What the article leaves me uncertain about, however, given that in modern times a cutting head is to at least a loose approximation a velocity-based transducer (meaning that it converts signal amplitude to groove velocity), is **how** the cutter is caused to cut at constant amplitude at those frequencies.I think I can answer that. Since the cutter isn't by any means a true constant velocity device, it gets equalized so it is. The CV equalizer is a pretty complex set of time constants; I'd hate to have to work them all out, since the actual cutter response is more like a loudspeaker with some odd peaks in it (complete with a 'nose' in the high frequencies, just like any raw driver response) than anything else.
Now the slick way to set up the whole thing would be to combine the qualities of both the RIAA curve into the CV equalizer curve so that the output was both at the same time but I suspect in order to preserve hair follicles this wasn't done and probably won't ever :)