Although I have only played CoC thus far, I love Chaosium's game system. As for the skills, what immediately occurred to me as an enhancement, would be a skill "tree" by era. In other words, some skills are related to each other in such a way that there might be enough in common between them that they could provide some cross-over capabilities. But in order to model this accurately and neatly, a tree would be needed, wherein more general skills serve as umbrella categories for more specialized skills. Of course, both technology, and culture progress (or, sometimes regress) from era to era, and this would provide an opportunity to create specialized skill trees based on eras - and this could be done in the form of monographs or supplements for the particular era/world/culture in question.
In terms of game mechanics, each monograph/supplement could detail both the chance for a more general skill to succeed at a more specialized skill, as well as how much a specialized skill might cross-contribute to a more general skill further up the trunk. One way to model this - that would make sense - would be for each monograph/supplement to first map out a skill tree. Then, the specialized branches of the skill tree could be categorized into tiers [e.g. Tier 1: 1920s firearms; Tier 2 1920s Handguns; Tier 2 1920s Guns w/Gunstocks, Tier 3: 1920s (Gunstocks) automatic weapons; Tier 3 1920s (Gunstocks) sharp-shooting weapons (rifles, etc); Tier 3 1920s; Tier 3 : 1920s (Gunstocks) scatter weapons (shotguns)]. Once the number of tiers for each root skill has been determined, we could then take the total number of skills from the highest tier (the most specialized tier) and this would determine how many points would be allocated to the tier below for each specialized skill learned; these points divided by the total number of points possible from the tier above (the more specialized tier) would provide the percentage for success for the tier below.
So, what about branches that don't have as many tiers as others? How should they be weighted? Well, progressing from tier to tier, for each branch that is missing a tier, a general weight could be assigned by taking the average number of skills of all of the skills under the same branch at that tier. This would then be a reasonable weight for the purpose of determining the points to be added to the branch on the tier below it (the more general skill, closer to the root skill).
Now, this could be a bit tricky during design due to the fact that cross-over can exist between skill branches with different skill roots (e.g. veterinary medicine would have some cross-over with human medicine; traditional medicine would have some cross-over with allopathic medicine, etc.). From the supplement/monograph designer's perspective, this could be addressed through the use of a custom software program that can handle such complex ontologies, and perform all of the same calculations that I mentioned above in a multi-dimensional topology (as opposed to a 2d topology where branches from different roots have no lateral branches between them). While this all sounds (ab initio) to be rather complex, such a software tool (even one made publicly available online) could be used by monograph/scenario designers to achieve both integration and consistency. As for changes in the core rules from edition to edition, such a software tool could be updated (if actually need-be) at the time (or shortly after) of the release of said new editions.
Although another challenge, for such an approach, would be deciding how trees would be provide cross-compatibility from era to era (e.g., how would experience with a tommy-gun provide skill with a more modern automatic weapon?), by using such a software management tool, such challenges could also be deftly mitigated.
Any thoughts? Any takers? [Electronic & Software gaming tools are increasingly the wave of the future - why not incorporate tools for the designers?]