A couple thoughts given your description.
Systems like the ZS, BTZS, et al are built on pieces of information and cobbled together into a workable "system". These systems, of necessity, move us a step or two or three away from the "real" theory and principles they are built on.
Here's a link to a discussion talking about one of the building blocks involved, ISO film speed. http://www.apug.org/forums/forum48/1...-bill-etc.html The builders of the various systems are trying to help us make good pictures even if our eyes glaze over when given this level of detail. To be blunt it has to be dumbed down a bit because a huge number of us don't really care about knowing the theory, so many details are left out.
A one of the complicating factors here IMO is the classic teacher/student relationship where students ask questions and teachers are supposed to provide answers. Teachers do their darndest to answer well but they are human and we regularly get the teachers opinions interjected, not always the real story. Along the way the developers of the systems interject their own thoughts to explain how things work and relate to each other.
Looking back over the last 50 years across almost any discipline that we might choose, we can see a huge variety of examples where our understanding has been updated.
In this piece by Phil Davis http://www.btzs.org/Articles/Sensito...20Part%205.pdf I think there are examples of the above issues.
In the middle coulomb of the second page (31) Phil makes the assertion that incident meters assume a 5-stop SBR because of an 18% transmittance. In the third coulomb he goes on to assert that a full sunlight measurement "almost inevitably results in of underexposure in the shadow areas of the subject".
IMO in both of these assertions Phil is personally trying to "fill in the banks" and by doing so he was unintentionally creating myths/misinformation.
To the best of my knowledge incident meters don't assume a given range, a single reading doesn't do anything except gather luminance info at a given point in space. There are meters that can spit out SBR but they need at least two readings. There has even been experimentation with "back leak" to allow this to be done with a single push of the button, I don't think this ever made it to the mainstream though.
Also incident meter domes aren't necessarily allowing exactly 18% of the light they "see" onto the sensor. The percentage of light the dome transmits to the sensor is irrelevant except to the meter's builder, the real percentage is just a number in the equation. For example, a builder could easily use a more transparent dome to improve low light usability and simply adjust the equation to make the displayed readings come out right.
Next, Phil's expectation with regard to shadow detail is to be frank, just his own. It's a purely subjective preference. Nothing wrong with that but that preference carries through into his system, into his equations, regardless of your preferences or mine.
Which brings us to the contrast issue you and Ben are trying to address, that when you input certain numbers into the BTZS system, from readings I assume are taken in the normal BTZS manner, the BTZS spits out answers that aren't getting you the results you hoped for. This failing is a direct result of Phil's preferences and biases.
Now I want to be clear here, this criticism doesn't mean I think the BTZS or the other systems are without value, for me it just means that they need to be taken with a grain of salt. Ben's patch is a good example of seasoning BTZS to taste.
We need to understand that the idiosyncrasies of any system do not necessarily represent the real physics of photography.