applications area is where the value lies now - searches we get now are based on smart algorithms to search lousy metadata. If that metadata can be enriched, searches can get a lot better.

In my view the social network approach is already driving a new sort of search, more intelligent than the first generation, as users add tagging, rating and recommendation metadata to basic search, forming the folk taxonomies or "folksonomies".

However, this has its limitations, its unpredictable both in terms of coherent metadata structure and rollout - so the challenge is to arrive at easily structured metadata. I can see 3 possible approaches to this for a new player:

(i) Carve off (valuable) pieces of the searchspace and structure better metadata on the content than the majors can provide in that niche (sorry, vertical is the new "in" word I am told) space.
(ii) Add value on top of an existing search platform by structuring better activity based metadata
(iii) Add value to "me" by knowing "me" better - and this probably means some form of opt in where you volunteer information for a more relevant search.

These will most likely ride on top of the exisiting search engines, like the Trexy system I wrote about earlier.

O'Reilly also quotes a post from Bill Burnham which notes that:

For start-ups, having core search become just another part of the Internet’s infrastructure is actually great news. This frees them from the huge capital costs required to build a competitive core search platform and instead lets them focus on building a great consumer/enterprise application.

So not only is building search applications the best strategic option at present from a competitive viewpoint, it actually is a better option from an infrastructure viewpoint too.

Burnham also notes that:

....some of the most interesting opportunities will come not from trying to improve the accuracy and context of a single query, but from looking at aggregate information about search indexes, results, and queries across time. In other words, the marriage of this search infrastructure, with persistent queries and advanced analytics will likely create an entirely new class of applications.

Quite - New Search needs to be integrated into the feedback system of the closed loop architecture of my life - know who I am, what I want, who is like me and wants the same. Oh, and it has to be simple, intuitive and available on my mobile 'phone too.

Not much to ask, is it