Simon,
> 1) Gear to rock interface is probably always the weak link in a belay
Yes, as modern gear is so strong.
> 2) Shock loading (dam I used the forbidden word) increases the risk of that failure
Well yes, but as been pointed out the increase in force is slight. This makes the real question: which is the best way to rig an anchor: risk a very small increase in force on a remaining arm, or try to equalise the force so an arm doesn't fail by using a sliding system? I guess the answer is sometimes the former, sometimes the latter.
> 3) Dynamic rope WITHIN the belay system will have a marginal affect on reducing the amount of sudden loading on the gear (not talking about rope from belay device to climber)
Is this true? I know the length of rope is small, but if might add to several meters, and importantly if contains knots. (Anyone remember some test data that was doing the rounds on the energy absorbed in knots in sports falls?)
> 4) The more marginal the gear the more this becomes a problem, ie sudden load onto a 4ft girth oak tree....who cares...onto an RP1!!!!!!!
Exactly, so the question is do you try and equalise a bunch of rp's with a sliding system to try and share the forces (at the risk of extension) or use a "static" i.e. non-moving system such as the rope or a big sling with an overhand?
One question that is not addressed enough is that when using systems with fixed arms (e.g. the rope) there is a tendency to set the thing in an arrangement for supporting the leader as she hangs at the belay. The real test will either come from the second failing off or the leader falling on the next pitch. The direction of pull from the second will change as gear is removed from the route, especially the last piece. The direction of pull when the leader falls will be onto the belay if they fall before placing the first piece, in the direction of the first piece, or possibly the direction of the second piece later if she is using two ropes. This means knowing the place she will place the first bit of gear in both ropes when you set the belay. No chance.
The direction of pull changes so much between these situations that the failure of any one arm will for some situations lead to a "shock" load on the remaining. (Example, three cams in a triangle, leader falls, lowest cam rips (because it is the one that takes the force, and was set nice and low so it could), the belayer flies up and "shock loads" the upper cams.) This shows that using the rope (or other system) does not solve the problem of extension. Belay failure is (as was pointed out by another poster) mainly an issue when the leader falls off, not when the second does.
The way I think about it is to reflect on the difference between extension that might do something positive (e.g. sliding-x's on bunches of rp's) and extension that does nothing but (slightly) increase the force on the rest of the belay when an arm fails, or usually more important, swings the belayer into something (or over the edge) and makes them let go of the rope!
Sorry for the length of that. I hope it seems at last half clear.
PS. The book: the contented baby, worked wonders for us. But I know the book is considered more controversial than chipping gritstone.