The "Less excursion is best" line as a general argument appears to hinge on 2 points;
that drivers are non linear through a substantial part of their operating range
that the distortion they produce is audible (in a way that has a negative impact on SQ)
Studies of the latter tend to point towards the hearing system being remarkably tolerant of distortion at low frequencies as far as I remember. I have no idea about the former, any data to share on this?
The other question is whether you are hearing (preferring) a change in spectral balance vs any changes in distortion.
Really, it boils down to the basic principle that when a driver isnt moving it creates no distortion, and when it is it does create distortion, with more movement equalling more distortion. Making the basic assumption that less distortion is better, then less excursion is then also automatically better.
Ive made the same arguments as yourself for a long time now, that as long as the threshold of distortion remains sufficiently low, then comparing one system to another is a bit of a moot point. Some people have actually been shown to prefer a certain amount of distortion over output that is too clean.
It might boil down to how particular you are. Ive found through experience that if you work to certain principles, i.e, assume nothing and simply work to reduce distortion as much as possible no matter what, then you wont go wrong, and I like to work that way myself. The figure of 10% distortion is obviously mentioned a lot if you study the field, but its also clear that some in the industry consider this too high (Ken Kriesel for example). There is then also a point I like to consider, and thats the point that I certainly feel there is a difference between obvious and non obvious distortion. Obvious distortion is, well, obvious, it makes you sit up and question what just happened. The less obvious distortion is what you begin to notice over time but cant pick up on as a distinct problem. Its like the difference between loud music, and music thats not loud, when they are actually both played out at the same spl. I wont try explain that Matt, as I know you understand what I mean by that comment. For me, being particular about your distortion in a way more like Ken is tends to help ensure the less obvious distortion issue never rears its ugly head. Obviously, we need to remember that there are several components to this part, and proper system set up and integration can be as big a factor in this as reducing driver distortion components to a minimum.
Ill have a look for some links, but John at AE made a nice post about xmax some 6 years ago now:
Acoustic Elegance • View topic - So What's the Deal with Xmax?
The bit in bold is not true, at least with the usual miniDSP model which provides you with the ability to create a bunch of biquad (IIR) filters. The distinguishing features being that they are computationally cheap and that they can only approximate a linear phase filter (and cannot be used for phase correction). To do that kind of work you need the opendrc model which opens up fir filtering to you (though latency becomes a real issue then).
Sorry Matt, I should have been more specific in general. The LTCs built into something like a DXD accounts for this. Ive not looked into the MiniDSP enough to see how that device deals with this, and wasnt meaning to imply that device would automatically solve these problems. It will however do a much better job than trying to simulate something like a LTC with a bunch of PEQ filters with say a BFD for example.