Maybe Gary has changed his tune on code in general. When I have talked to him in the past, his comments were always along the line of "operators don't really care about code the way you guys do" (before "enthusiasts" became an official market segment for him).
However, BM66 may be in a chicken and egg situation - if the game isn't selling well (I have no idea how well the Premiums are selling), how do you know if it's because of code status? What evidence is there that says major game sales are tied to the robustness of the code?
BM66 may be in a special situation - certainly everyone on Pinside or anyone that went to Expo should have known the game was released early with alpha code. But that didn't matter to those interested in an LE or SLE. But if you don't care about one of the limited editions, why would you buy a game early/now if you are concerned that it may never be "finished"? No reason to take that risk since the game is likely to be in production for three years or so. I had the same thought process - I was very interested in the game, but if I couldn't get an LE, I was going to wait and see.
I do think sales will increase as the code matures, but I hope sales aren't crippled just because the game has an early ding for alpha code. If Lyman does what we all hope he will, BM66 will turn out to be a sleeper hit.