|
|
|
|
|
|
by nickreese
1814 days ago
|
|
|
The main problem with search today is the lack of organically appearing "curation" found on the early internet. Today everyone knows the value of a link, so it curation via links doesn't happen organically any more like it used to... yet this is still a major/the primary part of Google's Search results. Until the "curation problem" is solved I am not hopeful of a search engine producing excellent results like Google did before 2010. Stoked to try this out, but teaching a machine to curate is hard. |
|
|
An idea for curation:
Content evaluated by usefulness towards solving some problem. Perhaps measured by its usefulness towards training a very narrow AI. Like what if AlphaGo’s only inputs were web pages on the topic of Go, and there were hundreds of instances of it all trained on different combinations of web pages, and competing against each other.
Probably the data<>contest mapping is itself intractable until we have AGI, at which point the whole question might be moot. But what it?