>>> For minimum belt creep, the platter and motor pulley should be as close as possible, so that the circumference of the platter is in contact with as much of the belt as is possible.
This overlooks the other (and more significant) half of the belt contact equation. It's true that placing a smaller circle closer to a larger circle increases the arc segment between the two tangent points around the larger circle, but it correspondengly reduces the arc segment between the two tangent points around the smaller circle. IOW, moving the motor closer INCREASES the belt contact patch around the platter (as Lewm said) but it also DECREASES the belt contact patch around the motor capstan.
As the contact patch around the smaller capstan will always be shorter than the contact patch around the larger platter, the capstan is where belt slippage is most likely to occur. Therefore, your setup shoudl be optimized to lengthen the belt patch around the motor capstan. This is accomplished by moving the motor and platter away from each other. How far? Far enough to optimize the benefit without incurring detriments from belt sag or elasticity.
Lew's argument to place the motors opposing each other to counterbalance tension on the platter bearing has merit, but it's not the only placement consideration. Motor orientation also matters. Assuming the motors are identical and running in synch, orienting them so that they point the same way would amplify any cogging effects (times 2, with 2 motors). Not a good thing.
One advantage of multiple motors is that, if synched and properly oriented, each can mitigate the cogging effects of the other. To do this you must orient them precisely out of phase with each other. How to do that? Find out how many poles the motors have. Lets say it's 12. Divide 360 by 12 (30), then divide that by 2 (15). Orient the motors such that one is twisted 15 degrees clockwise (or counter-clockwise) compared to the other.
Agree with Manitunc that proper belt tension is critical, particularly with a non-elastic belt material. That's what matters with regard to distance. It's not obvious to me, however, that the two motors need be equidistant. Aside from visual aesthetics and questions of available space, what's the benefit? One could even argue that differing distances could reduce the probability of simultaneous belt slippage events, since the contact patches of each setup would be different.
Agree with Lew that one good motor is a better solution than any greater number of flawed motors. The best you can hope for with multiple motors is that their flaws will somewhat cancel each other (as with the anti-cogging orientation described above). Better not to have the flaws in the first place. A non-cogging motor will always have less cogging than two cogging motors, however perfectly arranged. One motor is quieter than two. Etc...