3 ms·
Hidden inside the vaguely markov-chain/chinglish-like nonsense is an interesting idea: deduplicate the reference-information web. Even before the recent explos
by gojomo 17y ago
Hidden inside the vaguely markov-chain/chinglish-like nonsense is an interesting idea: deduplicate the reference-information web.
Even before the recent explosion of spablum (spam/pablum/made-for-adsense) sites, using the web to learn about something involved reading a lot of duplicate info. When learning about topic X, you might visit site #1 and learn A, B, C, D, E. Then, site #2 has B, C, D, F, G. Then, site #3 has A, C, E, G, H.
Even if H is the key thing you personally wanted to know, by the time you got to it, you'd encountered most other things 2 or more times, and even the 'gem' of a result, site #3, was 80% redundant.
A wonderful summary page would have A, B, C, D, E, F, G, and H, all in one place, properly contextualized. The best Wikipedia articles serve this role and that's why they're deservedly atop many search results. But the obvious utility of summary pages has prompted a race to create more of them, by automated and editorial means. When such attempts fall even a little bit short, they make the problem worse not better -- polluting the web with more permutations of the exact same information, hiding the rare gems in more muck.
I'd like to start a project to attack this problem head-on, and would love to hear from others who might want to collaborate.