In one

A wild, but interesting response to Joel Splosky. generated this interesting response:

Interesting comments. I think the two of you are looking at things from two different perspectives. Joel seems to be approaching things from the shrink-wrap/ISV developers perspective where the combination of the size of the runtime, as well as certain immaturities exist in the platform (that can and does result in broken code) only serve to cause headaches and support work for a shrinkwrap vendor. If you approach this from a software services provider perspective, the size of the runtime doesn't matter and breakage is minimized by comporate controls and additional billable hours.  [via Seven Reasons Why the API War is Not Lost After All]

Bang on and in one.

There are also a couple of web app usability problems comments...

The main point I think Joel was making is that webapps will continue to become more and more prevalent, as opposed to smart-client apps, because users don't mind....I don't agree with this - webapps are loved by System admins, but users often just have them forced upon them and really dislike them. Just today, a Web project (ASP.NET) in my team in a very large corporation was rejected by users for being unusable. And this wasn't a fault with the app, which was as specified - it was the lack of those 'smart-client' features that the users are used to in Office, etc.

Personally I agree (or I'm glad there are some users out there who agree), I've always thought it problematic to do apps inside a browser, but I've not thought it problematic to do apps with DHTML technology. IMHO, XAML exists because the browser can't be developed without walking into a wall of security concerns and screams of non-standard etc etc. So, write a new browser using its own markup language. All markup can be implemented in a runtime that instantiates objects - its really a very very short distance to XAML.