Just to clarify this point (Granted, it's beating a dead horse.):
(Classic parallel) IDE is actually a subset of the IBM AT's 16 bit ISA bus. Technically the only connection it has to the MFM/RLL ST-506 hard drive connector standard is from a software standpoint the original IDE drive's programming interface emulated the command set of the Western Digital WD1003 hard drive controller used in the original IBM AT. The "Hyperdrive" is only "essentially IDE" insofar that it uses a discrete controller chip on what's essentially a "PDS" bus. (ISA can pretty much be thought of as an 80286 PDS slot, with DMA and interrupt controller functions tacked-on.)
To understand the difference between ST-506 and IDE, well... think of the difference between a bare floppy drive and a USB floppy. A bare floppy drive is cabled to a controller integrated into the host computer, and said controller has the job of translating high-level data transfer and control commands into the appropriate signals to control the mechanical components of the drive directly. The limitation with this sort of system is that a given controller can only work with floppy drives who's mechanisms are self-similar enough to use the same set of mechanical commands. Without some fugly hacking it is thus impossible to use an existing floppy controller to control some new-fangled "superfloppy" that, say, has more heads or uses a faster data rate. A USB floppy, by contrast, adds "intelligence" the the drive itself, allowing the drive to reside on a general purpose bus which can be shared with other devices which use completely different mechanisms. (like hard drives, flash keys, etc.)
Early IDE drives fairly literally were just ST-506 hard drives with the stripped-down equivalent of a WD1003 integrated onto the drives's circuit board. It may seem somewhat counter-intuitive that integrating more "intelligence" onto the drive would make the drives cheaper than the drive portion of existing two-piece controller/drive combinations, but by integrating the controller on the board manufacturers were freed from having to make their drive mechanisms adhere to any existing limiting data encoding or transfer conventions, allowing the amount of data crammed onto a drive to increase much faster than it would be able to otherwise.
SCSI followed exactly the same evolutionary path. Early SCSI "drives" were generally composed of an ST-506-cabled MFM or RLL mechanism cabled to an external controller board containing an ST-506 controller and a SCSI transceiver chip, and only later did the drives start integrating the SCSI transceivers and controllers onto the drives themselves. Really the only difference is that SCSI was a "synthetic" standard designed from scratch to be a robust multi-purpose bus usable internally or externally on long cables with many devices while IDE is a simple, cheap shortcut designed for software compatibility with an existing standard (the AT BIOS) and capacity needs typical of small home and office computers in mind. (It was pretty rare for anyone to have more than one hard drive back then, so IDE being limited to two per cable seemed adequate at the time.)
Anyway, blawblawblaw. SCSI probably was the "correct" choice for Apple at least on the Mac Plus, since they were insisting at the time that hard drives needed to live externally and IDE isn't appropriate for cable lengths over about 2 feet, but it did saddle Mac machines with needing "overpriced" drives for a decade to come. Probably apropos considering how much they overcharged for their hardware generally, but nonetheless...