Between the network I use going down and Blogger going down, posting has become an exercise in masochism. That said, I wanted to point out Shirky's take on folksonomies, ie, massively distributed metatagging. I think this is the first interesting comment from the social software crowd in a long time.
I think it's clear at this point that massively distributed tags will replace professionally generated tags. It's not just that economics are superior (ie, it's relatively cheaper at the aggregate level, because the marginal cost to a surfer of adding a tag is minimal) - it's that the value prop is better aligned with consumer preferences. That is, the market - the Net, as it were - wants breadth, not depth. We want many things tagged shallowly, rather than a few things tagged with incredible precision and richness.
The support for this hypothesis is straightforward - the time people spend at Metafilter, checking out other people's photos and links on Flickr and del.icio.us, etc. You get the picture.
I don't buy Shirky's kayak metaphor - that no one can 'control' tagging at the aggregate level. I think it's always been that way - unless you're MS, it's difficult to control anything that's massively distributed. In fact, to argue this misses the whole point of using economics to analyse things. The price mechanism isn't a matter of control - it's inherently a massively distributed mechanism to coordinate expectations and outcomes.
Put another way, the gains to massively distributed models are
in the fact that 'control' is a nonissue - that coordination is cheap, and so the each agent can decide for him or herself what information to the aggregate. That doesn't mean that massively distributed metatagging is any more out of control than a farmers' market for tomatoes is - the same mechanisms are ultimately at work in both places.