Firstly, I think it’s important to recognize that the lick Richardson is demonstrating is a long sequence of alternate one-note-per-string playing, so what we’d now describe as “double escaped” picking, or what we’ve sometimes referred to as “crosspicking”.
I think what Richardson is experiencing is that the more strings you are crossing between notes, the greater the temptation for an “inside change” (a stroke that attacks the target string in the direction of the current string change) to become a “single escaped” stroke that buries itself between the strings after it hits the target, because it’s very tempting to get lazy and hit the target without initiating any change in slant (or whatever terminology we use now to refer to a change from one “escape orientation” to another). Meanwhile with an “outside change” (a stroke that attacks the target string in the direction opposite the current string change), the change in direction to come back and hit the target can serve as a kind of cue to also initiate the change in slant.
Though as many of @Troy’s interview subjects have demonstrated, a refined “double-escaped” technique can pretty much ignore any question of which changes are “inside” and which are “outside”.
Edit: And the linked CTC youtube video does a good job of explaining why in “single-escaped” picking, any discussion of “inside vs outside” is pretty much moot. It also makes a pretty persuasive case that in intermittently “double-escaped” licks, or what we’ve sometimes called “two way pickslanting” (like the “Paul Gilbert Lick”), “inside” and “outside” picking are more similar than people tend to realize, and the perception that one is more difficult than the other probably boils down to lack of awareness/understanding about the need to change the “slant” (escape orientation). And as @Prlgmnr notes, the potential to execute a lick with swiping can further obscure why one variation seems “easier” than another.
