Loading ...
Sorry, an error occurred while loading the content.

Re: [decentralization] Is P2P useful? or Hype?

Expand Messages
  • Seth Russell
    ... Another possible answer comes from the wildly successful implementation of the hUmAn2000 which adapted decentralized design early in their evolution. --
    Message 1 of 3 , Jul 20, 2000
    • 0 Attachment
      Lucas Gonze wrote:

      > As an industry we are obviously fad-driven, and even with all the
      > hype
      > about P2P, it is not clear that it is useful for a broad range of
      > problems. Clearly it is applicable to political filesharing
      > applications like Freenet or Gnutella. It is less clear that we
      > really
      > need it for anything else. So the question is: why bother?
      >
      > One possible answer comes from Adriana Iamnitchi's talk at Twist
      > 2000
      > [...]

      Another possible answer comes from the wildly successful
      implementation of the hUmAn2000 which adapted decentralized design
      early in their evolution.

      --
      Seth Russell
      http://robustai.net/ai/word_of_emouth.htm
      Click on the button ... see if you can catch me live!
      http://robustai.net/JournalOfMyLife/users/SethRussell.html
      Http://RobustAi.net/Ai/Conjecture.htm
    • Dave Winer
      I believe I can answer that question, without appealing to any fads. It s the joy of running a server. (Without the pain.) The first joy is watching the hits
      Message 2 of 3 , Jul 20, 2000
      • 0 Attachment
        I believe I can answer that question, without appealing to any fads.

        It's the joy of running a server. (Without the pain.)

        The first joy is watching the hits come in and wondering who all those
        people are.

        In Napster the hits don't come in all that fast, it takes a long time for
        most people to download a song. It's cool because you can really ponder who
        the person is. And even better, you can ask them! (Napster has integrated
        chat, most web servers don't, a shame.)

        I recognize this as a meditation I used to do with WebSTAR. I'd sit down in
        front of the server (it was in a different room) and watch the log scroll,
        wondering who these people were, coming from so far away, or from powerful
        companies, reading some obscure howto on the Frontier site, or whatever.
        It's the other side of writing, watching people surf your site. A very small
        percentage of us have had the oppty to do this, but what a useful meditation
        it is.

        The other cool thing about running a server, that Napster does not address,
        is the convenience of editing content. On my old Mac server (now long gone)
        I used to entertain my friends by opening the home page of my site in
        BBEdit, I'd change a word, press cmd-S, and then refresh the web browser.
        "That's easy!" my friend would say, yes, it's almost magical. Instead of 18
        steps to update your site, or even 3, it's 1. There's no more convenient way
        to edit for the Web than to have a server on your machine. Then it's a small
        matter to have the changes percolate up when you want to disconnect.

        That's at the core of what we're doing with Pike, which is a complete P2P
        development platform. I use it to edit Scripting News, it's still a little
        rough, but we're working on the roughness now.

        http://pikebeta.userland.com/

        Windows and Mac. Lots of stealth features. (It's basically Frontier with a
        friendly UI.)

        Dave


        ----- Original Message -----
        From: "Lucas Gonze" <lucas@...>
        To: <decentralization@egroups.com>
        Sent: Thursday, July 20, 2000 9:35 AM
        Subject: [decentralization] Is P2P useful? or Hype?


        > As an industry we are obviously fad-driven, and even with all the hype
        > about P2P, it is not clear that it is useful for a broad range of
        > problems. Clearly it is applicable to political filesharing
        > applications like Freenet or Gnutella. It is less clear that we really
        > need it for anything else. So the question is: why bother?
        >
        > One possible answer comes from Adriana Iamnitchi's talk at Twist 2000
        > where she compared the results of a project done with a decentralized
        > design with the results of the same project done with a distributed
        > design. The results were sketchy, but indicated that decentralization
        > was a performance loss for small networks (<50 nodes) and a win for
        > larger networks (>50 nodes).
        >
        > [Decentralized vs. distributed in this context means that authority is
        > delegated to worker nodes, instead of being kept in a single
        > controller. Adriana's results say loosely that decentralization starts
        > giving performance improvements at the point where a single controller
        > becomes a bottleneck.]
        >
        > ------------------------------------------------------------------------
        > Failed tests, classes skipped, forgotten locker combinations.
        > Remember the good 'ol days
        > http://click.egroups.com/1/7076/3/_/_/_/964100488/
        > ------------------------------------------------------------------------
        >
        > To unsubscribe from this group, send an email to:
        > decentralization-unsubscribe@egroups.com
        >
        >
        >
      • Lucas Gonze
        I have sent mail to her to see if she will talk about it herself, since my knowledge of the details is limited. For trust and incentive, though, I am pretty
        Message 3 of 3 , Jul 20, 2000
        • 0 Attachment
          I have sent mail to her to see if she will talk about it herself,
          since my knowledge of the details is limited. For trust and
          incentive, though, I am pretty much certain that there was no
          such thing; e.g. it was open trust and no incentivization. More
          to follow, I hope.

          - Lucas

          "Wesley M. Felter" wrote:
          >
          > On Thu, 20 Jul 2000, Lucas Gonze wrote:
          >
          > > One possible answer comes from Adriana Iamnitchi's talk at Twist 2000
          > > where she compared the results of a project done with a decentralized
          > > design with the results of the same project done with a distributed
          > > design. The results were sketchy, but indicated that decentralization
          > > was a performance loss for small networks (<50 nodes) and a win for
          > > larger networks (>50 nodes).
          >
          > For what project? What kind of network? What trust model? What incentive
          > model?
          >
          > Wesley Felter - wesf@... - http://www.cs.utexas.edu/users/wesf/
          >
          > ------------------------------------------------------------------------
          > Failed tests, classes skipped, forgotten locker combinations.
          > Remember the good 'ol days
          > http://click.egroups.com/1/7076/3/_/_/_/964138063/
          > ------------------------------------------------------------------------
          >
          > To unsubscribe from this group, send an email to:
          > decentralization-unsubscribe@egroups.com
        Your message has been successfully submitted and would be delivered to recipients shortly.