PaulyAlcorn :
I have often pondered the same type of technology, but it largely boils down to the vagaries of internal air patterns, fly height and vibration. The fly height of a head over the platter is amazingly small, and the head moves due to vibration and flutter, thus requiring hundreds of real time adjustments per second. The actual tracks themselves are not perfect concentric circles, they are often more of an oval(ish) shape and vary from platter to platter. In essence, the platters are not the same, and thus the track positions vary. The heads are lined up all at once; the voice coil actually moves the entire head assembly with all 12 heads (or however many are on the HDD) simultaneously. The read head would line up with the track, and move as it reads that single track, on one platter, but this would cause misalignment with the other tracks on the other platters.
In order for the system to work there would have to be independent movement of each individual head/arm. This would be much more complex and require a voice coil actuator for each assembly, which would add a tremendous amount of complexity. I suspect the real hindrance is the cost. I believe it is possible with today's technology to realize that goal, but doing it in a cost effective manner is what is beyond reach (imo).
Actually... I think you are ever-so-slightly mistaken on a number of fronts. Please allow me to show you where:
[1] head fly-height, small, etc.: there are NO externally applied 'adjustments' to the head-fly-height, and haven't been since the very first flying-head hard drives were invented. Its all about the aerodynamic stability of applying a significant but fixed force (from the springiness of the head mounting arms) to the heads that have a small polished air-berm carved in the front, which builds up and drags a pad of air under the head, allowing it to 'float'. While vibration and even position-relative-to-the-edge-of-the-disk make a difference in fly-height, these are all counterbalanced within 10% height flutter by the action of the air berm.
[2] The actual tracks may not be concentric circles, but slight ellipses, as you say. Though the platters are not the same, remember that the tracks are only 100 nm wide, and 130 nm center-to-center distant for an 8,000 GB drive. While the head CAN be modulated a few dozen nanometers side to side at the rotational rate (120 rps), when tracks are written, this kind of slewing is suppressed (otherwise tracks would overwrite tracks.) But yes: the track-over-track alignment can definitely vary even on fairly short time-scales.
[3] regarding whether the read head(s) of multiple PLATTERS line up is not material to my proposal: on a particular side of a platter, the tracks are QUITE WELL aligned with each other. If nominally (and in fact, by design) the tracks are 130 nanometers apart, then a series of read-write impressers/sensors along the trailing edge, each separated by N * 130 nm would do the trick. The tracks don't even need to be exactly adjacent. Since 'reading or writing" any particular track would always synch on the ''modulo 8 or 16'' base track, all the tracks would remain in alignment. Carefully choosing a magnetic head substrate material with the same thermal expansion coefficient as the disks would be important. So that even if the platters expand under thermal loading (they often run above 100 C!), the track-spacing scaling would be matched by sensor-head expansion.
[4] Again... this idea does not require separate actuators. It simply requires re-engineering the read/write heads to have multiple track sensors in a row at the back of the head assembly, and an onboard chip to take care of simultaneously reading and/or writing the data. Modern integration ought to make this now quite possible, and even possibly at very low incremental cost.
Given that hard disk manufacturers are loathe to sell $59.95 disks (they basically make near-nothing from such sales), and only make decent money from the top-end of their offering(s), it would seem like a juicy cherry indeed ... that would very likely have a BUNCH of corporate buyers, were the starkly increased data throughput speeds turn out to be sustainable, and data integrity not compromised in any way.
GoatGuy