From Simon St. Laurent:

The W3C has already poisoned markup substantially with W3C XML Schema's various notions of type.  While XQuery has taken a few steps to separate itself from the PSVI, its own notions of type substantially complicate XQuery and have infected XSLT 2.0 and XPath 2.0.

The PSVI is just one especially nasty aspect of the problems created by retrofitting notions of type which come from object-oriented and relational systems (conflicting notions, at that) onto markup.  I expect the inconsistencies and complications of this decision to take a heavier toll on XML and the W3C than namespace and URI issues have taken in the past.

The W3C needs to think seriously about what it wants from markup.  I think  the W3C has lost sight of the promise XML originally had for the Web, as a lightweight format for loosely-coupling information producers and consumers.  Instead, recent efforts seem to be aimed at recreating CORBA or DCOM, a style of technology I thought the Web had pretty much shown up over the last decade.

These seem like fundamentally architectural issues to me, needing some serious discussion with a heavy dose of courage.