Optimizations

[ghost]

Consider switching to a CISC-like ISA, rather than a RISC.

Consider adding an internal RAM to the @ symbol

Consider abstracting the running of these creatures; allows for an unbounded number of creatures

Consider adding register modifiers to algebraic Instructions: without these modifiers, these Instructions add and subtract from the value of the target symbol: not as a register, but as the numeric value of the insteuction

New Design

• the environment is an array of numbers called “symbols”
• Each symbol is represented as a bit string
• Two special symbols called boundaries enclose an organism
• Algebraic operations take from the organism’s dump register (contained in the virtual interpreter) unless they’re followed by a special “register operation designatir” symbol
• There exists an operation that spins off a new virtual interpreter. R9 will spin off a new interpreter starting at the address in the current interpreter’s 9th register. It dies if it attempts to interpret an instruction outside the boundaries it originally started in.

Design issues

• No way to choose ram size / register count for child organism
• Child organism starts with nothing in the dump register

[ghost]

Note: the shorter the minimal ancestor gene, the better the ISA (for this purpose)

[ghost]

Also consider adding a “probe” that moves to and evaluates the value of a symbol when that is needed

a) make it so that it takes time to move b) add a way to expel probes

If you add a way to detect foreign probes, and either (a) or (b), then it will be possible to defend against predation

[ghost]

The portions of the simulation that aren’t “part of the world” can be represented through the background. ie the interpreters can be represented at their current interpretation location as a bright background on that symbol. Further information can be encoded through the exact color. For example, the interpreter’s energy level (its dump register value) can be the brightness