Bob, are these raw scores or weighted scores? (I suspect that the weighted value--eheh--of the TPS-panel-to-orbit score would overwhelm any points the underside design might rack up, but then, I'm a layman...)
Also, it seems to me that the canard solution, while proven, requires an additional subsytem that is not required if the ACS performs the task. I'm tempted to call that one in favor of the underside solution on account of the simplicity gains. Am I right to do so?