...
I will add one conjecture to the list, though it is not something I ever confirmed with 't Hooft. It is possible that at that time 't Hooft knew the sign of the beta function but not its coefficient. His calculations employed dimensional regularization and dimensional subtraction. From these he would have known the sign of the renormalization constants. However, the fundamental definition of the beta function makes reference to the response of the theory to scale transformations. Dimensional regularization introduces a scale in a subtle way -- when one analytically continues away from the superficially scale invariant dimensions. How the traditional renormalization group is represented in this context is something that was worked out only a couple of years later.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...
a reasonable approximation to the relevant Chinese character
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.