> I do believe that the document paradigm is the best one for 
> the web for this reason and a number of other ones and I 
> think that fat client applications should remain a niche in 
> the web rather than the other way round. 

I certainly think the paradigm of sending a document is a nice and simple
one, and highly extensible to allow the document to contain various kinds of
active content in the cases where that's needed. (I also think it would have
been nice if it were more symmetric - sending documents in both directions.)

I do have some sympathy with the notion - paraphrasing Norm - that if you're
going to send a document to the browser, every string of Unicode characters
should be a valid document with a defined interpretation. I don't think
that's loosening the standards, it's tightening them - in the past with HTML
a subset of documents had a defined behaviour and the rest were displayed
however the browser chose. It doesn't stop authors who want to use a
stricter syntax with potential for error checking and validation from doing
that, but it does mean that the amateurs who churn out junk are going to
have their junk displayed the same way by every browser, which seems a step
forward from having every browser display it differently.

Michael Kay
http://www.saxonica.com/