On commoditisation, gear indicators, and web search - part 3

In part 1 and part 2, I'd written about commoditisation and some responses from web search engines. The responses included new features built on the top of their core engines, and personalisation. The next logical step would be using social networks. Myweb 2.0 seeks to do that. Can totally unsupervised machines come up with neat clusters like these in flickR? May be, but why not use the humans where possible? That's why del.icio.us, answers, etc., are being given additional attention even while there are people like Andrei Broder working on "searching without a box".

So, can these companies come up with all possible applications that can be built atop web search by themselves? If not, who'll do that and what will these companies do then? Web 2.0 is not just about user-generated content, but also about small enterprises and individuals building cool applications using web as a platform. So, if you want to build your own search engine, go build it! Just use the APIs from Amazon, Yahoo!, Google or numerous others. If you're an AJAX dummy, you can even get ready-made UI components. But, we all know that there's no free lunch. So, what would the platform guys do if they were to let anyone use their platform for commercial purposes? Yes, they're slowly evolving a business model around APIs. So, anybody listening?