I'm a pretty big fan of computational attacks. I think they are worth pursuing, and I've seen a small number that advance our understanding of the VMS text.
What I see lacking, however, when I look at the history of computational attacks on VMS text, is the acknowledgment that the spaces may not be spaces and the shapes that look like vowels may not be vowels. The possibility of null characters is also often overlooked. Many statistical tests assume one or more of these to be true and don't seem able to break away from this kind of thinking (along with a number of other assumptions).
What I see lacking, however, when I look at the history of computational attacks on VMS text, is the acknowledgment that the spaces may not be spaces and the shapes that look like vowels may not be vowels. The possibility of null characters is also often overlooked. Many statistical tests assume one or more of these to be true and don't seem able to break away from this kind of thinking (along with a number of other assumptions).