>Does anyone have a good reason why curation works better than search for mobile apps, but the opposite situation holds for web apps?
Web content is (relatively) static and indexable. It's essentially all "media".
Applications are not. Developing an algorithm that can deduce what an application does so it can be searched for is essentially a version of the halting problem, and even approximation pretty much requires "strong AI" to get even remotely close on all applications.
Thus, application directories must include a form for the developer to fill out to be searchable at all. Allowing the developers alone to control this means false information is too easy to spread, so curating steps in to weed out as much as possible.
As to ranking, yeah. A PageRank clone makes sense.
Web content is (relatively) static and indexable. It's essentially all "media".
Applications are not. Developing an algorithm that can deduce what an application does so it can be searched for is essentially a version of the halting problem, and even approximation pretty much requires "strong AI" to get even remotely close on all applications.
Thus, application directories must include a form for the developer to fill out to be searchable at all. Allowing the developers alone to control this means false information is too easy to spread, so curating steps in to weed out as much as possible.
As to ranking, yeah. A PageRank clone makes sense.