If vibrations are not eliminated at source, there are two things you can do. First, you could isolate (decouple) the vibration. Second, you could transfer the vibration to another object (couple) and dissipate it in a non-audible form of energy. Rather than think of these as two "schools of thought" which are in opposition, think of them as two ways to achieve the same result. They can be used together. It is not necessarily one or the other, although one may be better in a particular application.
Using spikes or cones under speakers or electronic components isolates vibration. Because there is less surface area between the vibrating object and what it is vibrating into, there is less energy transfer. To a mechanical engineer, the concept is "mechanical impedence". It is a differennt concept than electrical impedence to an electrical engineer. What happens to the energy which is not dissipated out of the component? Well, that's part of the job of the component's designer. For a speaker, the energy might be channelled and directed out of a speaker port. Or it might be damped by stuffing in the speaker. For an electronic component, the chassis might be damped with a compound to minimize metal chassis "ringing". Just as the cones or spikes keep energy from the component from going out, the impedence mismatch also prevents vibrations from the environment from coming in (excluding air borne vibration).
No spike or cone is perfect. Some energy will get through. Now you can use the second technique. Take the energy which vacates the component and convert it to a non-audible form (eg. heat, or shift it to a different frequency). The laws of thermodynamics state that energy cannot be created or destroyed; it is only converted into something else.
One thing you can do is to couple the component to something massive. A small vibration in a large mass will be more easily damped. If you put an amp on a massive stand, (such as metal filled with sand, or on a granite slab), the vibration can be dissipated in the stand. Because the vibration is relatively small, it is less likely to vibrate the stand and it will likely be converted into heat and dissipate. Coupling a speaker to the floor or wall allows the speaker to be coupled to something massive.
Unfortunately, mass is not the only variable. All matter will resonate at some frequencies. If the transferred energy causes the mass to resonate, the resonance could be audible, or come back into the component. Here is the problem with wood. It resonates. That's why violins and acoustic instruments are made of wood.
If you can couple to a concrete floor, you can get damping from mass without audible resonances. If you couple to wood, you get less mass, therefore less damping, plus you are likely to get audible resonances. Coupling to wood structures and not getting resonances is extremely difficult. It's not impossible, but it's hard to do unless the vibrations are small. Turntables can be coupled to walls with a wall mount turntable stand, but a speaker simply produces too much vibration. It's luck (and construction) more than anything else if it happens to work for your speakers. And it doesn't prevent structural vibrations (footsteps for example) from going back into the speaker. After all, the vibrations go in both directions) Decoupling with spikes is more likely to be effective for speakers than coupling. It doesn't connect you to resonant wood, and it makes it harder for environmental vibrations to travel into the speaker.
Remember that you can use both.
Put an electronic component on cones or spikes and it helps to prevent vibration from entering the rack and vice versa. If vibration does enter the rack, use a heavy, massive rack that is non-resonant at audible frequencies so that the energy that does escape through the spikes is damped. Now put the rack on spikes so that energy in the rack doesn't escape into the resonant floor and so that resonant energy doesn't come up into the rack. Put the rack on a granite slab so that energy from the rack is dissipated by the mass of the non-resonant slab. Put the slab on spikes so energy from the slab doesn't go into the floor and resonate, and so that floor energy doesn't come into the slab. This is an extreme example, You take it as far as you want given your budget and your evaluation of whether it's cost effective or not in terms of audibility. The point is that it's a series of decoupling techniques that limit energy transfer, combined with coupling to massive, non-resonant objects to damp the energy that does escape.
It's the same thing with speakers if you choose to couple, only instead of a massive rack, you're using a massive house structure.
Put speakers on spikes to decouple. This helps to keep the energy in the speaker where it is damped in the speaker by the cabinet or stuffing. The spikes also limit energy coming into the speaker from the outer environment. In addition, the spikes also keep the speaker energy from energizing the resonant floor.
Because electronic component vibration is small, coupling to a massive, non-resonant rack is a good way to go, especially since you need a rack anyway to hold the gear. You can then add cones as a final tweak if you think that it makes a difference for whatever miniscule vibrations which are left, or if there are environmental vibrations coming into the rack and then into the component(or put spikes on the rack)
For speakers, the amount of vibration is so large, the only thing massive enough to couple to for damping is your house. But if your house is wood, coupling is a problem because of the resonant wood. So the more efficient approach for speakers is to decouple with spikes.
So for reasons of practicality, tame component resonances by coupling to massive, non-resonant racks and tweak with cones; tame speaker resonances by decoupling with spikes.
