

Right - the water has inertia in a straight line (as does the bucket). When they both try to go straight the string prevents it, accelerating them in a new direction. At each moment you look at the circular path the water’s inertia wants to go in a straight line (tangent to the circle). So at each instant it is behaving exactly as if you had been running in a straight line and stopped.
What I meant about the geometry and axial tilt - imagine that instead of a bucket you had a dinner plate with a bucket handle. So the water was all at the level of the top of the bucket rim on a plate. As soon as you stopped the plate would flip and the water would splash off. Likewise, if you had the string connected only to the bottom edges of the bucket rather than its handle, as soon as you stopped the bucket would flip due to the water’s strong inertial force against the side of the bucket. The setup doesn’t tilt the bucket in the right way to keep the water contained and impart the new acceleration upon it.
Funny. I just gave away my desktop with that same chipset & CPU. It was a really good hardware generation.
No plans for a replacement in the near future.