Oh, but we do know - In order to be compatible with the existent languages, it's going to have to look similar to what we have now. It will have to keep 8 bit bytes instead of going wider that's what IBM came up with in the 1950s, and it will have to be a stack-oriented machine that looks like a VAX so it can run C programs. Unicode will always be a second-class character set behind ASCII because it has to look like it runs Unix, and we will always use IEEE floating point with all its inaccuracy because using scaling decimal data types just makes too much sense and we can't have that.