Mercora

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

~ the goodbye ~

mercora sat in a category i knew existed but never installed. by 2004 i was 20 and on ttnet dial-up sliding into early adsl, and the streaming-music window in turkey was mp3 files copied off internet cafe machines and winamp playing local libraries. the idea of streaming from a stranger's home library over the open internet was clever but the bandwidth story did not work for turkish home connections, and the licensed-in-the-US part of the pitch was more or less invisible outside north america anyway.

mercora was announced by srivats sampath in 2003 out of sunnyvale. sampath had co-founded mcafee.com and served as its CEO through the antivirus-subscription boom, exiting after network associates absorbed the consumer business. the mercora pitch used what he had learned about paid subscription mechanics and applied it to music. the twist was that the "streaming" was actually a peer-to-peer transfer from another user's collection, wrapped in a licensing layer that paid soundexchange and the composers.

the client shipped in june 2004 for windows. users installed mercora and pointed it at their existing music folder, becoming one of several sources anyone else could tune into. the interface was a radio dial with genre and mood filters that pulled matching tracks from whoever happened to be online. licensed at the platform layer, unlike napster or kazaa, but architecturally a p2p file-share dressed up as a broadcast.

~ the chronicle ~

~ what it was ~

mercora at peak was a genuinely inventive product. the catalog scaled with the user base because every user's local library was potentially streamable. by 2005 press coverage put the catalog at more than three million songs. that was competitive with itunes at the time and larger than any legal streaming service that ran off centralized servers. the search worked because the query resolver could route a request to whichever online user actually had the track.

in september 2006 mercora shipped a mobile client for windows mobile smartphones, letting users tune into the same p2p radio network from their pockets over 2G data. that predated most centralized mobile streaming by a couple of years. in 2007 the international broadcasting convention gave the mercora radio product a technology award, mostly for the licensing architecture.

~ the decline ~

in april 2006 mercora dropped the paid subscription tier and moved fully to ad-supported free listening, powered by a google-adsense-style contextual ad system it called madwords. the shift was a bet that scale would come faster with no paywall, and that ads would pick up the revenue.

the scale did not come. pandora launched publicly in 2005 with a cleaner one-click radio model and no p2p install to explain. last.fm shipped scrobbling in 2002 and was building a social layer that made the mercora social pieces feel late. the free-ad-supported tier in music streaming was a viable market only if you had tens of millions of active listeners, and mercora had a small fraction of that.

in early 2007 the service rebranded to social.fm with a redesigned profile layer explicitly modeled on facebook. that was a defensive move against last.fm's ad-market position. the rebrand did not move the needle on user growth or ad revenue. the P2P architecture that had been the technical differentiator was also expensive to keep running as user churn increased and the average online population shrank.

~ what got buried with it ~

on 4 august 2008 the register reported that social.fm had closed with no press release, no CEO statement and no user announcement beyond a note on the site. the servers went dark and the domain lapsed shortly after. the small ad market for music streaming in 2008 was already going almost entirely to pandora and last.fm on the US side and to the european incumbents on the other. the mercora structural bet that a licensed P2P radio network could beat centralized streaming on catalog cost did not survive the moment when the centralized players got their licensing and their ad rates figured out.

the P2P-as-broadcast idea itself did not come back. by 2011 spotify had launched in the US on a centralized-cache model that pretended to be P2P internally but never exposed users as streaming sources to each other. the pure P2P radio idea died with mercora and never got resurrected because the bandwidth savings stopped mattering once server-side streaming got cheap enough. mercora had built the right architecture for 2004 bandwidth costs and got outrun by moore's law on both storage and last-mile capacity.

@}-->--

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

« previous  ·  random  ·  next »