There is no science to the idea that source impedance has anything to do with loudspeaker control. It doesn't. The voice coil resistance totally dominates the damping term. For damping to halve the source impedance would need to equal the voice coil. A non zero output impedance means the damping component of the speaker system changes from one defined by Zvc, to one defined by Zvc + Zs. So for a typical driver, from about 6Ω to 6.08Ω. A DF of 1000 would mean a change from 6Ω to 6.008Ω. The difference in damping here between a DF of 100 and one of 1000 is thus 1.2%. Given the voice coil resistance is changing dynamically by vastly more than this in response to the signal, we can be pretty sure that this is simply not an issue. The crossover adds typically 0.5Ω to 1.0Ω as well. This is in series with the damping resistance, so the effect is even smaller with passive crossovers. Why the existing series reistance inside a speaker is continually ignored in arguments about the nature of "control" of speaker movement is one of the great mysteries. It ignores fundamental understanding of circuit theory and the basics of dynamic loudspeaker physics. The well known analysis by Dick Pierce goes over this in careful detail. The physics has been well understood for many decades and is exemplified in the Theile Small analysis of loudspeaker operation.
When it comes to choosing speaker cables, I wonder if there is some misunderstanding derived from this viewpoint from
Benchmark.
Note that even Benchmark agree, control of the speaker has nothing to do with damping factor.
This missive from Benchmark requires some careful reading, as it provides a very specific and limited set of constraints. It also does
not provide a guide for how speakers should be evaluated. Nor does it actually provide any justification as to why you should purchase an amplifier with a very low output impedance, although they hint that perhaps you should be buying one of their very nice products.
Basically Benchmark are pointing out that speaker impedance variations coupled with a non-zero feed impedance can cause small variations in the speaker's frequency response. I don't think anyone ever suggested otherwise. What Benchmark then do is provide a very stringent requirement - no more than 0.1dB variation across the response range - and then calculate the maximum permissible feed impedance needed to meet this requirement for a given speaker. Thus the combination of speaker cable and amplifier impedances. It should be noted that their requirement is sufficiently stringent that even with speaker cables of zero length it cannot be met by the vast majority of amplifiers on the market for speakers that exhibit a messy load.
Does this matter for testing and auditioning loudspeakers? Mostly no. Their justification for the requirement is directed at auditioning
amplifiers in controlled ABX settings when connected to very difficult speakers. Amplifiers, not speakers. This is important.
If one is designing, testing or auditioning speakers it makes sense to do so with an amplifier of the specification the speaker is likely to be used with. If the majority of amplifiers have an output impedance of say 0.08Ω, design to that, along with an allowance for a reasonable speaker cable. If the drive impedance varies dramatically from this and the speaker has some nasty impedance variations, the speaker will change its frequency response very slightly. Maybe +/- 0.2 dB. Note, this variation can be in either direction, a lower source impedance (aka higher damping factor, fatter speaker cables) will also cause variation. Very high DF is not a panacea that guarantees perfection. It simply provides the limit of change in one direction.
Benchmark are concerned about different
amplifiers sounding different when connected to the same speakers when the only difference the amplifiers present is a different output impedance. If you are auditioning the amplifier, especially if you are conducting well controlled ABX tests between amplifiers, you could get a misleading result, depending upon what you are testing for, if input feed impedance into the speakers is not controlled. The converse is not true. Auditioning different speakers is not made more accurate by a very low source impedance. You should audition with the source impedance they will actually be used with, or more importantly, were designed for.
What does this mean for testing speakers?. The speaker's impedance is currently measured and presented. That gives you a starting point on the frequency response variations can expect from changes in source impedance. Unless your own amplifier and cables have exactly the same impedance as the test system's you can expect a well understood and definable slight variation in frequency response. However given that most people have modern amplifiers that don't have wildly different output impedances, mostly sitting around 100, there is good reason to expect that the results, in practice, will be very close.
For auditioning by ear, an argument can be made that extremes of feed impedance should be avoided. Auditioning might be best done with a source impedance that best reflects the expected real life use case. Not everyone has a Benchmark amplifier. This is especially true, as the designer of the speaker will almost certainly not have designed or tested the speaker assuming a very low feed impedance. They will (at least should) assume the most likely case. All computer based speaker design software systems I have seen include a term for source impedance. Only the very naive set this to zero. This is key.
Speaker designers assume a reasonable source impedance from the outset. They do not assume an extraordinarily low one.
In the end, it is possible that small changes in output impedance of an amplifier can be used to subtly vary the frequency response of a speaker. Very subtle. Given the variations is response intrinsic to loudspeakers in a real life setting these variations are vanishingly small. A very low source impedance does not guarantee the best possible accuracy of frequency response. It provides the limiting case of one side of the possible variations. The best (and designed) frequency response may well be found at a different impedance. Typically the designer of the speaker will have chosen such a source impedance.