GUI guy was trying to make the point that as software engineers, it is not our job to develop the "intelligence" of the system, but to take the groundwork of other specialists and essentially create that in software. I vehemently disagreed, as I am (and have been) a systems engineer, and it is my belief that I need to understand the system as a whole and design the software to meet the needs of the system.
I was perfectly okay with making this an agree-to-disagree kind of thing, but GUI guy believes that my thinking on this is not only wrong, but dangerous to the longevity of the company as a whole.
I think this is a fundamental difference of view between an application developer and an embedded systems developer. Applications rarely deal with physical systems, they almost entirely deal with abstractions of the physical, or purely virtual systems (like a database) that have fairly static and structured delineations of interfaces. Embedded guys have to deal with the nitty-gritty and dirty details of the real world: things like friction, backlash, nonlinear resistance, boundary conditions, limited resources, and the like.
I've noticed this trend in other places as well. It's really similar to the difference between a physicist and a mathematician (in a rather crude undergraduate sense): a physicist uses math as a tool to try and deal with real-world situations, where a mathematician works in a virtual world that is generally well-behaved. If you've ever taken a physics lab course, you know that the actual numbers that you get almost never correspond to the exact thing that is modeled by the mathematical formulas.
Case in point: terminal velocity. The mathematical formula essentially pits the acceleration of a body due to gravity (9.8 m/s^2) with a function that is based on velocity and area of the object (wind resistance). When the two are equal, the body no longer accelerates, and reaches a steady state velocity. Of course, in practice, the actual terminal velocity changes because of things like air currents, different atmospheric densities, different gravitational fields (distance from the center of the earth, tides, etc.). All of those things can be modeled with a sufficiently complex equation, but those complexities are added to explain real-world ideosyncracies.
It's also reminiscent of the split that I have with a certain brainy type who has a major interest in AI, where I pursue things like emergent behavior as being key to understanding consciousness, and he abhors emergent behavior as an abberation in his otherwise clean understanding of "intelligence". I embrace chaos, he sets his boundaries so he doesn't have to deal with it.
For me, understanding how a system works is crucial to designing the code which makes the system function. Otherwise, I'm just a code monkey.