[eu-gene] can a program be generative, interactive, creative?

[Artem Baguinski]

On 21 Feb 2006, at 06:37, Michael Gogins wrote:

> There is a deeper question inside this, whether Turing machines can  
> effectively model natural law. The answer to this is probably no,  
> since quantum theory predicts true randomness in natural phenomena,  
> which lies beyond the ability of any program to compute (see  
> Chaitin's Omega). Finite programs can only be pseudo-random.

Are you claiming to be infinite then?

here you go:

10 INPUT "Enter something: "; X$
20 PRINT "Something: "; X$
30 GOTO 10

finite program implementing infinite random output.

"But", I hear you object, "it's not the program that generates the  
random output, it's the user!"

Well, I just made a system from the parts i had available - one  
human, one BASIC interpreter, one conventional turing equivalent  
computer.

I could substitute parts of the system with something else - e.g.  
write the program in assembly language, read the input from adc  
connected to some noisy electronic cirtcuit, use a computer that  
isn't turing equivalent.

Starting with this simple example (well, I lie, but just for the sake  
of argument) I codeveloped the initial version of Artificial  
Paradises 0202[1] - a virtual machine that treats its various input  
channels (audio, video, raw file system, network traffic, ...) as  
source of data/code for programs running on it. This programs don't  
exist until the VM receives any input. The programs modify the VM's  
memory (including each other and themselves, since VM has von  
Neumann's architecture and data and code reside in the same memory)  
and generate output (visuals, audio, ...). The computer it runs on,  
the VM itself and any program that may run on it are completelly  
deterministic, but because of the inherent complexity of the inputs  
the behaviours generated were hard to predict in the beginning.

After some time though you could learn the potential repertuar of the  
system (after all it only had 16 instructions, including NOP) and,  
although you couldn't predict what will happen next, you wouldn't be  
surprised when it happened.

I dare to say this program was constantly generating something NEW,  
within the constraints of its "world". The fact that the mapping of  
the inputs to the outputs and mapping of the input to the mapping  
itself (somebody told in the thread that programs don't go to the  
meta level? well this one does) is deterministic doesn't mean  
nothing, where is your proof that your mind is non deterministic?

The new behaviours that the AP0202 generates are determined by its  
inputs, just like the paintings are determined by the impressions of  
the artist. Our creative mind would not be such if it wasn't  
connected to the rest of our body. And even together they won't be  
creative if they spent their life in a closed barrel. And just like  
you can't be creative without the environment, the machine needs it  
to be creative (or to simulate creativity, although for me there is  
NO distinction) too.

Hence the wikipeadia's

QUOTE

2. Interactive/Behavioural: music generated by a system component  
that ostensibly has no inputs. That is, 'not transformational' (Rowe  
1991; Lippe 1997:34; Winkler 1998).

UNQUOTE

may sound almost void, if it weren't for "ostensibly".

What that word wants to say is that the inputs are there, but the  
authour doesn't find it iportant to draw attention to them. When we  
call a piece "generative" or "interactive" we just mean that we find  
important certain its aspects (e.g. apparent lack of I->O  
transformation or reaction to spectator's actions), while almost or  
completely ignoring others. It isn't like there's some Department of  
True Meanings of Words which has to approve the use of "interactive",  
"generative", "random" or "conscious".

[1]: http://1010.co.uk/ap0202.html