I'll disagree with the technical details of your second paragraph, CountMike.
The PWM control system in computer fans that use that name is NOT the same as the way PWM power sources are used to control DC motors in larger industrial applications. (No, you did not say it was - I'm just pointing out to others that it is different.) In the brushless computer PWM fan, the fan receives 12 VDC from Pin #2 at all times. The circuitry in the motor case that does the brushless switching to simulate the effect of a brush-and-commutator system in standard DC motors also has an added chip that applies the PWM signal supplied from Pin #4 to that 12 VDC power source before it flows through the motor's windings, and at that point the power supplied does look very much like a common Pulse Width Modulated DC supply to industrial DC motors. By the way, the PWM signal operates at 20 to 25 KHz usually in these computer systems. But the resulting voltage fed to the motor windings actually stays at 12 VDC peak, switching from 12 V to 0 V over and over as the PWM signal directs. The result, of course, is that the current flow through the windings also switches from max to none, etc., producing torque pulses. But since the torque is not at max all the time, and the motor is experiencing resistance to torque (it is driving fan blades to move air) it does not turn at the motor's rated max speed unless the PWM signal actually calls for 100% On (i.e., full speed).
Doing things this way instead of the way older larger PWM power supplies do it has a few advantages. It means the mobo circuits for the fan headers do not have to have DC Amplifier outputs that can supply PWM-pulsed power at 12 VDC max and up to 1.0 A at 100% On state - they can be simpler and very similar to what already was on mobos for the older Voltage Control Mode. The circuit elements inside the motor that do the commutator simulation job required by small brushless fans can still receive a smooth reliable DC supply from Pin #2. (Note that makers of these fans specifically tell you NOT to try to feed such motors with PWM power that is already modulated into a pulsing source, as older methods do for large DC motors.) And, it made it much easier to build backward compatibility features into the new 4-pin PWM fan designs so that mobo makers and users have a relatively easy way to adjust to either fan type.
Regarding the speed signal sent by a fan back to its host mobo header, as far as I can tell the mobo speed control system does NOT use that at all - it is NOT a speed feedback control system. The control loop actually is a TEMPERATURE controller that uses Temperature measured at a relevant sensor as the feedback information. The control loop simply sends out signals to manipulate the fan speed to achieve its temperature target and actually does not care what the speed is. (In fact, if you have a "2-pin fan" - that is, a fan designed to connect to a PSU source only with no control and no speed signal output from the fan - that fan CAN be connected to a mobo header using the older Voltage Control Mode and its speed can be controlled perfectly by that header without any speed signal.) BUT, users are interested, and that signal is already available inside the brushless motor - it is generated for use in the brushless switching circuits to synchronize the switching action with rotor position - so it's easy to access and display. PLUS, there is a more practical use of the speed signal, too. That is FAILURE detection, the secondary function of each fan header.