From Jobst Brandt:
Tread patterns have no effect on surfaces in which they leave no impression. That is to say, if the road is harder than the tire, a tread pattern does not improve traction. That smooth tires have better dry traction is probably accepted by most bicyclists, but wet pavement still appears to raise doubts even though motorcycles have shown that tread patterns do not improve wet traction.
A window-cleaning squeegee demonstrates this effect well. Even with a new sharp edge, it glides effortlessly over wet glass leaving a microscopic layer of water behind to evaporate. On a second swipe, the squeegee sticks to the dry glass. This example should make apparent that the lubricating water layer cannot be removed by tire tread, and that only the micro-grit of the road surface can penetrate this layer to give traction. For this reason, metal plates, paint stripes, and railway tracks are incorrigibly slippery.
Besides having better wet and dry traction, smooth tread also has lower rolling resistance, because its rubber does not deform into tread voids. Rubber being essentially incompressible, deforms like a water filled balloon, changing shape, but not volume. For a tire with tread voids, its rubber bulges under load and rebounds with less force than the deforming force. This internal damping causes the energy losses of rolling resistance. In contrast the smooth tread transmits the load to the loss-free pneumatic compliance of the tire.
In curves, tread features squirm to allow walking and ultimately, early breakout. This is best demonstrated on knobby MTB tires, some of which track so poorly that they are difficult to ride no-hands.