How probable is the success of such a discussion, do you think?  We
cannot even agree on specific options, and here you suggest to take
this disagreement to a much higher and abstract level.

That's a good point. If "success" means "agreement on a single guideline", I think it is 0.
But I'd say that "agreement that everybody wants A, B or C as main guidelines" (and may
still disagree on how to achieve them) would still be progress.

I think that the "higher level discussion" could also be less emotional and less multi-directional, 
by not looking at specific options. I was surprised to see in the other discussion some messages 
arguing that ibuffer is too confusing compared to list-buffers, and others arguing that undo/redo 
is inferior to the undo-ring model to the point of being unusable.

If anything, thinking about such guidelines should give indications as to what the "themes" 
should be. I imagine that everybody has an internal model of what a barebones emacs should 
be (e.g. actually barebones, or simplified settings, or superior paradigms in full...), and while 
there are certainly thousands of combinations of options to realize those models, there can't be 
that many models. 

In that sense, agreeing on a subset of goals might be an easier task than agreeing on a set 
of options. 

But that might indeed be wishful thinking, and I didn't realize at the time of posting that it makes 
the discussion essentially a duplicate of the one on themes. Sorry about that.

Thibaut