Joris, and all,
I do think that numerable emerging projects (like Kevin Kiernan's 'Edition Production Tool' (EPT) and Sebastian Rahtz' Open Office TEI filter) are doing an increasingly good job at both abstracting away form the TEI-particulars and hiding the major amount of time and work involved in applying XML by hand. Which is what good software tools should be all about: abstracting from technical particulars and supporting any process in such a way the it becomes faster, more efficient and easier to use. Thus providing the tools that answer to the scholars intuition when handling and structuring text, seems a good way forward to me. That mark up recedes into the background due to such abstracting developments is a good thing. Just like you don't need to know the XML particulars of an OTD-XML file to write a perfectly comprehensible letter in Open Office, you shouldn't need to know the TEI-XML particulars when structuring a text for literary research purposes.
The difference between writing a letter in OpenOffice and structuring a text for research is that the former is not necessarily an intellectual effort. The intellectual effort has been made by those that studied people's letter-writing habits and needs and designed the Open Office XML format.
You're right of course in advocating the use of tools that wherever possible hide the complexity of markup. But what these tools can hide is only the admittedly not very pleasant sight of lots of brackets and ampersands. All of the other complexities (e.g. what in TEI terms would be called add's and del's and unclear's and gap's and app's and rdg's and all of these nested and/or overlapping...) are there because digital editions do things that are complex. You need to understand them one way or the other.
The research value is in the act of structuring, not in the particular tagset used and not even in the mark up model applied.
This may be true if you are analysing a text for your own research purposes. But we should also try to create enduring digital resources. For durability, ease of maintenance and mutual intelligibility it is important to try to stick to standards (and therefore, to learn them).
Peter