Popular Science Predictions Exchange

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

~ the obit ~

PPX opened on 11 June 2007 at ppx.popsci.com, a prediction market bolted onto the side of a science magazine. Signing up got you POP$250,000 of house money, with another POP$10,000 if you filled in the survey. Shares ran from POP$0 to POP$100, three new listings came out each week at POP$50 apiece, and a prediction that came true paid POP$100 per share while one that failed paid nothing.

The markets were sorted into six categories, running from Consumer Tech at one end to Extreme Science at the other. The questions were the ones a science magazine would ask: whether Facebook would go public by 2008, and whether anybody would be living in Dongtan, the Chinese eco-city, by 2010. You could buy, sell, short or cover. That is more verbs than most readers of a monthly science magazine had ever needed.

The software was not built in-house. It came from the group behind the Hollywood Stock Exchange, running their virtual specialist application underneath a PopSci skin. HSX had been pricing unreleased movies in fake dollars since 1996, and PPX was that same machinery pointed at science instead of box office.

~ the rap sheet ~

~ how the price was set ~

The interesting part, for anyone who looked closely at it, is that the share price on PPX was not produced by buyers and sellers meeting. It was set mechanically, by counting transactions. More trades in a direction moved the number in that direction, independent of who was trading or what they believed.

That is an exploitable rule and people exploited it. Register several accounts, run trades between them, and you could walk the price of a prediction wherever you wanted it. Scraping the site with a bot made the whole operation faster. None of this required any opinion about whether a touchscreen iPod was coming.

What the magazine had built, then, was a game wearing the vocabulary of a market. That matters because the pitch for prediction markets in those years was precisely that the price carried information: thousands of small stakes aggregating into a better forecast than any expert. A price fixed by transaction volume carries nothing except transaction volume.

~ the hall of fame ~

On 8 May 2009 PopSci announced that the exchange would close on the 31st of that month. There were 33,339 registered accounts at the time. The top hundred traders were given a hall of fame listing, which was the whole of the payout, POP$ having been designed from the start with no door out into real money.

Replay runs out on a working PPX front page dated 9 March 2009, so the closing weeks are missing and the farewell notice, if there was one, is gone. For years afterward the subdomain answered crawlers with 301s and 403s in alternating captures, and at some point it stopped resolving altogether. The logo survives in the archive's asset store: two-tone orange and blue capitals reading PPX, with THE POPSCI PREDICTIONS EXCHANGE set small beside them.

~ reading it from here ~

PPX ran through the exact two years in which the biggest claims were made for the idea that a crowd with play money could be pointed at any question and produce a better answer than the people who studied it. I followed that argument from Turkey in English-language tech coverage and never held an account on any of these things, which in the end saved me from learning the lesson the hard way. What stayed with me is the pricing rule. Tens of thousands of people spent two years trading against a number that did not mean what they were told it meant.

@}-->--

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

« previous  ·  random  ·  next »