Augiedoggy is correct that if the system is designed correctly, it shouldn't matter. As he points out, things can get "off" pretty easily. It's been my experience that the nature of homebrewing is too inconsistent to have a "correctly designed" system because what works for one type of beer doesn't for another. Consider this:
Brew Session # 1: a low gravity session ale; 8 lbs of grain and a water to grist ratio of 1.25 qts/lb. You carefully tuned your PID controller with water ahead of time, and everything works pretty well.
Brew Session #2: a Barleywine; 16 lbs of grain and a water to grist ratio of 1.00 qts/lb because you can't fit it all into the mash tun with a higher ratio. Now your carefully tuned PID burns your wort. What happened???
Your system was tuned for very little grain and a lot of water. It worked ok as long as the beer you brewed was close to that. With the Barleywine, the higher amount of grain combined with relatively less water means flow rate was a lot lower. The higher sugars in the fluid burned on the heater because it was tuned for a higher flow condition.
Our processes change too much from batch to batch to be able to have consistent temperature control if the control probe is measuring the mash temperature directly. Putting the probe just after the heater makes the system able to respond almost immediately to changes, and this minimizes the impact of the the differences between batches.