Chris lz
The accuracy of a triple-INS installation may be very good, but it is not perfect. Over the course of a six-hour flight it would not be unusual to see an error in the region of 1 - 3 nautical miles, and greater on exceptional occasions. The error is likely to be cumulative, i.e. the longer the flight the larger the error. It can only be identified by comparison with a known fix point, so poor fixing performance (as opposed to a system malfunction) may not be immediately apparent to the crew. For these reasons Flight Management Systems navigational computations used radio navaid fixing, principally DME/DME, to update the INS-generated position.
In the case of the Antarctic flight no such updating would have been available, so the crew were relying on raw INS data.. For the en route navigation this would have been entirely satisfactory - the ANZ experience of operating DC10s over the Pacific and North Atlantic would have told them that. As wiggy has suggested most if not all companies would prohibit descent below MSA based solely on INS position, for the above reasons plus the risk of inaccurate data entry. But that is for a descent in IMC.
The principal requirement for a VMC descent below MSA is that safe terrain clearance must be maintained by visual means throughout the descent. Provided that requirement is met then there can be no objection to the use of INS. Whether to fly the direct INS track or to use the INS data to assist situational awareness while following a different track would depend on the circumstances at the time. Note these are general comments about VMC descents as I do not know how the requirements were specified in the ANZ SOPs of the time.
The visual misidentification was a fatal error. That was an opportunity to verify INS accuracy beyond doubt. Had the visual position been correctly identified the crew would surely have realised there was something seriously amiss with their INS position and maintained altitude to await a corroborative fix. As it was, confirmation bias seemed to set in. The INS could not be so wrong. A 27 nm error in their INS performance was quite contrary to their experience
henry crun
I agree with you. To the best of my knowledge civilian aircraft can only use the DME function of TACAN, not the bearing information.