i actually just found this very interesting read.
http://history.nasa.gov/SP-4219/Chapter3.html
Per this account, everyone might have been a bit correct. What's clear is that the 38 was designed with laminar non-compressible flow in mind. In the compressible region, both the aerodynamic flow AND the aerodynimic pressures (hence forces) on the plane were not properly calculated.
The only thing not clearly explained (by this article) was why the controls became 'locked' which I assume meant that the stick and evelators couldn't be moved.
Speculation on my part could be the at the shock Normal created by the airfoil settled on the elevator and created and the high pressures which created a force that the pilot could not overcome. OR that the booms and tail section of the plane deflected enough under these (higher than calculated) forces so that the bearing loads, gaps, or other clearances necessary for smooth operation of cable, rod, and hinge mechanisms closed up resulting in the controlls locking up.
Perhaps the controls lock up is related to the aero elastic study quoted, and "tuck under" and dive [brakes] issues were related to the compressibilty as also stated. Also it could be that the aero elastic deflections calculated in later years where compressible loading could be properly calculated SUPLIMENT the P-38 explanations of the 1930s not contradict them.
With respect to quoting Kelly Johnson or others that worked on the program at the time, remember that these are snapshots in time. Often time engineers will propose a theory and come up with a fix. If the fix works, the theory is excepted. But in cases like this one, where compressible Aerodynamics were not understood and could not be calculated for the many different parts of the airplane, the theory of what was wrong, put forth by the engineers (Johnson included),may have been incomplete. If the Aeroeastic deflection was caused by the changes in loading do to the changes in flow dynamics in the compressible region than fixing the flow dynamics with the dive [brake] may have solved both issues without the P-38 team ever knowing the second one was part of the explanation. Suplimantal explanations could be developed years or even decades later as modeling and analysis techniques improved.
Summary: Because all these problems were occuring at the same time, under the same conditions, they may all be correct but should not singularly be described as "the" compessibility problem.