With Alps SKCL/SKCM, there are well-documented visually-observable changes: height of the switchplate, indentation of the actuator leaf nub, redesign of the slits, reversible and non-reversible sliders, removal of slits, and a wide variety of numbering and branding forms suggesting multiple factories and reduced quality control. This gives reasonable credibility to the claims that the switch feel degraded as a result of product alterations. In fact, SKCM White was recorded as SKCMAQ in the 1994 catalogue (shown as pine, I think — Thunderbird's message search seems to be hopelessly broken in recent versions), but SKCMCQ in a later data sheet that shows bamboo white alongside pine yellow (SKCLAR), which hints that bamboo switches were considered separate products (with a suggestion that the Far East contract manufacturers switched to bamboo first).
Vintage Cherry MX Black is much worse, as the chief reported visual difference is the Cherry logo. In fact, the switch contacts were redesigned at some point (with some clone switches copying the original design), but whether that has any correspondence with vintage/modern blacks, I don't know. Lots of people consider the better feel of vintage blacks to be a myth, but it does stand to reason that Cherry would have been looking for ways to cut costs to compete with the clone manufacturers in the Far East, and to compete with other high-end manufacturers, just as Alps would have had to. Chicony in particular were known for frequently retooling to incompatible switches, presumably to whatever cost less at the time — Cherry MX, Yali, Alps, Himake, Futaba and Mitsumi having all been used at some point. Some FK-2001 keyboards used an Omron switch never seen before or since — similar to the B3G-S, but smaller.
With Alps, the first significant change in feel was from SKCM Blue to SKCM White — the latter doesn't feel as soft, clean or smooth. Thus far, nobody has actually proven any differences in feel. Due to variations in degradation, I assume you'd need to measure lots of keyboards and average out the force graphs. SPARC's graph showing white, blue and black depicts a horrible force curve for black Alps that matches how it feels, but I don't know if that's actually bamboo black or not; blue and white have different graphs, but the roughness of the graphs suggests that no averaging was done, and from experience I can say that white Alps can degrade terribly.
To prove the cause of the changes, you'd need to measure the springs (turns, length, wire gauge etc), analyse the plastics, and so forth. Alps used dry lubricant with the older switches, that seemed to stop, but nothing's proven yet. There are a lot of subtleties, such as Matias quiet click switches having a white slider instead of the intended orange after the orange dye was found to interfere with the frictional properties of the plastic and thereby the feel of the switch. You get a better click out of a Matias click switch if you substitute in an OA2 click leaf, that seems to look the same.
It's a huge combination of subtleties — steel composition and processing, metal thickness, plastic selection, part shape, keycap weight etc. Rubber domes aren't inherently bad, but their poor reputation comes in part due to their correlation with cost cutting: the kind of companies who would replace discrete switch modules with domes and membranes aren't going to care much if anything about how the switches feel. Companies with a good reputation (e.g. Topre and Key Tronic) still care about the feel even when using domes. Likewise, mechanical switches require a lot of knowledge and experimentation to get right, and it's a balance that's easy to interfere with.
