I have long held that 100% of calculation inconsistencies can be avoided. I am changing that position, somewhat reluctantly, given strong evidence that this is a sound theoretical position, yet it is not a sound practical position.
While it is the case that you can make such calculation inconsistencies go away, it is not the case that you can do so and make the set of financial reporting disclosures which one would desire to make from an accounting perspective.
The reason I am reluctant to take this position is because I see many XBRL instances which now sloppily hide behind the fact that calculation inconsistencies CAN exist, now have calculation errors which should NOT exist intermingled with true calculation inconsistencies and those calculationerrors go undetected. This is unfortunate, but I cry uncle and admit that it is unavoidable in many cases.
My new position is that XBRL calculations or XBRL Formulas should exist which prove that everything which should foot or cross cast, does in fact foot and cross cast. You prove that, then you can safely ignore the calculation validation results which do report calculation inconsistencies.
Regretfully from the perspective of an XBRL instance consumer, this makes things tougher. My answer is that a good analysis application will show you things which do add up and will somehow mask the confusing calculation inconsistencies which really mean nothing.