Social structure as genetic encoding

[Michael Allan]

Hi David,

> I remember talking about this a few years back, and I still don't
> get what the problem implementing what you describe as recombinant
> text is. As ever you have given an important and central concept
> that many people over the years have struggled with, a catchy and
> simple title - but since 2005 we have had several systems that
> actually implement all the features you describe for recombinant
> text - a p2p population of text fragment variants in which
> attribution to each fragment is carried and traceable across the
> variations - its called distributed version control (DVCS) - or the
> de facto standard - git.

I described the difference vs. recombinant text here:
http://zelea.com/project/textbender/d/overview.xht#Distributed-Revision-Control
The crucial thing is the genetic structure.  It allows you to transfer
changes that are out of literal context.  Here's a simple example:

   1. Starting population, two individuals:

         D    I live in the United Kindgom.
              My name is David.

         M    J'habite au Canada.
              Je m'appelle Michael.

   2. Translational mutation of D yields:

         D'   My name is David.
              I live in the United Kindgom.

         M    J'habite au Canada.
              Je m'appelle Michael.

   3. Recombination, D' to M yields:

         D'   My name is David.
              I live in the United Kindgom.

         M'   Je m'appelle Michael.
              J'habite au Canada.

A complex example might be a creative literary application in which
authors can cherry-pick characters from one version of the story to
another.  Or variations of plot.  Or atmosphere.  Revision control
systems can't do this kind of thing, because it requires a genetic
structure (abstract context) beyond the literal context.

> > I can't go into details here (this is just a napkin sketch), but I
> > want to say that, technically, this is a genetic structure.
> 
> Explore more? I'm looking at the moment to do a phD in this sort of
> area - so biological / morphological / genetic modelling with
> parallels to political organisation - so any thoughts we can develop
> here could help frame the proposal. ...

Sure.  This should clarify what I mean by genetic structure:
http://zelea.com/w/User:Mike-ZeleaCom/G/p/vohall#PS
Looking at the right side, the black forms appear to be genes and the
coloured contents appear to be alleles.  If that's a valid view, then
it's surprising and strange: surprising because the genetic structure
was deliberately abandoned for Votorola, and now here it returns; and
strange because it returns as a social construct external to the
individual.  I never heard of that before.

But maybe I understand it a little better, now.  Looking again at the
alleles, I see that each is restricted (owing to the practice of
atomic variation) to a sub-population defined by a single branch in
social space.  From there it may issue to become prevalent in the
population (wild allele), or to there it may retreat if already wild.
These two alternatives (wild or unwild) are controlled by the new
selection procedure of "patch relay" (assuming it is adopted).  This
is now a social procedure, no longer an individual one.  So I'm
guessing it all turns on this change.  Alleles are forced to a social
level where it becomes at least possible for genes to appear, too.
And lo, they do appear!

Why?  I guess they appear out of practical necessity.  Genes are an
efficient mechanism for communicating variation.  They serve evolution
by providing long haul channels for recombination.  This is how nature
communicates her designs.  We just borrow the technique for human
purposes.  (Anyway, that's how I understand it.)

Mike


David Bovill said:
> On 17 January 2013 02:20, Michael Allan <mike at zelea.com> wrote:
> 
> >
> > I discovered something strange and interesting.  Recall that Votorola
> > is recombinant text + transitive voting.  In order to keep it simple,
> > I was forced to drop the genetic encoding of the texts.  All I kept
> > was the population structure and text transfers, as explained here:
> >
> > http://lists.thataway.org/scripts/wa-THATAWAY.exe?A2=ind0906E&L=NCDD-DISCUSSION&F=&S=&P=83
> >
> 
> I remember talking about this a few years back, and I still don't get what
> the problem implementing what you describe as recombinant text is. As ever
> you have given an important and central concept that many people over the
> years have struggled with, a catchy and simple title - but since 2005 we
> have had several systems that actually implement all the features you
> describe for recombinant text - a p2p population of text fragment variants
> in which attribution to each fragment is carried and traceable across the
> variations - its called distributed version control (DVCS) - or the de
> facto standard - git.
> 
> I've always proposed using a DVCS combined with p2p messaging as the basic
> infrastructure for distributed LD - a server based implementation is a
> convenience, but structuring the core concepts and data model around a
> distributed set of apps (mobile, desktop and server based) - has a number
> of additional benefits.
> 
> I can't go into details here (this is just a napkin sketch), but I
> > want to say that, technically, this is a genetic structure.
> >
> 
> Explore more? I'm looking at the moment to do a phD in this sort of area -
> so biological / morphological / genetic modelling with parallels to
> political organisation - so any thoughts we can develop here could help
> frame the proposal. In an earlier life I studies the evolutionary genetics
> of the immune system. This time I'm thinking more along the lines of the
> morphology of tissue structures, or modelling the structure of tissues
> (probably the immune system) using topological concepts rather than direct
> spatial modelling. The immune system is interesting because of it's a
> natural fit with agent theory (and therefore Latour), and also because the
> genetics are unusual (the generation of diversity), and would seem to lend
> themselves to a topological analysis of interactions (DAG).