Postol is more than a bit of a nut. He has some valid points regarding real weaknesses (past and present) in certain US systems, but he constantly exaggerates things massively out of proportion, and baselessly assumes that zero improvement is possible over time, while dismissing many major real practical solutions to the issues he raises for unclear reasons (which are rarely sound or well-reasoned).
Many of his arguments hinge on results from 30+ year old tests of initial prototype systems, and frequently assume that a virtually infinite number of unrealistically perfect decoys can be packed into tightly space and throw weight limited missile platforms.
His arguments that IR-only seekers are incapable of reliable decoy discrimination against certain forms of decoys are shaky but have some legitimate foundations. However he avoids addressing the issue of X band radar. A LRDR-style site employing X band radar can eliminate upwards of 90% of the "issues" he brings up with decoy discrimination.
I do agree that with sub-X band radar solutions, it's probably impossible to reliably discriminate all forms of decoys, and that he has some very valid critiques there.
It is also quite amusing that he constantly nitpicks the oddest minor details in many tests, even when these nitpicks are irrelevant or no longer present in newer tests. He also seems incapable of nuance, and believes developmental tests should jump straight from prototype to max-difficulty intercept simulations, which is hilariously unrealistic.