In pure functional programming, so well mentioned in some PG's articles, you have only functions - no numbers. You can make numbers from functions, though. Then, you may have a basis - a set of functions which are sufficient to represent arbitrarily complex algorithm, i.e., Turing complete set; see here - <a href="http://dkeenan.com/Lambda/" rel="nofollow">http://dkeenan.com/Lambda/</a> .<p>Such a set may consist from only one function. So, the whole program is a structure of calling that function. Writing down such a program, we obviously don't need to write the function - it comes by default; we only need to write the structure, which can be represented by ( and ) . In a different way, than what I see here, but still - one bit per parenthesis, very simple processor, lots of memory and lots of steps for simple operations - a Turing tarpit...