I had a thought. I feel like there is a flaw in here somewhere, but I'm going to put it out for critique.
Consider the Coke bottle analogy for DCI. The faster you open the bottle, the more likely it is to fizz. This is analogous to the faster you ascend, the more likely your body is to "fizz".
As we all know, the closer you are to the surface the more important it is to ascend slowly. In other words, if you ascend at 30 feet per minute from 100' to the surface, the ascent from 100 to 90 corresponds to opening the Coke very slowly. The ascent from 90 - 80 is opening the Coke a little more quickly. And so on. Ascending from 10 to the surface, still at 30 fpm, corresponds to opening the Coke much more quickly. The last 10' is always the most important part of the ascent.
Consider an ascent from depth where you are bumping up against a GF99 of 95 the whole way. As you go up (presuming a constant ascent rate), the rate of pressure change increases steadily. You go up 10' every 20 seconds, but the change in pressure every 20 seconds gets bigger and bigger.
The change in pressure from 100' to 90' is going from 4 ATA to (approx) 3.7 ATA. That is around an 8% drop in pressure. The change from 10' to the surface is going from ~1.3 to 1.0. That is a ~23% drop. So, 8% per 20 seconds versus 23% per 20 seconds.
My thought is that bumping up against GF95 during that last 10 feet is where the real danger is. Perhaps bumping up against 95 all the way up TO the last stop (of 10 or 20) is much less dangerous than it might seem on initial consideration. Then staying at that stop until your SurfGF is, for example, 60 really makes the last 10' very safe (statistically speaking).
In other words, maybe the way the folks that I have talked to are doing their dives is not at all as crazy as it might sound?
I think we're in danger of the simplification getting too simple. The coke bottle is a good metaphor for new divers, but as we all know the physiology is a lot more complex.
Yes, for the reasons you discuss, it's really important to use the ascent rate on which the models are based especially when shallow, but it is conflating concepts to say that "10 feet is where the real danger is" based on the fact that relative pressure change is greater with depth changes at that point. That's true but we're talking about a
stationary diver at 10', who by definition (in this case) has the leading compartment at 95% of the M value or less. Physiologically, I'm not sure why it would be any worse to bump against 95% at 10' versus 20' or 30' or 40'.
Your computer doesn't clear you to ascend to 10' until your GF ceiling is 10' or shallower. The only time you "bump" against your GF ceiling is when you first arrive a stop. That ceiling continues to lift, a little faster now what you've ascended and increased the pressure gradient for offgassing. When it lifts above the depth of your next stop increment, your computer will clear to go up again. For example, towards the end of your 30' stop, your ceiling will be 21'. When it's 20 or a bit less, the computer will let you ascend to 20'.
There may be some more complicated math and physiology about which compartments matter the most (are the least tolerant of supersaturation), etc., but to the extent that there is a difference between the risk associated with bumping against your GF limit at 10' versus some other depth, the story would be much more complicated than (and I believe unrelated to) the "rate of change" concept. For example, one good reason why it is more risk to surface at 95% of your m-value than to hang around on your stops at 95% of your m-value, is that you're now doing all sorts of things that could contribute to a hit, like climbing ladders and moving gear around.
Where the phenomenon you discuss is pertinent is when you're doing your deco on a fixed line in rough seas so your "depth" is changing as waves pass over or you're hanging on a deco bar and getting jerked up and down by the boat with the same effect. On the last stop, you might be a 5' one minute at 15' the next - and that's a problem.
Custom GFs allow you add conservatism by setting your personal GF ceiling at something less than the M-value. By allowing you to set the slope of the line, you can also bias deco towards deeper stops and have a profile that looked more like a bubble model. That was a thing for a while, but the NEDU study led many to question that practice. Some, like the folks you spoke to, seem inclined to throw the baby out with the bathwater.
As I think the researcher you spoke to seemed to imply, we have some good empirical data on "conventionally" sloped GF factors. Jumping to something like a 95% all the way to the last stop seems to me to be a layperson's "if I take twice the dosage, I'll get better twice as fast" approach to medicine.
