Trove

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

~ the obituary ~

For about a year, somewhere around 2012, the Facebook feed filled up with lines about what other people were reading. Not links they had shared. Articles they had merely opened, announced automatically on their behalf, with the name of the app attached to the bottom. One of the apps generating those lines was the Washington Post Social Reader, and the team that built it is the team that later built Trove.

I never installed any of them, and almost nobody around here did, since the app was tied to an American newspaper. The output was unavoidable anyway, because that was the design: the reading was the broadcast. Facebook's Open Graph in 2011 let an app publish your activity without asking each time, and news publishers understood immediately what that was worth. The Social Reader got to 17 million monthly users on that mechanic. Then Facebook adjusted how those stories surfaced in the feed and the number collapsed, which is the standard ending for anything built on somebody else's distribution.

The lab behind it was set up in 2009 as Washington Post Labs, the Post Company's own engineering group. In 2010 it bought a personalized news service called iCurrent that became the underlying technology, and the first thing carrying the Trove name appeared in April 2011 as a web product.

Then the ownership got strange. In August 2013 Jeff Bezos bought the Washington Post newspaper for 250 million dollars, and the Graham family, who had owned it for four generations, kept everything else and renamed the parent Graham Holdings. The software lab was one of the things they kept. It was renamed Trove and pointed at building a consumer product with no newspaper attached to it at all.

~ the case file ~

~ who it was actually for ~

The relaunched app landed on 22 January 2014 on iOS and the web, with Android that July, run by Vijay Ravindran out of Washington with a second office in San Francisco. The premise was that you would build named collections, called troves, around subjects you cared about, and that following a person's trove beat following a publication, because a person filters and a publication just publishes.

CNBC described it as a Pinterest board for news and the New York Times called it a treasure for news junkies. Both are accurate and both are the problem. News junkies were not a growth market in 2014, and everyone else had already handed this exact job to Twitter, which did it worse but did it in a place they were already standing.

~ what it left behind ~

The features it shipped in its last year read like a company trying doors. Smartpick, in January 2015, read your Twitter timeline and sorted the linked articles into categories for you, which is a genuinely good idea and was also an admission that people were not going to curate anything by hand. Conversations followed in October. There were builds for the Apple Watch and for the Amazon Echo, which is a great deal of surface area for an app that still needed one thing to work.

In December 2015 Graham Holdings folded Trove into SocialCode, an advertising technology business it also owned, and that was the end of it. Nothing was sold and nothing was rescued. The team went to work on ad tooling.

The idea did not die with it. Every feed you read now is ranked by what people like you engaged with, which is the Trove premise implemented by a machine instead of by a named human you chose to follow. The version where you pick the person turned out to be a hobby. The version where the system picks for you turned out to be the industry.

@}-->--

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

« previous  ·  random  ·  next »