Dear Philip,
I completely agree with your overarching point that we need consistency to guide our approach to help us gain further insights into the fundamentals of nature. It struck me, though, that when you gave the examples of Maxwell, Einstein and Dirac, these were in a sense completely different situations from the one that relies on arguments pertaining to black holes to derive new consistency-based insights.
In each of the historic cases, there was at least a prospect that the assumptions on which the consistency-based arguments rested could eventually be checked by experiment. What prospect do we have for that when it comes to black holes? The power of the consistency-based arguments we derive from these assumptions is only as great as the consistency of the assumptions themselves. It seems to me that if we cannot check our most basic assumptions about black holes experimentally, then there is a real danger that we could have overlooked inconsistencies in them, and will continue do so. This could then lead us to derive false arguments even though they are consistent with the assumptions. Instances like the recent firewall debate only strengthen my suspicion that this may in fact be the case.
It seems to me that a more reliable way to use consistency as a guiding principle would be to apply it to a situation that has at least a fighting chance to be eventually subjected to experimental test. I agree that it is not easy to find such situations where both GR and QT come into play, but consistency-based arguments based on that kind of a situation would seem that much more compelling.
As you may know, I am also pursuing an idea based on the notion that spacetime emerges from a lower-dimensional analog, namely that quantum theory tells us that pre-measurement states are spacetime manifestations of lower-dimensional objects and that a "measurement" is really the mechanism by which actual spacetime objects emerge out of these. Several people have told me that this reminds them of the holographic principle, although I am myself remain skeptical of that. It is good, though, that you explore the holographic principle from an angle that others have apparently neglected, as hopefully this will increase the chance the whole issue will be more clearly defined. Why do you think is it the case that over the last 20 years, the application of necklace lie algebras has not been taken up by the string theory community?
All the best,
Armin