Another consideration - the water in the glass is a relatively undamped (no pun intended) system and once set in motion could slosh around as a sort of resonance. The actual driving impulse from the shutter could be a very short and quickly damped pulse within the camera mechanism. As such, the internal camera vibrations could be settled out during a large percentage of the actual exposure; e.g., a half millisecond wiggle during a 1/2 second exposure won't even register, even though the water sloshes a bit. I also seem to recall some discussions about shutter vibration being a bigger problem at 1/25 or so than really slow speeds for that reason, but it may have been with regard to a very specific camera and tripod.