Preamp Output Capacitor: Mundorf Supreme vs. Supreme Silver Oil

AG, not necessarily.  A 0.68uf cap into a 100K loaf will have a rolloff that is -3 dB at 23 Hz.  That is not going to sound “thin” on most systems.  The 2uf cap extends the -3 dB freq to 8 Hz.  There would be virtually no sonic difference between these two caps provided they are otherwise identical (same brand, same model) at least on most systems.

You are right, there are a few criteria.
1. I am not talking "most" systems. I am discussing excellent systems. Many average Joes probably won’t care that much though. But it is important to others.

2. I am not considering bumping the woofer’s Qts above 0,707 in order to artificially boost the system’s low end response. The problem with that is a phase deviation, bloating, or even synthetic bass (RCA Radiotron Designers Handbook, 4th edition, 26+ engineers). And can one ever really obtain flat response with such a measure?

3. Another problem is that dbs add. By that I mean each components deviation is additive. If we have three components, we add the db variation of each. For instance, -0,2db plus -0,15db plus -0,2db gives us -.55db from flat. A deviation varies vs frequency, and exists over many octaves. (RCA Radiotron Designers Handbook, 4th edition, 26+ engineers explains this plus the next paragraph.)

Another concern is if we limit the bass response, we also need to limit the high frequency response. Visa versa as well. (RCA Radiotron Designers Handbook,)
We need to keep each component’s frequency response as flat as possible, minimal deviation to protect the musical information.

4. I am discussing capacitors in absolute sonic terms (not measurements), so as to not cover over another fault in a poor/average design. The capacitors I am talking about are not a 716P orange drops.

5. Every time one compromises sonic quality in a part, a stage, in a component, the synergy will not be as optimum as it could be. Design and parts quality is very important.
By the way, imo, -3db at 23 hz is extremely poor frequency response unless one is purposely constructing a high pass filter to limit the bass response, which is useful for certain purposes. Cheers
ag

The sound character of the particular output capacitor will have a much greater effect than the different measured cutoff frequency with a 100K load (23 vs 8 Hz).  
For example, a Mundorf Silver/Gold/Oil sounds “thin” regardless of cutoff frequency; an older InfiniCap would sound even more thin and lightweight.  On the other hand, a .68uf Copper V-Cap with a 100K load will have deep, powerful, full-bodied bass despite the measured response being down 3db at 23 Hz.

It is mistake.
For R=100K and C=0.68uF F3 is 2.34 Hz

You're right!  That should teach me to do math in my head.  I dropped a digit.  

Of course that makes the issue of the value of the cap even less significant sonically.  With tube gear where the input impedance is typically 100K or more, a 0.68uf coupling cap is fine.