Re: OC Register

From: Adam L. Beberg (beberg@mithral.com)
Date: Thu Feb 22 2001 - 22:17:14 PST


On Thu, 22 Feb 2001, Gregory Alan Bolcer wrote:
> The problem is, centralizing stuff is hard. P2P lets you do the
> easy stuff first with only centralizing the metadata and leaves the
> harder problem of migrating the rest of the stuff for later.

OK Greg, now I know you're on drugs. Running a centralized system is orders
of magnitude easier then running anything with decentralized elements to it.
IT managers wake up in the middle of the night from nightmares about this
stuff for a reason.

...

On Thu, 22 Feb 2001, Gregory Alan Bolcer wrote:
> > wild idea - isn't it more realistic to say it goes up to:
> > benefit = users * [avg users "buddy list" size on protocol]
> > My "buddy list" is about 1-2k people - friends, associates, and lists -
> > spread across about a dozen protocols - aim, icq, email, majordomo, etc.
> > My benefit is limited to the number of people in my world using the product,
> > certainly not exponential. Of course, if you consider the protocol TCP...
>
> They typically measure network effects, not by actual connections
> but *potential* actual connections. You're just assigning a chance value.
> I bet you don't believe the 6 degrees theory either. 8-)

Nope, it's more like 4.5 degress now they say..
Standards would be a really nice thing to push up the "buddy list size" in
my more real-world formula. That is a much better goal I think.

> > > "You're using the distribution power of the network without losing
> > > control,"
> >
> > Ah, the holy grail... *chuckles*
>
> It's all a matter of what you centralize and what users
> allow to be centralized willingly. The holy grail is
> decentralized distribution with centralized payments.

Yea, you find that grail you just let us know ok ;)

> > Not conceiveable, provable possible and usually easy. Why is everyone
> > treating this like it's a solveable problem? Oh wait, i'm not supposed to
> > say that am I...
>
>
> > Has anyone? Everyone is too busy trying to lock everyone into their
> > proprietary protocol and namespace because THAT is how you make money.
>
> You smell that? That gasoline smell of Adam torching P2P. Throwing
> a match on the whole industry. Smells like...
> victory. Someday this infrastructure war's gonna end...

What industry? That implies someone is willing to pay other then VCs.

It will end as soon as something comes into the public domain (no, i dont
mean "open source"(tm) ) that works and doesnt require central servers that
cost money to run... you are looking for a cup, I'd suggest going for an
ordinary looking wooden one. That or some genetic rewrites that remove all
the self-interest genes from humans :)

In the meantime, we'll just have to end it another way... *grin* soon...

> Nothing wrong with piggybacking information, content, whatever
> on the incoming status of an outgoing HTTP call.

Nothing at all, as long as you have a nice convienient SERVER somewhere to
bounce off of with a stable address. Well drat, that would cost money, oh
and that would be client-server... double-drat... damn 50 years of computer
science for being right all along!

*LMAO*

P.S. anyone else going to the Spring 2001 Internet2 Member Meeting?

- Adam L. "Duncan" Beberg
  http://www.iit.edu/~beberg/
  beberg@mithral.com



This archive was generated by hypermail 2b29 : Fri Apr 27 2001 - 23:18:20 PDT