Disregard this. a PC with a failing HD will still get to the post screen and show an image.
Not so!
yes it is. the BIOS only starts issueing BIOS commands upon beginning the Plug & play detection.
In this case the ATA "Identify drive" command would freeze the machine during boot- but would freeze at the PNP load screen.
You also can't flood a bus with I/O errors without an initial I/O request...
However- I'm still wrong, since a faulty drive could easily be flawed to the point of consuming too much power during boot, preventing the PS from being able to pass POWER_GOOD or PS_ON and thus allow the CPU to exit the reset loop.
I can almost see it via the logic board but I cannot think of a reasonable way that the logic board could flood the IDE bus enough to prevent the BIOS from initializing the video adapter, even without any IDE commands being issued, a flood of ATA messages across the bus would likely be ignored. BUT- a lot of clean-room BIOS code is very timing dependent and so could be influenced in this way to prevent video initialization. I would imagine the necessary amount of flooding needed to freeze before video initialization would vary based on the IDE speed implemented at the BIOS level as well as the speed of the processor.