It's going to be a lurching ride
The interpretation of the obstacle might be correct, but I'd rather instead follow the continuous acceleration interpretation (also mentioned by Kira).
The disk is immobile while you are within 20 feet of it. If you move more than 20 feet away from it, the disk follows you so that it remains within 20 feet of you.
The question, then, becomes: how fast is the disk going to accelerate? This is not specified in the spell. I find this quite unfortunately since the disk can be up to 100 ft. from the Wizard, blocked by an obstacle, and thus as soon as the obstacle is removed the disk should come to within 20 ft.
Now, in the absurd case of an infinite, or very high acceleration, any object on the disk would in this situation be ejected either at the start of the acceleration or the end of the deceleration. It does not seem very useful, a disk which would lose everything onto it at the slightest obstacle, so I'll assume that in the case the disk accelerates and decelerates moderately enough that anything on the disk stays on the disk (though it might topple cups and the like)
All right, so the disk starts to accelerate smoothly, the players might feel very clever...
... right until the Wizard starts having difficulty hanging onto the ledge of the cart as the winds push him back, and indeed everything onto the disk is pushed back by the wind until it topples off the disk. But let's imagine the players solve this issue, they might feel very clever...
... right until the Wizard cannot keep his eyes peeled to direct the disk because of the wind. But let's imagine the players solve this issue, they might feel very clever...
... right until air starts to heat around the Wizard (friction creates heat after all). But let's imagine the players solve this issue, they might feel very clever...
... right until the Wizard's reaction time starts to be quite insufficient to dodge obstacles. But let's imagine the players solve this issue, they might feel very clever...
... well, I could continue for some time, but the ultimate issue of reaching relativistic speeds has already been explored by Randall Monroe, of xkcd fame, in its What If 1: Relativistic Baseball article.
Oh, and of course, let's also talk about deceleration. There are two ways to decelerate should the Wizard suddenly be within 20 ft. of the disk:
- smooth deceleration, aka inertia
- immediate deceleration, aka catapult
The latter might be a little violent, and contradicts the smooth acceleration thing, so I would instead advise a smooth deceleration. That is, a disk that does not stop until it hits an obstacle or is given sufficient time to return to a null speed (like, proportional to the time spend accelerating).
The main advantage of smooth acceleration and deceleration is that it does reward players' creativity, as long as they manage to avoid infinite acceleration and handle the inertia. At the same time though, there is no stable speed here. The disk is either accelerating because the Wizard is more than 20 ft. or decelerating because the Wizard is less than 20 ft.; that makes for a lurching ride, not everybody might appreciate.
Oh, and I seem to remember something about the disk hovering close to the ground; hope your Wizard pilot avoids the cliffs!