about 8 years on 2009-07-12

(UNFINISHED DRAFT) RIP SEARCH 1999-2009, WE HARDLY KNEW YE


note:  i am posting this as a draft because, quite frankly, i don't have the time to refine it and won't for a while...

there have been several quite insightful posts analyzing the diffusion of information around michael jackson's death.  the point has been made a few times that the 'social' web -- twitter, digg, facebook, etc picked up on and diffused information about his death far faster than google et al. were able to react.  in fact, google news falsely sensed an attack sensed an attack shut down and for the second time i can recently remember.

people have been hinting, let me come right out and say it - open ad supported search is dead*.  it is as over as yahoo directories were in 1998.  it was a 10 year accidental blip on a continuum, and here are just a few reasons why:

1.  search ranking is a war of attrition - access to arms is flattening with cloud computing, and they are way more of them then there are of you.

2.  the types of questions that search answers are commodetized - (bing has this right).  the value of information is a function of probability, and google's results are quickly becoming highly probable.  there is no value in knowing the answer when everyone else also knows the answer.

3.  biology tends to get things right.  a networks of peers, not a central dbs, dominate bio.

4.  push will always be faster than pull.  search is a pull/crawl metaphor, where social is push.

5.  'the commons' always get messed up.

the good news - despite some crazy recent advances in quantum computing - point one humanity.

the ?bad? news - the world is not getting flat - the map is just no longer geographic.  we will look back at a brief period where access to information appeared to be flattening, and watch social flip it back in the opposite direction.

a final teaser - search is democratic, social is not... so here comes the most fundamental divide of the haves/have-nots ever.  

*deep search, tech to cull complicated datasets, etc.  that is only just beginning --- and if marooned in realtime tells us anything, accumulating and efficiently scanning databases takes the cake.


original swl blogposts and letters 2007-2010