I looked at this minus-million thing again yesterday when my brother brought it up. I'd seen it before, but we were talking about search engine crappiness and alternatives.
Since I remembered Blekko (met Greg, the founder guy, in SF once) I pulled it up and we searched for yunnan, a Provice in China, as a test. Most of the results were for commercial tour operators or thinly veiled redirects for such.
The million-missing one on the other hand turned up more interesting or 'bespoke' content.
I am ignorant of such matters but would have thought a spamassassin-style Bayesian model based on sentiment analysis, advertising frequency, update frequency, hosting location, content originality or any similar clump of readily obtainable metrics would be enough to usefully cull the vast majority of the useless modern stuff.
I mean, if I want commercial tour operators, I'll tell the search engine by typing something like "prices" or "companies" or "costs" or whatnot.
The other realization we had, doing this on an iPad, was that search engine interfaces positively suck. They're still stuck in the 90s. With a touch-based interface, there should be a more interactive model for query refinement than text editing. Like, uncheck [x] commercial sites.
Since I remembered Blekko (met Greg, the founder guy, in SF once) I pulled it up and we searched for yunnan, a Provice in China, as a test. Most of the results were for commercial tour operators or thinly veiled redirects for such.
The million-missing one on the other hand turned up more interesting or 'bespoke' content.
I am ignorant of such matters but would have thought a spamassassin-style Bayesian model based on sentiment analysis, advertising frequency, update frequency, hosting location, content originality or any similar clump of readily obtainable metrics would be enough to usefully cull the vast majority of the useless modern stuff.
I mean, if I want commercial tour operators, I'll tell the search engine by typing something like "prices" or "companies" or "costs" or whatnot.
The other realization we had, doing this on an iPad, was that search engine interfaces positively suck. They're still stuck in the 90s. With a touch-based interface, there should be a more interactive model for query refinement than text editing. Like, uncheck [x] commercial sites.