I spent some more time geeking out on the data last night. Here are my conclusions:
1. All of this is equalization/extension theory is primarily relevant in only three situations: A factor-two fall directly onto a belay, catching a second who somehow takes a relatively hard fall onto the belay, and when building a two or more piece protection point (this happens a lot on sketchy trad climbs and also on ice). These are situations where it would be real nice if the anchor equalized well under load, and then didn't shock-load the other pieces if one failed.
2. Unfortunately, it's about impossible to get any sort of realistic equalization out of a multi-piece anchor (with the gear we commonly use as climbers). If you go to page 29 of the study there's a low-friction equalization situation (equalette) shown there. Even in this perfect situation the pre-drop load totals per piece are obviously different (and could be improved by adding another sliding X etc.,), but even looking at the total "per leg" it's clear it isn't anwhere near perfectly equalized. The rest of the sliding X stuff etc. are worse (with the exception of page 28, but the extension problem is horrendous). A big smooth anodized aluminium 'biner might improve things a little, but even with knots etc. the real problem is what happens when one piece fails and the anchor extends. Yes, you cold tie limiting knots etc., but it looks to me like any extension is violent, which makes sense if you think about it (relatively static webbing or limited cord on a sliding X--bang).
3. In this study, and this is only one study, extension in the anchor is more problematic than poor equalization in terms of the max forces generated on the anchor. That's a real big departure from previous studies I've read.
All of this may and likely will change with the higher forces involved in a factor two or other high-impact situation with a lighter belayer and a larger fall force...
My basic idea that one piece in the belay must be capable of handling very high forces hasn't changed. I want one absolutely for sure bomber nut, cam, screw, whatever. Two absolutely bomber pieces are better, hell throw a third one in for grins. Two or even three or ten "maybe" quality pieces just aren't good enough. If I'm "equalizing" a stubby, an icicle and a shit pin for a piece before punching it up a difficult bit of alpine terrain I'm going to assume that the entire piece is only as good as the strongest individual piece.
I remember a helicopter pilot explaining the term "Jesus nut" to me. He didn't mean a super-religious person, he meant the nut that held his main rotor on. If that broke the only thing left to do was pray to Jesus. In a belay I want one super-solid "Jesus nut" that will hopefully hold any impact I can foresee and then some. And, because I'm not into the whole one-god thing too much, I'll put in another Jesus nut... And still try to limit extension to some point, and even roughly equalize it all.
And this may all change again once JimE gets some more research done, or I see another study done differently. I doubt that the basic concept of having one "for sure!" piece and preferrably two is going to change. And if I can't get that level of security then I'm gambling with two lives.