Sphere

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

~ the goodbye ~

Sphere was founded in San Francisco in 2005 by Tony Conrad as chief executive with Martin Remy as chief technology officer and Steve Nieker as chief information officer, with Toni Schneider advising. The product was blog search, which in 2005 was an unsolved problem. Web search ranked pages by who linked to them, a signal that accumulates slowly, and blog posts needed to be ranked within hours of being written by people with almost no inbound links at all. Sphere's answer combined semantic matching against the query with separate authority scoring for the blogger, and the public site was running by 2006.

The competitive picture at launch was thin, which is the part worth remembering. Google had no blog search product yet. Technorati had the category more or less to itself and was already struggling with its own index. For about two years there was a real race to be the search engine for the part of the web that updated constantly, and Sphere was in it.

The piece that got noticed was Sphere It, a button for the browser toolbar. You would be reading a news article, click it, and Sphere would run a semantic analysis of the text on the page and hand back blog posts discussing the same thing. It needed JavaScript enabled, which in 2006 was still worth saying out loud. The same machinery was sold to publishers as an embeddable widget, so blog results appeared contextually inside news stories on sites that had no blog index of their own.

That distribution business is what AOL bought. The acquisition of Sphere Source Inc was announced on 14 April 2008 at a price reported at upward of twenty five million dollars, and Sphere was run as a wholly owned subsidiary inside AOL's programming division.

~ the chronicle ~

~ how it landed ~

Sphere began as a destination you would visit and ended as plumbing you would never see. The consumer search engine at sphere.com was the thing the company was named after, but the revenue came from being the supplier of the related posts box that sat under somebody else's article.

That box is now on every content site on the internet, and it is mostly paid placement with a thin pretense of relevance. Sphere built an earlier version of it that was actually trying to answer a question about the page you were on. The format survived and the intention did not, which is a fairly common way for a good idea to persist.

~ the aol years ~

In October 2009 AOL renamed the engine Surphace and launched a self-serve version called S4. In November it announced that the AOL News front page would itself be rebranded as Sphere, so the name was being moved off the product and onto a news site. Tim Armstrong, newly running AOL, was against it, and the whole thing was reversed within weeks. What survived the reversal was a redirect: sphere.com sent you to AOL News, and the captures from January 2010 onward show nothing but that.

Tony Conrad left AOL in November 2009 and stayed attached as a special venture advisor, which is the arrangement founders get when everyone wants to part on good terms. In February 2011 AOL sold Surphace to Outbrain, meaning a content recommendation company sold its content recommendation product to a content recommendation company. Conrad's next thing was about.me, which he sold to AOL as well.

~ what it was for ~

By 2006 I had TTNet broadband and was midway through university, which is exactly the window when a blog search engine should have been useful to me, and it was not. The reason is boring and structural. Sphere indexed the English language blogosphere, and the Turkish one it could reach was shallow enough that any query in Turkish returned either nothing or the same four sites.

The assumption underneath every blog search engine of that era was that the blogosphere was a single conversation you could rank. It was really a few dozen conversations sorted by language, most of them invisible to each other, and the English one was simply the one with the search engines pointed at it. Google Blog Search arrived while Technorati decayed, and the whole category emptied out within a few years, which suggests the problem was never quite the one anybody was solving.

@}-->--

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

« previous  ·  random  ·  next »