"The point here, is that the flickr team did not wake up one morning and think: “You know, if we captured THIS kind of data, we could create this mashup; so let’s create an application.” Instead, they re-used data they were already capturing, and brought out something very interesting indeed. By creating tools which match their data (and could be used with other data of the same kinds), flickr is able to expose layers of value from the rich-pickings of their own data-cloud. The good stuff is where the data are." Yes, it is.

"The technology will probably improve, but in lieu of the promised emergent web AI, we need to build more small tools, more games to bootstrap datasets, and more simple ways of encouraging people to play their part in the semantic web without ever having to explain what it is." tt++.