Repost.us

» see repost.us as it lived, on the wayback machine

~ the obit ~

In 2011 a small American company decided that a news article should behave the way a YouTube video behaves. You would not copy the text out. You would paste a snippet of code, and the article would arrive in your reader's browser from the publisher's own server, carrying the publisher's ads and the publisher's analytics along with it. The original stayed canonical in search. Attribution stopped being a courtesy that bloggers either remembered or did not, and became a property of the delivery mechanism.

Repost.us launched on 1 March 2011, built by Free Range Content, Inc., whose chief executive John Pettitt had already founded CyberSource and Beyond.com in earlier decades. The pitch landed at a moment when the web was arguing loudly about aggregation, scraping and the ethics of the excerpt, and Repost offered a technical answer instead of a legal one. Launch partners included the translation outlet Worldcrunch and the sports network FanSided. By 2012 PR Newswire was pushing wire copy through the same pipe.

At its largest the system held 8 million articles from roughly 8,500 participating sites and served them to more than 100,000 destination domains. It ran its own link shortener at s.tt. Both numbers were quoted by the company itself in the notice announcing that it was done.

~ the rap sheet ~

~ what it actually did ~

The clever part was that nothing was ever duplicated. A Repost embed was a live window into the origin server, so the publisher kept the pageview, kept the ad impression and kept the ability to correct or withdraw the piece after it had already spread. A blogger who embedded an article could not strip the byline, because the byline was not theirs to edit.

This solved a real grievance. The complaint that drove the whole aggregation fight of 2009 to 2012 was that someone else was monetizing your reporting. Repost made it structurally impossible to do that by accident, and it did so without asking anyone to sign a licensing contract first.

~ the cracks ~

The flaw showed up on the demand side and it never closed. Publishers signed up eagerly, because being embeddable cost them nothing and might bring traffic. The sites that were supposed to do the embedding mostly did not. A blogger who wants to write about an article wants to argue with it, pull one paragraph and add four of their own. Dropping somebody else's entire finished piece into your page is a strange thing to want, and the people who did want it were usually the content farms Repost was built to make unnecessary.

I was working as an engineer in Ankara through these years, and this was publisher plumbing rather than anything that reached a reader in Turkey. The idea is still good enough that I find it slightly annoying nobody has made it work since. What killed Repost was not the technology failing. It was that a two-sided market needs both sides.

~ the day the embeds went dark ~

The announcement came in late June 2014, and it was unusually plain about cause: "Although the technology worked well, the business model didn't." Logins stopped immediately and no new embeds could be created. The company's own wording set the end date: "On July 31 2014 we will shutdown the Repost api servers and all embedded content will cease to be available."

Because every embed was a live window, there was nothing left behind when the window closed. Articles that had been reposted across thousands of blogs did not degrade into stale copies; they simply vanished from the pages that had carried them, leaving a hole where a story used to render. API calls were redirected to an endpoint that returned errors on purpose, so that callers would fail fast instead of hanging, and the s.tt shortener was kept alive until the last day of 2014 so that old links would still land somewhere. Then repost.us stopped resolving, which is still true today.

@}-->--

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

« previous  ·  random  ·  next »