WinMX

» see winmx.com as it lived, on the wayback machine

~ the last words ~

WinMX showed up in 2000 as an OpenNap client. Napster was already in legal trouble at that point, and a lot of users were looking for the next thing. WinMX was that next thing for a while, until May 2001, when version 2 introduced Frontcode's own WPN protocol and a peer cache system that Frontcode itself ran. That decision is what made the September 2005 shutdown so quick. The peer cache servers were the choke point, and Frontcode owned all of them.

By 2005 WinMX had about 2.1 million users and was the most-used source for online music for a stretch of that year. It was the program you opened when you could not find a track on Kazaa, or when Kazaa's results were 90% fake files with the wrong bitrate. WinMX had longer queues but better hit rates. The chat rooms inside it were also a quiet weird community of their own, full of people trading album rips and arguing about MP3 encoders.

I used WinMX during the early university years on a 56k modem. The math was brutal. A 4MB MP3 took roughly 15 minutes if the source had a decent upload, and the queue position was usually somewhere between 40 and 200. You would line up six tracks, leave the computer overnight, and check what survived in the morning. By 2006 ttnet's first broadband packages arrived and WinMX suddenly felt fast, which is funny because by then the network was a community-patched ghost of itself.

~ the record ~

~ how it ended ~

The shutdown happened over about a week. Frontcode received a cease-and-desist letter from the RIAA on September 13, 2005, demanding either content filtering or full closure. Eight days later the homepage and the peer cache servers went offline, and the official client could no longer connect to anything. Two days after that, community developers started releasing patches that redirected DNS lookups to community-run peer caches, and the network limped back up. By 2008 there was a unified "WinMX Community Patch" that maintained the infrastructure on volunteer hardware. The traffic by then was a fraction of what it had been, because every active user had already moved on to BitTorrent or to private trackers.

~ the funny part ~

Frontcode never released the source code. The community kept the network alive entirely through reverse-engineered DNS redirects and cache replacement. It worked, technically, but the original company's silence after the shutdown was so complete that the community patches felt like maintaining the corpse of a friend who had never sent a goodbye message.

~ the legacy ~

WinMX's peer cache architecture is one of the cleanest case studies of why a P2P system with a centralized coordination layer can die in a single afternoon. BitTorrent learned from that lesson, mostly. Trackers were the equivalent choke point, and the move to DHT-based torrents in 2005-2006 was partly a response to seeing what happened when one cease-and-desist letter could turn off a 2-million-user network. The DHT generation was the last one to have to learn this directly.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »