Re: [arch-d] Splintering (fragmentation) vs Centralization vs Users

[Hesham ElBakoury <[e-mail]>]

Constantinos (one of the speaker) supports decentralization since
"Centralised systems have built-in security that’s however vulnerable as
they act as single points of attack and failure".

Hesham

On Fri, May 5, 2023, 3:02 PM Brian E Carpenter <[e-mail]>
wrote:

> On 06-May-23 00:16, Toerless Eckert wrote:
> > Brian,
> >
> > When you say "centralization", what do you mean ?
>
> I mean that, to continue your example, there is basically exactly
> one on-line bookseller in the world, which has all the properties of
> a monopolist. How that is implemented in terms of servers, private
> networks, CDNs, distributed data centres, etc. is a separate question.
>
> For continuation of the argument, see Christian Huitema's message.
>
>     Brian
>
> >
> > If one Internet bookseller has one server in Seattle for all its global
> > customers, is that centralized, or is it decentralized because its just
> one
> > out of many equal booksellers with servers on the Internet ?
> >
> > If the same bookseller has millions of servers to buy books all over the
> planet,
> > is that decentralized now, or do you call that centralized, because the
> Internet
> > enabled that one bookseller to eliminate most others ?
> >
> > How can therer even be an answer to these questions without admitting
> that
> > we do have an interest in the higher-layer reasons and implications of
> our
> > protocol work, and that we can not even make technical judgmenets with
> that
> > "use-case" discussion.
> >
> > And i thought we also already agreed that we needed to be more precise
> in terminology,
> > such as attempting not to use centralized/decentralized for everything
> unqualified.
> > E.g.: monopolization was bought up as another good term to distinguish
> cases.
> >
> > Cheers
> >      Toerless
> >
> > On Fri, May 05, 2023 at 09:39:17AM +1200, Brian E Carpenter wrote:
> >> Hi,
> >>
> >> After a little off-list discussion, I have a few more general thoughts
> >> on this topic. (I won't identify the other person in that discussion,
> >> to respect their privacy.)
> >>
> >> I mentioned that some security technology that we develop could be
> >> "dual use", e.g. useful both for privacy and useful for walled gardens.
> >> So perhaps we should be careful when evaluating new ideas that they
> >> cannot be used for undesirable purposes as well as the intended purpose.
> >> If we consider that both excessive centralization and excessive
> >> splintering (a.k.a. fragmentation) are bad things, does a new technology
> >> drive those tendendencies? Could we design it differently to avoid
> >> this?
> >>
> >> Is there scope for IAB guidance to the IETF about what aspects of
> >> protocols, especially security protocols, might encourage or discourage
> >> either centralization or splintering?
> >>
> >> That could be a productive use of the IAB's resources where we might
> >> have some impact. Discussion of wider societal, commercial and
> >> political issues in the IAB and IETF would get nowhere, and in my
> >> opinion is best left to ISOC.
> >>
> >> There's very clearly a 3-way tussle, and that makes all discussion
> >> difficult, especially since each national government has different
> >> goals. ASCII art:
> >>
> >>                  Users
> >>             (freedom of action,
> >>                  privacy)
> >>                  /    \
> >>                 /      \
> >>                /        \
> >>        National          Global
> >>     governments -------- businesses
> >>     (defend or          (capture &
> >>      control             exploit
> >>      citizens &          customers)
> >>      economy)
> >>
> >> Regards
> >>     Brian Carpenter
> >>
> >> _______________________________________________
> >> Architecture-discuss mailing list
> >> [e-mail]
> >> https://www.ietf.org/mailman/listinfo/architecture-discuss
> >
>
> _______________________________________________
> Architecture-discuss mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/architecture-discuss
>