Pleasure and Joy

[bennyboy]

So, You have the binary representative of 2 to the (8*64*1000000000) bit value exponent, but those bits in the hard drive are meant to follow a set of rules limiting the diversity of each cell's capabilities to a simple instruction. For example, register the number 6 to variable F. Binary is written in numerous character sets like 8 digit sets (0 or 1 being the possible value for each digit), but the range for a set of 8 is limited to 256 characters. For logical operands in the CPU, they are only as complex as the maker decides to implement. For example, an instruction in Windows could be that which originates of trigonometric operands or even allocating a list to the hard drive or even including integer data from another drive to be operated on in the active drive (the classic "call"). The bits themselves are just placeholders that fit these limited functions as needed. The bits can be ordered together to be utmost complex but at the root, the cell for each bit is a basic little switch. The possibilities are only as deep as the machine language which in itself is simple but frustrating.

The mechanics are limited, but the variety of possible outcomes has no practical limits given an evolutionary approach to learning. In a pure artificial neural network, the programmer wouldn't even need to know what hardware the "behavior" would be output on. The main limitation is actually the ability to expose the system to a sufficient number of trials for it to refine its output.