Okay, thru a not terribly generous contribution from Colorado PERA, A6 valve research continues...
I've pulled mine, and even though I always have had the filter screen over it, it has still accumulated an interesting amount of very fine grained particulate. Examination has not been able to establish whether it managed to get between the plunger and the gasket, but it also cannot be excluded.
The particulate matter has hardened to a resistant mass. Submergence for an interval of less than 5 minutes in a solution of hot water and Dr. Bonner's soap loosend some, but not all, of the mass, and exposure to a flow of hot tap water succeeded in flushing it thru the valve. A residual amount of the mass has remained, and is being exposed to longer duration submergence.
The actual mechanics of the valve has been examined, and it seems to have [at least] two potential problems:
First is actual leakage caused by the plunger being elevated above the diaphram by the accumulated particulate.
Second is a weakening of the resistance to air pressure causing the valve to release early.
Paradoxically, the leakage problem caused by the particulate accumulation could work to counter the problem of early release, although either could contribute to the problem of a floppy floor. Data collection practices in early phases of this research failed to distinquish between the different problems, so primary cause cannot be established.
It is noted that during field research there was no presence of sustained bubbling or of a hissing sound.
There is a possible connection between the two identified problems areas. It is possible that some form of weak material bond exists between the plunger and diaphram. The bond could be intentionally introduced by addition of a bonding agent such as a glue, or be surface adhesion from two initially clean materials during the manufacturing process. A pressure triggered release of the plunger would break the bond from either source, and the introduction of particulates could further weaken the bond. How this would contribute to early release is not clear, although it could in theory make it harder for delayed release past 2# pressure to occur (assuming the secondary particulate bond was weaker than the primary created during construction).
Possible directions in future research may consist of more experimentation with the addition of a new, unused valve as an experimental control, or just washing the existing valve with 303 and seeing if it fixes it.