Amplicate

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

~ the obituary ~

Amplicate had one question and asked it about everything. Do people love this or hate this. It read what people were posting on Twitter and elsewhere, decided which side of the line each statement fell on, and kept a running count per topic. Its tagline was Making Your Opinion Count, which is a pun it earned, since counting was literally the whole product. By the end it claimed 450 million opinions over 1,160,700 topics, and the front page rotated through them: Barcelona, flying, Jet Li, Reba, Netflix, beef, Mesa, whatever was moving that hour.

Juan Alvarez founded it in 2009. The press it got was mostly from the business side, because the obvious customer was a company that wanted to know how much its own name was hated this week. PCWorld wrote it up in 2010 for tracking love and hate of enterprise software vendors, which gives you the flavor of the pitch: sentiment analysis sold as a scoreboard, in an era when every startup deck had the word social in it twice.

On 26 February 2013 the company posted that Amplicate was "evolving" and "merging efforts with a complimentary project," spelled exactly like that. In March it was acquired by Grand Central Holdings. No merged project was ever announced under either name, and the hundreds of millions of opinions the site had been collecting since 2009 went with it.

~ the case file ~

~ love and hate, nothing else ~

The binary was the design. There was no four-star rating and no comment thread to argue in. A topic page told you how many people loved the thing and how many hated it, and you could add yourself to one pile. It was crude by the standards of the sentiment tools that came later, and it was legible in a way those tools are not: you could look at a page and immediately know what it was telling you.

Crude also means wrong a lot. Sarcasm reads as praise to a counter, and a company's own employees post about it more than its customers do. Nobody at the time pretended otherwise, which is almost refreshing to read back now, in a decade where the same guesses get shipped with a confidence interval attached and called a model.

~ the year it took to actually die ~

The announcement was February 2013 and the site did not go anywhere. It kept serving normally through the spring, with the last healthy page on 14 June 2013. From 24 July onward it returned 502 errors, week after week, into the following winter. Then on 6 January 2014 it came back, fully working, topic carousel and all, and ran for nearly four more months. The last capture of a working Amplicate is 27 April 2014. By 27 June the domain did not resolve at all.

A site that answers for sixteen months after its own goodbye post is not being maintained, it is being forgotten in stages, probably on a cloud instance somebody kept paying for and then did not. The ArchiveTeam page for it flagged the practical problem at the time: there was no export. If you had been adding your opinions since 2009, the only way to keep them was to copy them into a text file by hand.

~ from here ~

This one never came near me. In 2009 I was in Ankara in my mid twenties, and Turkish Twitter was still small enough that the people on it mostly knew each other; the explosion came later, around 2013, for reasons that had nothing to do with brand sentiment. Amplicate was reading English, and an opinion counter that cannot read your language has nothing to say about anything you care about. I found the name in an archive index seventeen years after it launched, under a note explaining that none of the data got out.

@}-->--

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

« previous  ·  random  ·  next »