I'd like to have proof that Chaitin's constant for lambda calculus and/or SKI combinatory logic is pretty small. I've found some approximations (accurate to about 63 binary digits) for truing machines and as expected they were low, below 1% in fact. I can't find any bounds for lambda calculus or combinatory logic. I'd really just like to show that it is less than 10% to 20%. Does anyone know of such a paper or how to get such an awnser?
I also found this but it doesn't have a reference for how it got that answer. If correct it would show that the probability is between 0.125 and 0.0625 which is great for me! Does anyone have a citation for this?