Scott Ambler has been a prominent critic of the OMG's Model-Driven Architecture, but has started to examine some scenarios for "Agile MDA" here.
I think many of Scott's insights are useful, and admire his sense of history - it's rare to see people of our generation understand the history of CASE. One thing he doesn't make the connection on - the OMG's MDA can trace its lineage directly back to the failure of 1980s CASE tools, via interim standards such as CDIF, AD/Cycle, PCTE, and IRDS.
