I believe his imagined compression scheme compresses variables so that when there are N variables in scope, it takes log(N) bits to refer to a variable, and similarly for methods.
If the source code compressor is well-written (and specialized to the language in question), then the programmer can achieve compression by using the practices mentioned above. itaibn0 has the right idea.
Well, only three of your practices have anything to do with compression.
I believe his imagined compression scheme compresses variables so that when there are N variables in scope, it takes log(N) bits to refer to a variable, and similarly for methods.
If the source code compressor is well-written (and specialized to the language in question), then the programmer can achieve compression by using the practices mentioned above. itaibn0 has the right idea.