The Culture of Secrecy, Disinformation, and , Propaganda...
Ray Dillinger
bear at sonic.net
Mon Apr 16 00:13:58 PDT 2001
On Mon, 16 Apr 2001, Jim Choate wrote:
>
>On Fri, 13 Apr 2001, Ray Dillinger wrote:
>
>> In a fully distributed state, the number of elements that have to
>> fail in order to make the system not work is the same as the number
>> of elements in the system. Fully distributed systems (as in plan D)
>> have the structure of water.
>
>Actually that's not accurate, it depends on the particular 'hyper-cycles'
>that the various components rely on in their existance.
>
>Simply because a system is distributed doesn't imply that each 'node' or
>participant is identical and non-unique (which is the problem with your
>water comparison). Even in a fully distributed crypto-anarchy system the
>individual are unique and non-interchangeable.
I was offering a definition for 'fully distributed.' It is possible
for a system to be distributed without being fully distributed. The
existence of structures, priveleges, or roles in some parts which
cannot arise in other parts, or cycles where nodes can be excluded
if other nodes are removed, are possible in some distributed systems.
The subset of distributed systems in which they are not possible, I
call 'fully distributed'.
Napster is an example of a system which is partially distributed.
If it were fully distributed, you could pull the plugs out of the
servers at napster and the users would never notice.
Usenet is an example of a system which is fully distributed. If
all the backbone nodes went down tomorrow, a thousand linux geeks
across the country could work out the news routing software and
could put it back up without them inside of a week. Basically,
they've already got all the software they need to deal with NNTP.
There is specialization, but it is specialization chosen, not
imposed by the protocol, and if the situation required it, the
remaining nodes would simply change their specialization.
>Otherwise the entire 'pay
>yoru way' would fall down, there'd be no reason to pay anything for
>anything. If you were all interchangeable you'd already have whatever
>it was we were talking about exchanging.
Not necessarily. I might have the means to produce it, but not
the time or inclination or expertise. Specialization, among
humans in meatspace, does not happen because people have different
types of hands. It happens because people have choices about what
they can do and they choose to do one thing instead of another.
If the system isn't fully distributed, there are choices about
what role to take in it that you can't make even if you want to.
Napster users couldn't choose to set up their own site as an
indexing node, for example; it was a reserved role. Some of them
might have specialized in rap and others in pop, as a choice, and
they had reasons to trade with one another. But none could replace
a lost indexing node, so it was possible to shut the system down
and put all the traders out of business, by shutting down one
node. A critical error in design, tolerated because they were
trying to set up a "toll booth" at the weak point in the structure.
Government and many other organizations work the same way; there
are irreplaceable nodes within them that, if shut down, spell
the death of the organization. (Hint: it ain't the president of
the US). What some here call crypto anarchy, may well wind up
being only the fully distributed form of government.
>> It is interesting that trees need water, but water does not need
>> trees.
>Bullshit. If it wasn't for the biosphere trapping hydrogen in water it'd
>be gone a long time ago (ie no water).
Water exists in places where no trees are. Consider Titan.
Bear
More information about the Testlist
mailing list