>>Fabiooooo! What's he talking about?
>
>I think Yuri talk about a wide security program.
>What is interesting on a conceptual level, but in practice it is not manageable.
>Yuri wants a sort of protection between different programmers.
>To do this you should attach to the variables of rights of use by subroutine created by different programmers,
>( like NTFS ACL ).
>This could generate more control over programming,
>but in my opinion makes development impractical.

It's the mental acrobatics implied that make this impractical in the first place. While we don't expect this kind of (almost) undocumented behavior to change ever, the mere fact that it's security based on the behavior of a rarely used feature makes it suspect.

>Despite the myriad attempts at automatic documentation and meta-languages
>​​for describing the meaning of the software written,
>the reality is that what really works is that the programmer is able to keep in his head.

That's on the same level of complexity as proper machine translation from english (or mandarin or japanese - the highly polysemic languages), or just about anything else that requires so much semantics. As long as there are calculated jumps (i.e. your basic IF, Case, do while, select where &lcCondition, magic numbers) in the code, there's no AI yet that could understand what did the poet have in mind. Its output would necessarily be as incomprehensible as the code it tries to interpret.