Actually, none of these protocols for reducing the appearance of bias, suggested or implemented, can themselves prevent bias. They not only can't stop it, but one could argue that, in the long run, they make it more ingrained by forcing appraisers to be continually conscious and aware of the issue. Many will learn to resent the forced obfuscation of reality and seek pathways to bypass appearances through self-indoctrination to the methods that simply amplify existing bias through cloaking procedures that effectively ensure bias becomes procedure.
We see evidence of the above, time and again, on this forum. It should be quite obvious.
What is needed, like or not, are strict protocols for improving appraisal methods, including the application of mathematical constraints on the Sales Grid and the use of statistics. Unfortunately, the statistics used to understand such protocols requires a good knowledge of Matrix Algebra and some calculus, as well as some programming skills. ... Until the skill sets become more availabe - things will ONLY get worse.
Neural Networks and MARS regression are not particularly challenging; understanding them is certainly in the realm of the upper 2-3% of intelligence. Maybe even the upper 5%, - of males in any case (that is just a fact).