Read interesting post here that mentions constant throughput timers

Was wondering, if using constant throughput, however you configure the load, it will try to generate same throughput (e.g. hits/sec). So how does that relate when you vary # of threads, increasing/decreasing and see how system responds.

Is that sort of like this car traffic example:

* more cars enter/exit the road, stop and go traffic. When go, go fast 60 mph, at stops, slow down to near 0. Average speed/throughput is 35 mph.

* road traffic stable in that very few cars enter/exit, but all cars tend to go 35 mph (perhaps speed limit on local road)

In both cases traffic pattern different but average throughput ends up being same. Is that how you would relate to use of constant throughput timer with varying # of threads and ramp up times?

If not, can someone elaborate? I'm just a novice with web load testing and JMeter.