There's an interesting issue here that may seem subtle but I believe it's worth highlighting.
You mention Evidence Explained (EE) by E. S. Mills. Citations are a great start and all newcomers should listen to your words. But how should they be recorded?
EE gives both a printed style and a list of parameters for the citations of a huge number of sources. Does this mean that we should write them all long-hand? Does it mean that we should enter the parameters into a software product and let the software generate that printed style whenever we ask for a report?
If we're sharing our data with someone else - someone who may use a citation style different to CMOS, or who may come from a different country to ourselves - then what are we actually sharing with them? Will it make sense to them?
These are questions that have been discussed many times in forums such as BetterGEDCOM, and a scheme must eventually be part of any new data standard.
The essential point here is that what you record isn't the same as what you see later. This will depend on whether you rely on software to store and manipulate your data, or whether you do it the old way with written documents. It's important because giving up total control and relying on software isn't a natural progression for everyone. If you're someone with hand-crafted citations in your data then it may be because you distrust software to make an accurate job of it, or the product you use just isn't up to the job, or you use especially intricate citations (e.g. with multiple inline references when explaining your reasoning).
This differentiation between "data" (i.e. what is stored) and "presentation" (i.e. what you see later, whether on-screen or on-paper, and including charts) is not confined to just sources and citations. It applies to all our data, including narrative text - something I'd love to elaborate on but it would overload this response.
In the case of simple citations, a product could rightly record just the parameter values of the reference. This would mean that the selection of a corresponding printed style such as CMOS, or the application of locale preferences such as date formatting, could be done by the product of the respective end-user. They are not part of any data exchanged with another party. With some thought, this could be extended to include complex citations such as those that include multiple sources and/or author annotation.