3 ms·

We're surely working all the time to make search "more sophisticated"; many special case queries are already smart (from "2+2" to geo, stocks etc., you don't ne

by opinali 15y ago

We're surely working all the time to make search "more sophisticated"; many special case queries are already smart (from "2+2" to geo, stocks etc., you don't need to go to special sites like calculator, maps and finance). And we surely have plans to go way beyond, but generally, this is Hard Stuff(TM). For things like journal articles, the information is often behind paywalls like ACM, and even when it's not, specialized engins like citeseer are hard to beat because the info has very special organization needs like collecting and measuring citations. On your most advanced requirements, I think only an Asimov's positronic robot would be that smart ;) unless there's a specific effort to curate this data... which requires tons of human labor, so ads served to the very small amount of people who needs this service will not make it viable. It's the same problem we have with patents (see http://arstechnica.com/tech-policy/news/2012/03/opinion-the-problem-with-software-patents-they-dont-scale.ars http://arstechnica.com/tech-policy/news/2012/03/opinion-the-...).

waterlesscloud 15y ago
Listing all the reasons it's hard is why the area is ripe for someone else to do it.
- opinali 15y ago
  No; these reasons show why the are is ripe for anyone to do it. The only fallacy in Paul Graham's comment (or in possible interpretations of it) is that Google has a weak spot there so it's an opportunity for somebody else to beat Google. Trust me, we have a ton of resources dedicated to improvements in search and we have lots of cool things coming down the pipe, although maybe not in the velocity that one could dream (e.g. something like intelligent research for scientific papers is firmly in the sci-fi realm today, at least for fully-automated computing). BTW, Paul's article has a big #fail when he mentions code search as a possible idea; dude, we do have that and it's amazing (but unfortunately we recently shut it down; not sure if this will eventually resurface as part of some other product). All that said, of course some company can always make an effort dedicated to a specialized niche that we are overlooking and beat Google Search in that niche; an excellent example of that is Wolfram Alpha. Still, not a big deal; to really "beat Google" you need a new general-purpose, full-Web search engine that beats Google's. Not impossible either, but the barrier to entry is simply colossal and it amazes me that people don't realize that and dream that it just takes some cool new idea or clever new algorithm to do that--we are not anymore in 1998, when Google, still working off a garage, started to beat the current top engines like Yahoo! and AltaVista.

by opinali 15y ago

waterlesscloud 15y ago

Listing all the reasons it's hard is why the area is ripe for someone else to do it.

opinali 15y ago
No; these reasons show why the are is ripe for anyone to do it. The only fallacy in Paul Graham's comment (or in possible interpretations of it) is that Google has a weak spot there so it's an opportunity for somebody else to beat Google. Trust me, we have a ton of resources dedicated to improvements in search and we have lots of cool things coming down the pipe, although maybe not in the velocity that one could dream (e.g. something like intelligent research for scientific papers is firmly in the sci-fi realm today, at least for fully-automated computing). BTW, Paul's article has a big #fail when he mentions code search as a possible idea; dude, we do have that and it's amazing (but unfortunately we recently shut it down; not sure if this will eventually resurface as part of some other product). All that said, of course some company can always make an effort dedicated to a specialized niche that we are overlooking and beat Google Search in that niche; an excellent example of that is Wolfram Alpha. Still, not a big deal; to really "beat Google" you need a new general-purpose, full-Web search engine that beats Google's. Not impossible either, but the barrier to entry is simply colossal and it amazes me that people don't realize that and dream that it just takes some cool new idea or clever new algorithm to do that--we are not anymore in 1998, when Google, still working off a garage, started to beat the current top engines like Yahoo! and AltaVista.