Re: WEB : Mapping out communal cyberspace

D. Owen Rowley (owen@autodesk.com)
Thu, 16 Jun 1994 12:40:29 -0700


> From: Brian Behlendorf <brian@wired.com>
> On Tue, 14 Jun 1994, Mark Waks wrote:
> > Brian writes:
> > >This topic (grand unified shared coordinate space vs. local coordinates
> > >only) is one which sparked a lot of, er, conversation between Mark
> > >and I at WWW'94. Mark has his CP proposal embedded in his papers on
> > >VRML, but he and I have both agreed that the development of VRML must
> > >come about totally independent of the development of the Cyberspace
> > >Protocol. This topic should be revisited, in my opinion, after we have
> > >achieved the first level of a distributed VRML system.

> > Can it be? Maybe I'm being dense, but it seems to me that navigation
> > is integral to the concept, and probably needs to be integral to the
> > language. I'm not clear on how one can really accomplish anything
> > *without* at least rudimentary navigation. And this question seems
> > to cut to the heart of navigation -- if the language permits a
> > web-like mode of navigation (where any scene can point to any
> > arbitrary other scene), then you're never going to *get* a unified
> > coordinate space.

Then we are both dense, I think you've navigated the topic, and found
precisely the crux of the issue.

> I'm going to indulge in this at length for one post, mainly to show
> why I don't think a shared 3-d registered coordinate space for all of
> cyberspace is needed. Mark, forgive me. :) A CP mailing list/work
> group might be called for, but here's why CP and VRML can (MUST) be
> separate entities.

Indulge away.

>
> Look at it this way - we already have an addressing scheme for the
> net, and that's IP numbers. Now, IP numbers are, for most intents
> and purposes, random - Wired's subdomain is 140.174.72., so you know
> every IP address starting with those 24 bits belongs to Wired. 140.174
> is (one of?) the Class B subdomains for The Little Garden, which is
> where we get our feed from. So, everyone under that shares a common
> IP server, TLG, but *semantically* we don't share much more than that.
> Now, all the machines under 140.*... who are they? ANYone. So
> essentially, IP-number space isn't semantically meaningful.
> 140.174.72.1 could be light-years different from 139.174.72.1.

there is an interesting aspect to cyberspace, many ( perhaps most or all)
of the conventional wisdom we are used to in objective-space, undergoes
a *sort-of* inversion.
In objective-space semanticaly meaningful address based on sequential numbering
MUST be in sequential proximity to facilitate navigation. this is because our
physical self resides in objective space. The vast bulk of cyberspace resides
within the cerebral cortex of the users, and we control the rendering technology
so we are not constrained to the dictates of physical proximity for cyberspatial
navigation.

> Now, I understand the CP proposal (the Labyrinth groups' proposal
> for a shared 3-d address space) to essentially be the same as
> IP-number space, with many times more bits for aligning yourself.
> The argument was made that the main benefit for this is that items
> which are semantically linked together could be positioned near
> one another, and thus "traveling" from one subject to another would
> take you across the spectrum of concepts and ideas encompassed between
> those two subjects. This sounds great, really - and while the shared
> system makes sense from a scalability point of view, it just doesn't
> seem to stand up to a pragmatic view.

Pardon my boldness, but unless you have spent some time in cyberspace,
attempts to delineate whats pragmatic and whats not are sorta..
uh.. not very relevant.

Some of us have been privileged to have opportunitty to spend lotso time in
graphicly represented cyberspaces, and have gained an understanding of this
issue at a visceral level.

as an *observed medium*, having observed it gives one a certain advantage
when describing it, and having not observed it a disadvantage.

> Say, for example, NCSA is the first group to register their address,
> and goes to coordinate 0,0,0. Now, surely there will be other companies
> who want to be hip and will try and align themselves with NCSA, so
> links at 0,1,0, 1,0,0, 0,0,1, etc. are made. Now, do these links
> have volume in 3-space?

No, volume is a measurement of objective space - it measures *real estate*.
data-space is *unreal estate*, and linking unreal estate does not require
proximity of address (proximity of address better not mandate linkage either!)

>- Mark's paper has Wired's web site in the
> shape of a banana. :) Who will do the link/object registration?

the owner of the site.

> Who
> will decide how much cyberspace an object can contain?

In objective space real estate is limited and finite, they ain't making any more
(at least not terrestially) thus has intrinsic value by being

In Data space unreal estate is unlimited and infinite ( theres that inversion!)
we have all you want, thus has no intrinsic value until developed.

un-real estate development is the job ahead of us, and vrml will be the
way we will accomplish the tasks.

>Who will
> resolve property disputes? Will ghettos be created? Will there be
> stratification? See, what I *love* about the web now is that everyone
> is an equal - it is as easy/difficult for me to access Joe Blow's
> World Wide Web Site as it is NCSA's. In the CP proposal, it could be
> VERY hard to get that same level of even-playing-level feel, even with
> wormholes.

thats only because the *net* you play on is anarchic in nature.

no one network or internetwork will serve all needs all the time.
just as we zone objective space we will zone cyberspace.

if your network *home* is your own network, it will really be your castle,
or more appropriately your fiefdom.

LUX ./. owen