Posts Tagged ‘crowd wisdom’

Range voting is cool

Posted: 10 July 2008 in Uncategorized
Tags: , , , , , , ,
3

I have just joined a reading group related to prediction markets and crowd wisdom. One topic that has come up is range voting.  Our current election system in the US is known as plurality voting:  the person with the most votes wins.  This is fine in elections with two candidates, but when there are more candidates it is very bad.  More on that in a second.  Range voting takes a different approach and lets you assign a value on a scale to each candidate.  Let’s suppose the scale is 1 to 10, with 10 being your favorite choice.  The person with the most points at the end of the day wins.  This is much, much better.  So much better.

Bayesian regret is very simply the difference between the best case scenario and what actually happened measured in terms of utility.  The obvious question is “what does utility mean in a voting system?”  It’s not exactly something that’s easy to quantify.  Will I gain financially if Obama is elected?  Will I gain satisfaction if Nader is elected?  Will I lose my mind if McCain is elected?

Warren Smith has this solution:  generate a multitude of random simulations and see which method comes out on top.  It’s impossible to generate every possible permutation of voters in an election of any significant size, so he uses Monte Carlo methods to randomly sample the space.  This has the added advantage of alleviating the need to quantify utility for real people.  Just assign them a random value according to some model of voter mentality.  Smith points out that’s not perfect since these simplified models cannot exactly match real human behavior.  But if you run the simulation enough, it all comes out in the wash.

The result is that range voting beats everything every time in his simulations.  Even when considering strategic voters, range voting performs better against other voting styles with strategic voters.  Strategic voters are those who choose the lesser of the evils that they think might win.  Plurality voting is actually one of the worst methods when it comes to elections with more than two candidates.  So our voting system reinforces the two-party system we have in America.  Go figure.  The high numbers our method gets on Bayesian regret plays out in everyday life.  There are all these rallies to get people out to vote.  Most won’t, since they are disillusioned with the whole process.  Why shouldn’t they be?  It blows.  Read the paper or check out RangeVoting.org for details.

Instead of rocking the vote, let’s change it altogether.  Less Bayesian regret.  I should sell T-shirts.

less bayesian regret

Range voting: less bayesian regret

Predictify!

Posted: 11 December 2007 in Uncategorized
Tags: , , , , , ,
2

I wrote about Predictify a while back. It’s basically a website that pays users for predicting world events. When I first wrote about them, I presented the sample question: “How long will Michael Vick’s sentence be?” Well, the verdict came down and my prediction was very close. I predicted 24 months and the dirty bastard got 23. Total payout for me: $6.07. Not bad for 30 seconds effort.

The site appears to be doing well. Currently, there are 24 open polls with large cash payout potential. I was pretty skeptical it would succeed (and that still has yet to play out fully), but it would seem that the guys who predicted its success are going to be looking at nice payouts of their own.

Predict this!

Posted: 8 October 2007 in Uncategorized
Tags: , , , , , , ,
1

Predict the following:

  • How long will Michael Vick’s prison sentence be?
  • How many iPhones will Apple sell by year’s end?
  • How much will Hillary Clinton raise in funds during the fourth quarter of 2007?

How accurate is the crowd?  Futures markets have done a good job at predicting how the market will go for certain commodities.  This fact inspired DARPA (the Defense Advanced Research Projects Agency) to create a project that sought to build a futures market around terrorism.  People would be able to bid on where and when terrorist events would occur (or be attempted).  The public backlash from this idea was enormous.  Senator Barbara Boxer raised quite the furor and demanded that the then Information Awareness Office director Admiral John Poindexter be fired.  He did lose his job over the affair, but like most Washington insiders, soon found a job on a board of directors.

I actually don’t think the Policy Analysis Market (as it was called) was such a bad idea.  It seemed that the three main critcisms against it were

  1. it would encourage bidders to actually encourage terrorist acts in order to earn payoffs
  2. it’s grotesque and immoral
  3. pure vitriol

As for point (1), I can kinda see where they’re coming from.  That would definitely be a bad side effect if it were to occur.  However, I think the fact that the government knows exactly who is bidding on what, it’s also very dangerous for that person.  Also, the thing had $8 million in funding.  Not exactly winning the lottery considering that that money had to be spread around creating the system and as a payoff for each particular security to be traded.  It was basically an advanced tip service that takes advantage of collective wisdom.  Point (2) I’m also not so sure about, since we’re only talking speculation here.  Is it immoral for me to predict that another soldier will die in Iraq tomorrow?  I’d argue instead that the market, if it did anything at all, would save lives.  And as for (3), the level of outrage coming from Senators Boxer and Wyden at the time (this all happened back in 2003) was disgusting.  You’d think that DARPA had suggested we start using our babies as a food source.

I could see a worst case scenario cropping up if such a market existed, though.  What would happen if this market suddenly predicted that Iran would launch a nuclear attack on American troops in Iraq (or a conventional attack for that matter)?  If the market had experienced success before and gained some credibility, this could be considered a justification for war.  That would be a very bad thing indeed.

So anyhow, after meandering way off point on a 4-year-old digression, let me come back around to my main point:  Predictify.  This is a new website that seeks to capitalize on crowd wisdom by predicting events ranging from the mundane to the presidential election.  Users can predict outcomes for free and premium users can post polls that collect extra demographics from the users.  Certain polls come with prize pots ranging in the hundreds of dollars.

After playing around a bit and making a few predictions, I’ve found it mildly interesting.  There are slight problems with the interface.  For example, if you’re browsing political predictions and submit a prediction, you’re presented with a link to go back to predicting.  This link takes you back to the main page rather than the subtopic you were just at.  I found this to be annoying since I had to go back to browsing the topic.  Also, the top polls on the main page are starting to get slightly spammy.  Predict the outcome of some motorcross race in Las Vegas this weekend for example.  Predict the yearly earnings of some company no one has ever heard of.  Predict how many speed dates SpeedDates.com will have by the end of the year.

I think Predictify will have to do something to prevent this sort of detritis from clogging up the site.  The interesting polls that will bring in the visitors are getting lost in the clutter of these junk pedalers.