I reached some similar conclusions in my overnight reflections on this topic. Basically, even the responses so far to this thread suggest that when it comes to the "basic" chronograph, performance of both integral and modular solutions exceeds the mainstream market requirements. I actually seems that both kinds of designs also exceed the performance requirements of picky folks like the PuristS!
That said, I think that Clayton's principles still apply. I, for one, find it useful to be able to understand the paragraph that I just wrote above in terms of a conceptual construct (that of meeting or exceeding mainstream market requirements as a key driver of modularity or integrality). In addition, I started thinking about the JLC Duometre, for instance. It goes beyond the "basic" chrono requirements with its claim that it reduces the drag of the complication on the base movement, thereby making the timepiece more accurate whether the chrono is running or not. It also has additional complications such as the 1/6 seconds running display.
My initial thought was to characterize the Duometre as an integral design -- improving on specific and fixed performance dimensions through a "carved from a single block" approach -- but I was confounded by the emergence of the moon phase Duometre based on the same "double wing" design. I'm not certain whether the majority of the new caliber is identical to that of the Duometre chrono, but let's say that it is -- in this case I'd characterize what's going on as an improvement in the interface between two highly integral modules. To be more specific, it seems that JLC has linked together two separate movements through the extremely simple (and low-energy-drain) mechanism of the stop-seconds display, with a power supply on each side removing the "thick book" requirement that the power for the complication flow through the interface. The chrono or moon phase complications themselves, however, are highly integral in nature -- although within an overall modular architecture.
To me the dual wing design (if I understand it correctly) is further testimony to JLC's ability to pick exactly the right spot to locate its interfaces, and to make those interfaces as simple as possible, thereby making a modular architecture work almost as well as a fully integral one (and at the same time allowing a number of variations on the basic dual wing design via various complications).
To come full circle, I agree with you that in areas that have been "perfected" a long time ago, and in which modular designs exceed market requirements for core performance, the choice of modular vs. integral becomes a matter of personal taste and individual definition of "purity" of design (vacuum tube pre-amps come to mind here...). Still, I very much like having explanatory frameworks like Clay's that probe behind the surface of things to allow us to understand why, for instance, many of the PuristS polled so far seem fairly indifferent to integral vs. modular or why the Duometre design improves performance on certain dimensions without resorting to a fully integral approach.
Perhaps enough on this topic from me for now! Thanks for your interest in the topic and for asking my views.
Best,
Gary