Prismatic

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

~ the eulogy ~

prismatic was the clever one in a category that kept dying. it used machine learning to rank the web against a model of what you liked, and the people who built it were the kind of engineers who wrote papers, not just apps. i found the idea genuinely interesting from a distance and never made it part of my day.

bradford cross and aria haghighi founded it in 2010 and opened it to the public in june 2012. the pitch was an interest graph. instead of you choosing feeds, it watched what you read and inferred the topics and sources you cared about, then ranked everything else against that. the tech press treated it as the smart future of news reading, and for a while it was the app engineers showed each other.

the smart future did not pay. in july 2015 prismatic pivoted to selling its interest-graph engine to publishers and finance firms as an api, hoping the technology could earn where the reader had not. that did not grow either. on december 20, 2015 it shut down the consumer apps and the api on the same day, with a note admitting content distribution was a hard business and that it had failed to grow fast enough to keep going. i respected the machine learning more than i ever wanted the product. by 2015 i had watched enough of these readers die to not bother installing one more.

~ the dossier ~

~ the category ~

prismatic died into the same pile as zite, circa, and news360, the interest-graph readers that all bet an algorithm could choose your news better than you could. being the smartest of them did not help. the problem was never the ranking, it was that too few people wanted a machine standing between them and the news, and the ones who did were not enough to fund the servers.

~ what it left ~

the interest graph won everywhere except as a product you opened on purpose. the same idea, watch a person and rank the world against them, runs the feeds people actually use now, inside apple news and the big social timelines. prismatic did the work early and could not turn it into a business. the recommendation machine it was proud of is now just the water everyone swims in, uncredited.

@}-->--

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

« previous  ·  random  ·  next »