i replaced the noisiest whispers with 50k stepped alps. the 33ks came out at that time.
If you wired the 50k Alps the same as the stock FP III attenuator then you still have a dead short at the input when the volume is at zero. I.E. your source is shorted, again a
bad thing! Not knowing the configuration of your attenuators I can't comment on what happens one step, or more above "zero volume."
So between the grid and cathode of say a 6cg7 is a short?
Between the grid and cathode doesn't apply here as it has nothing to do with the input impedance. The Grid in a stock FP III is 15.22k to ground, the cathode is an LED forward bias to ground.