Why does conventionally set up audio normally fail to 'conjure'?

There are many factors at work, but the fundamental reason is that insufficient experimenting and research has gone into understanding how people can pick up on subtle clues in what they're hearing, usually unconscious, which allows them to easily separate 'real' sounds from an attempted mimicking of such via a high fidelity system. Most playback manifests clearly audible "flaws", which then makes it trivially easy for the mind to "spot the fake", and once this happens the mind will automatically keep referencing these "defects", reinforcing one's awareness that the illusion has failed. An analogy would be watching a stage magician after being made aware somehow of how he does it - the tricks will be quite obvious, and part of one will be somewhat surprised that others in the audience don't also see how the manipulation occurs - your attention has been drawn, irreversibly, to the "chinks in the armour" of the stage act, and while you may be impressed by the magician's skill you no longer marvel at the "trick" itself.

It is unfortunate that during the history of audio reproduction that more in depth examination of the crucial factors didn't happen - many of the pioneers had an intuitive understanding of what was necessary, but this didn't evolve into an industry wide set of highly effective standards and rules which everyone could usefully reference; Paul Klipsch is a good example of such a person.

So, the illusion usually fails ... again, why? The answer, because the distortion, meaning elements of the audible sound which don't match the recording content, is too obvious, and intrusive. Most of the industry with a technical focus has become infatuated with easily obtainable raw numbers on behaviours of the individual components of a system: the source mechanism, the amplifier, the speakers exist in completely different spaces, are seen as separate units on a test bench; the overall performance of a complete system in its final environment is seen as something that is too hard to measure - that aspect which is in fact the most important of all is never sensibly dealt with. Unfortunately, the types of distortion which will always cause the illusion to collapse or fail to manifest easily arise in an end system, many times as a result of interaction between the components and the environment in which it finds itself. The fact that the "measurers" are remarkably blind to this is a bind that the audio industry is currently trapped in, and there are no signs as yet of any meaningful progress ...

But fortunately this does not stop the individual being able to bypass these limitation of thinking, and now and again one can read how this has been done by someone who has chanced upon a lucky combination of factors that reduce the key distortion sufficiently.