There was a panel discussion with John Bristowe, Scott Hanselman, Joe Homnick, Joe Lindsay, Terry Mohn, Ted Neward.  the highlight for me was that there still isn't good tool support, or a good story from Microsoft, on how to manage services once they have been deployed.  Other points:

• Scott mentioned that his company provide services to 30% of all online US Banks and 50% of them rely on web services. So they are out there and being used.
• An excellent point was made there isn't good tool support (or a good story from Microsoft) on how to manage services once they are deployed. The tools to ensure that services are up and running and providing guaranteed levels of services are not here yet.
• There was some discussion about whether we should care about angle brackets or not. Scott made the point that we should focus on Infoset. John Bristowe made the point that it can be useful to understand the specs and know what is happening on the wire.
• Ted mentioned that editing WSDL is too hard and no-one at the event other than Scott had actually done it. Scott made the point that most of us had edited HTML files because we did not like the way FrontPage formatted them, so why hadn't anyone done the same with their WSDL?
• Some good points from the more business-focussed members of the panel that they are more interested in tools making developers productive than in having developers that understand the plumbing in web services.
• Discussion about what the definition of SOA is. Answers included 'components with a longer wire', 'objects with explicit boundaries', 'hooking shit together'.  I had a useful discussion aftewards with Joe Homnick that helped me see that SOA is most useful as a concept when talking with businesses about how to architect projects.  As I've mentioned before I'm still not convinced that SOA is definable, but it is based on top of good architectural concepts such as encapsulation, data hiding, loose-coupling, service discovery, messaging patterns and asynchronous processing.  So if we need to call this SOA in order to get all excited about these topics then it may be a necessary evil.
• I followed with the question 'what problems is SOA trying to solve?' answers included high-level re-use and interoperability.
• Discussion about location transparency not being a good idea. Ted suggested the original idea was really about being able to write code without having to know the location of the components, rather than hiding the fact that this was an expensive cross-network call to a component.
• Discussion about the fact that Amazon uses Web Services well but that XML over HTTP is more popular than their SOAP implementation. There was some discussion about alternatives to SOAP such as REST, but they were seen as more limited compared to SOAP. The key for me is that tool vendors need to implement great tools for SOAP so that it is easy to use as these competing approaches.
• The more business focussed members of the panel made the point that the technology is not the main thing, it's supporting business functions. For them the point of the Interoperability work is that it les them go out and buy products that they know can talk together.