should speakers be coupled or decoupled?

It depends upon your type of floor. Couple if your floor is rock solid and won't transfer significant amounts of vibration into your speaker. Decouple if your floor is relatively elastic and capable of transmitting vibration into your speaker. Think concrete slab construction ten miles away from any industry or highway versus a wood frame house with suspended flooring above an subway line.

It's sounds odd but spikes both couple and decouple.

By coupling a speaker to a stand and the floor, there is mass damping of vibrations from the speaker. It has to do with Newton's First Law of Motion. When the speaker drivers move forward, there is an equal but opposite force pushing back on the speaker cabinet. Energy is wasted moving something other than the driver, i.e the speaker cabinet. But if the speaker is coupled to mass, it can't move. Just like you can't push your hand through a brick wall. So with mass coupling, energy is moving only the speaker driver. You want coupling to mass.

On the other hand, vibrations can move from the house into the speaker. So you want to decouple the speaker from those vibrations. That's where the spike is better than a an unspiked platform stand on the floor. The tiny point of the spike creates an impedence mismatch between speaker and floor.(That's mechanical impedence, not electrical impedence. They're different concepts.) In other words, it's hard for vibrations to pass through the small point into the speaker, so it decouples the speaker, although it doesn't help with airborne vibrations.

So you couple to mass, and decouple from external vibrations. A spike does both.

A ball bearing type of device will work like a spike as it has a small contact area. But it has to hold in place. If it can move or roll around, it won't work as well as a spike.

Markphd, I understand what you are saying and feel that your logic is sound, but recently I have read reports and reviews from credible and reliable sources that claim that the use of a decoupling device such as Symposium's Ropllerblock jrs., can and do outperform spikes, when placed under a loudspeaker. Now a device of this type, basically a ball bearing of some sort sandwiched between two discs with a concave surface, will allow a fairly large amount of movement of the speaker cabinets, even if they are fairly heavy. Do you have any theories as to why this result may have been reported? Are these people hearing things? I certainly hope that, one such as yourself, wouldn't go down that road as it would seem to be, well for a lack of a better term, a cop out way to explain what these industry professionals report to be hearing.

On a lesser expensive level, I suspect one can ascertain for themselves what's best for their situation by trying both avenues... At least they'll find out what they can live with that way... and should.

you may enjoy using both methods simultaneously. Coupling say the stands, yet decoupling the speaker from them.

I'd go that way before I'd opt for some $200 + or way more expensive items that suit one path or the other best.... just to get some idea.

There's tons of threads here on this activity - philosophy, and it always comes down to what you will endure or enjoy the most. So, try both at some moderate level and go from there.

Remember, it's your ears and your wallet.

Happy listening!