Jim: I'll try not to be too long winded!!!  However ... we have done extensve
testing regarding the effects of neg vs pos etchback on both rigid and flex
multilayers.  Bottom line ... pos etchback is ALWAYS preferred. Neg allows
separation at the PTH barrel. Excessive etchback (pos) can result in plating
folds which is not desireable.  The 55110 guidelines for pos etchback are pretty
good. 

With multi-layer rigid flex there is another problem.  The use of acrylics in
the flex sections, when extended into the rigid sections (standard flex
multi-layer), can result in excessive Z axis expansion during thermal cycling
(acrylic CTE is higher than the other materials and it is hydroscopic).  This
results in barrel cracking and / or separation.  Pos etchback will help ... but
the solution is to eliminate or reduce acrylic (10% max). A number of PCB fab
houses have now made this their standard product.

As far as reliability ... we have run Rel. Enhancement Tests and traced failures
to butt joints (no etchback) or single point contacts (minimal pos. etchback).
Negative etchback was a DISASTER! (Note: etchback is only one of a number of
contributing factors to PTH failure. Another one is the method / occasion of
rework.  The thermal stress to the PTH can be excessive if too many rework
cycles with too much heat in a localized area is applied. Another is the type of
copper, HTE is preferred.)

Our experience suggests that up to 8 layers is relatively safe if designed
properly and fab'd with HTE Cu and pos etchback.  As the layers increase above 8
all factors become more and more critical.  

I've really only scratched the surface ... we have lots more info if you need
it.  I hope I've been of help.

Vince Giardini
Parker Hannifin, GESD
(516)231-3737