| A major change is that good search algorithms or giant server farms could become sufficiently useless that a remote search engine makes no economic sense. I don't see that happening in the near term but there are sci-fi scenarios in which it might happen; I'll sketch them so the pattern becomes clear. EG: - user's local storage and local cpu / throughput goes through a quantum leap and most users most of the time have locally cached copies of any content they might find useful, and the cpu power to search it all efficiently - some quantum leap in bandwidth and other IT infrastructure means you can scour large chunks of the web directly in a short time period (eg: if everyone had terabit fiber-to-the-home and backend infrastructure were built out enough you could just download entire websites and full-text search them in the time it takes now to search @ google) - some quantum leap in "software agents" and safely-executing remote code comes to pass and users send out long-lived, virus-like agents that constantly scour-and-index-the-web-for-them (executing at the site of the content host This is basically a sci-fi "what-if" but as a thought exercise it gets the point across: right now it makes sense to have a centralized spider-and-archive-and-index-and-rank facility b/c the web is so huge and its pipes are so tiny compared to the # of end users and their individual computing capacities; a sufficiently radical shift in end-user capacities would make it moot. All probabilities for the above I'd put at .01% or something similarly lame. Weaker versions of the scenario include anything that reduces the use of the web and consequently reduces the need for search services: - next-generation rich-client driven chat + forums + social-networks become the dominant areas of internet activity (with the web relegated to low-grade entertainment, reference material, and shopping); users don't need external search engines for these activities and the service-operators' databanks are deliberately opaque to conventional spiders - despite all the naysaying the semantic web takes off in some fashion, with enough widely-used, useful-enough ontologies and so forth that user-run agents and user-run spam filters and user-run relevance+credibility assessors can auto-navigate the web to find what the user is looking for without passing through a conventional search engine ...etc., which I'd move up to .1% or so likelihood. |