[exprla-devel] Re: [XPL] Re: [xpl-fog] To the Back Burner

[reid_spencer <ras...@re...>]

--- In xpl-dev@y..., cagle@o... wrote:
Lucas,

Welcome to our little universe here.  It's still very much in the 
quark soup
stage, but there are indications that light will break through any 
moment
now.

> My main working principle is that very large scale distributed
> systems can't be programmed explicitly.  Their behavior is emergent.
> To get them to do useful things you have to look for existing
> behaviors and use them as leverage.  the converse is that you have 
to
> forget about fixed specs except in well controlled subnets.

I'm very much inclined to agree with you on that, although I would 
also
caution that we have an existing paradigm that is nearly fifty years 
old now
that works very much on the opposite approach -- that everything 
within a
system must be designed from the bottom up, that a computer can't 
readily
program itself (although that is essentially what any optimizing 
parser
does) and that emergent behaviors are largely aberations to the 
system,
rather than an expected characteristic of large scale distributed 
networks.
Changing that mindset, while necessary, won't be easy.

Emergent behavior engineering is still very much in its infancy, and 
is
going to require that we start looking at programming as a holistic, 
rather
than a reductionist problem. XML is a big part of that, because it 
inverts
the emphasis that data has.  In a reductionist mode, data is self-
contained,
discrete, a part of an object -- the sum of the data of a system is 
the sum
of the data of its attendant parts.  Reductionist programming works 
when the
domain is small, but by its very nature is incapable of handling data 
that
exists is a part of the environment, unless explicitly made aware of 
that
data.  Holistic programming on the other hand sees everything as 
data, with
programs effectively being producers or consumers of that data -- 
organisms,
if you will, albeit very low level ones, which live within the data
environment. Reductionist components cannot evolve in the face of 
changing
data requirements, and thus have to be sustained by very artificial
constraints.  Holistic data organisms, on the other hand, will likely 
need
to evolve, but that means that the internal structures of those 
organisms
may reach a point of complexity beyond that of people to understand.
Moreover, such organisms themselves may be consumed.

The one difference, however, between a biological and a cybernetic
environment is that the principle constraints on a biological 
environment
are energy based, while those in a cybernetic environment are time 
based.
The implications of this are being worked out even now, but it does 
change
the complexion of the environment considerably, cf. Vernor Vinge, 
among many
others.

> One of those behaviors is data clouds, which already exist in the
> form of usenet messages, chain mail, viruses, etc.  I believe that
> this will not be a problem once intermediate nodes have detailed
> control over what they pass through and why.  Uncontrolled message
> distribution like in Gnutella leads to big messes, but carefully
> controlled message distribution should lead to clouds that are only
> as
> big as they need to be.
Gnutella is actually an interesting model -- it takes advantage of 
the many
to many relationships that the net has to offer. That it is currently 
used
to distribute mp3s is fairly secondarily to the fact that it is a 
voluntary
mechanism for creating an accessible network of resources without 
there
needing to be an explicit concept of source. The refining of this 
concept
will come over time -- as someone who inadvertantly ended up pinging 
every
newsgroup in Usenet once (when Usenet itself was once smaller), I see
Gnutella's notion as a harbinger of things to come. You access a local
cloud, which can then work into ever increasing arcs of distribution, 
one
local cloud at a time.

> Another behavior is the mutation of protocols, because specialized
> vocabularies are inevitable.  Though worldOS nodes prefer their
> native transport and message format, they can also have converters
> for
> other transports and formats.  one immediate application for this
> principle will be an adapter for 7-bit ascii.  A 7 bit ascii device
> can establish a relationship with a node that understands it, then
> use
> it as a gateway to nodes that don't understand 7 bit ascii.
>
> There are a couple ways that WorldOS might be useful to your 
projects
> here.  One is that it's a flexible app server which is not bound by
> counterrevolutionary WWW thinking.  You can plug in transports,
> protocols, and functions fairly easily and not have to replicate 
that
> code all over the place.  Another is that it makes it pretty trivial
> to work with xml.
>
> > Can't WorldOS make it impossible for someone to force me to go to
> their site
> > to buy, when I can exchange freely with whomever I wish and not be
> > interfered with?
>
> you bet!  vive la resistance!
I'm currently working on an essay right now that talks about the fact 
that
we are moving into a paradigm where bartering is not only possible, 
but will
in fact be the primary economic transaction; moreover, that money, 
which is
in essence a reputation marker issued by a government, will end up 
becoming
subservient to reputation markers based upon more fungable standards, 
such
as expertise, sex appeal (think about the rise of cyber-models), 
musical or
literary talent, good deeds, and the like. As we end up entering into 
the
cyber age, the distinctions between the real and the virtual world 
will
continue to blur, I anticipate a point in the not too distance future 
where
there really is no money in the traditional sense, where you can buy 
a weeks
worth of groceries on the basis of a sonnet and a song, and where the
primary assets of a country are measured in the number of cultural 
producers
and not the amount of gold or weapons it has. That's the potential 
for XML,
one that I foresee taking place in our lifetime (and perhaps helped 
along by
this very group).

-- Kurt Cagle



> If there are any problems using the system for your projects let me
> know.  mailto:lucas@g...
>
>
>
>
> --------------------------------------------------------------------
----
> Wrox Wireless Developer Conference, Amsterdam, July 10-12. Choose 
from
> 40+ technical sessions covering application of WAP, XML, ASP, Java 
and
> C++ to mobile computing. Get your ticket to the future today!
> http://click.egroups.com/1/5689/1/_/809694/_/961392229/
> --------------------------------------------------------------------
----
>
> To unsubscribe from this group, send an email to:
> xpl-unsubscribe@o...
>
>
>
>
--- End forwarded message ---