Well, the argument is that that if the flow from the tap is fast enough to keep 1 full, the additional head of pressure will fill 5 faster and then 2 faster than the water can escape from 2 into 4, therefore 2 will fill first.
pgy1+1/2pv1^2=pgy2+1/2pv2^2
I don't agree, I think the guy arguing that case (who does have a physics degree) is not taking account of the following:
1. The flow from the tap is unknown. If it were high enough, it could fill 1 before it fills 5
2. The exit tube from 2 into 4 will remain at zero pressure because 4 will never fill past the entry point, so the extra speed of flow into 2 would have to be very much higher for it to fill faster than it could exit into 4.
3. There are no specific measurements to calculate an accurate flow rate. The puzzle is designed to demonstrate how water finds a level in any shape vessel or arrangement of tubing, not the dynamics of water flow rate, given pressure, diameter and viscosity. Whilst the tube from 2 to 4 does look thinner, it is not a scale drawing and without any specific measurements, there is no way to know if it would be small enough to cause 2 to fill first.