I don’t think this correct. Behe’s argument depends on which is the dominant pattern, or in other words, what the net effect is.
If there are a few examples of increasing genetic sophistication/function, but far more abundant examples of decreasing function such that the overall trend is diminishing function over time, then Darwins mechanism would be incapable of producing sophisticated lifeforms. The net effect would be a loss of function.
Conversely, if there are only a few examples of diminishing function and far more numerous examples of increasing function, then it’s clear that Darwin’s mechanism is capable of producing increasingly sophisticated lifeforms, despite the occasional “devolution” setback. In other words, the net effect would be positive (increasing genetic function).
Thus determining the dominant pattern is the important question here. In my work with stickleback evolution, it seems like loss of function is the dominant pattern. Loss of function mutations at Eda are responsible for differences in lateral plating, loss of function mutation at Pitx1 are responsible for changes in pelvic morphology, and numerous other loss of function mutations underly evolution in school behaviour, body size, trophic morphology etc. As far as I am aware, nearly all the notable evolutionary changes have occurred through loss of function mutations. While examples of gains in genetic function could be present, it would be fair to say that the “highly derived” populations of stickleback possess a reduced amount of overall genetic function.
So I don’t think offering a few counter examples would be refutation of Behe’s argument. To be convincing, the overall pattern needs to be demonstrated. I hope Behe is arguing for the overall pattern in his book, and I hope any critique of it would do the same. I think the best way to do this is look at genome wide function over time, not individual genes - such as Lenski’s LTEE.