A possible definition of life.

Peter Bentley ucacpjb at cs.ucl.ac.uk
Fri May 2 13:04:44 EST 1997


> > Hmmm... Suppose I wrote a computer program that was capable of creating
> > another program that was capable of creating another... and so on. Say it
> > uses genetic programming to evolve its children from scratch.
> > This is replication of 'the ability to replicate', and not self-replication
> > (the parent only passes the ability to replicate to its offspring, never any
> > information about itself). This would be replication without the passing on
> > of genes/information about the parent. It would also be replication without
> > evolution.
>
> I thought you said "evolve its children from scratch"?  Sounds like
> evolution to me...and sounds pretty much like what we do with
> computer programmes right now (ie: create them from scratch).  And
> "replication" means to copy: and if you pass on no info, you are
> creating, not replicating.  Pretty much what Big G did, if you are
> into that B-).

Well, that's not quite what I meant. Yes, evolution would be used to generate
the new children. But since every parent would create its children from scratch,
without information being passed on, evolution towards better entities (programs
better able to generate programs that can generate programs...) would not and could
not occur.

> > Would such an entity be alive? If so, then 'self-replication' and 'self-coding'
> > would be superfluous to life.
>
> An interesting proposition...but I suspect not, if it didn't pass on
> anything of itself.
>                      Ed Rybicki, PhD

You seem to be suggesting that life requires self-replicating entities to pass on
information about themselves to their offspring. That's almost saying that all life
must evolve, isn't it?

Actually, upon consideration, my hypothetical entities must pass *something* on
to their offspring: they must pass on the need to become capable of generating new
programs themselves. Uh... this is getting difficult... Let me put it another way:

Life (as we know it, captain) passes on information about itself through the
generations. If a piece of information (gene) increases an entity's liklihood of
being successful, surviving, reproducing, or however you want to put it, then the
information tends to survive, otherwise it is lost.

Might not an alternative form of life (not as we know it, captain) only pass on a
*requirement* that its children must be able to generate *from scratch* offspring
also capable of generating new offspring..? It might be a sort of higher-order,
highly varied form of life. For example, we evolved from scratch. We might create a
clever computer program from scratch, which might create some obscure entity based
around wierd unfathomable maths, which might create...

Get the idea?

Peter.




More information about the Mol-evol mailing list