Water turbulence will not normally decrease flow rate.

Its effect tends to be that it limits flow rate.

The comparison with traffic is far from perfect; water has no intrinsic destination but follows the path of least resistance. Most traffic has specific destinations which are at odds with the specific destinations of other traffic sharing intersections. Moreover, the path of least resistance for an automobile could easily be to move in completely the wrong direction, and less resistance is often to be encountered simply by staying put; autos are as subject to inertia as are water particles to gravity, but gravity drives the water and inertia is the first thing to overcome in dealing with automobile transit.

I wonder, though, whether street traffic couldn't better modeled with a chronological inversion of water flow. Traffic moves TOWARD destinations in somewhat the way water moves AWAY FROM sources.

Could we just flip the timelines from left to right and get a better fit?