Hubdub.com is a platform and community that combines the value of user-generated news aggregation with passionate prediction markets. It tracks only the stories that people decide are worth creating a market around. Successful predictors make the leaderboard in topics such as Technology, Politics and Sports, which they can link back to their blogs, sites or profiles on social networks.
Hubdub was one of the most popular prediction market sites, but the site closed in April after a 2.5 year run. However, some Hubdub fans are in the process of developing an open source substitute at news-guru.com. Intrade remains the most popular real-money prediction market.
I used Hubdub for a few months and I have a long-standing fascination with prediction markets and collective intelligence (so much so that I am considering starting a site with a similar aim myself), so I was eager to ask Hubdub's founder, Nigel Eccles, a few questions. Nigel graciously agreed to respond, and his answers are below. Thanks, Nigel!
---
Q. Is it correct that your reason for closing Hubdub was to turn your scarce attention to FanDuel which was performing better and you felt had more potential? If so, my question is why – broadly or specifically – do you think Hubdub did not perform better and why do you think it did not have more potential?
A. That is correct. Lots of people loved playing Hubdub but its only means of monetization was through ads. You need millions of users to make an advertising business work and Hubdub didn't have that.
Q. Can you give me a sense of the ongoing costs to run Hubdub? What were the primary cost drivers? Do you see any reliable way to reduce these costs?
A. The main direct costs were category editors (about $1,500 per month) and server costs (about $250 per month). On top of that there was application maintenance time. You could definitely reduce the category editor costs if you use a community model.
Q. What worked (surprisingly) well with Hubdub? In other words, what features or improvements ended up being really good decisions?
A. Leaderboards, commenting
Q. In hindsight, what were the major things you would have done differently?
A. Related news stories never really worked that well. I would have done it on Facebook given another go.
Q. I assume you would be willing to sell Hubdub. What in your estimation is a reasonable asking price?
A. There are significant packaging and transaction costs for us so we have set the price at a minimum of $100,000.
Q. I'd like to get your thoughts on the role of money – real vs. fake vs. none – in predictions. In your estimation, how problematic are the following disadvantages resulting from requiring betting (with either real or fake money) on predictions?
Quibbles over the truth / starting values / appropriateness of question
Easier with play money
Paying people to moderate
With play money you can use a community model
Inability to bet on questions which don’t have clear answers (e.g., how many barrels of oil were spilled in the Gulf?)
Can do with play money. You could use something called a Keynsian beauty contest for that.
Reluctance to bet on outcomes which will not happen until far in the future
No easy solution to that with play or real money.
Reluctance to bet on questions which are especially hard to answer
Again no easy solution
Q. Do you think the inaccuracy of using no money is worse than the combination of the above disadvantages? In other words, I am wondering why you think money (real or fake) is better than no money.
A. You need to have some currency otherwise users do not care how they predict. Having a currency drives the accuracy. There is a paper which compares accuracy between real and play money markets. You should Google for it. [I believe he is referring to Emile Servan-Schreiber's paper.]
Q. Any other thoughts or comments on Hubdub or prediction markets more generally that you'd like to share?
A. I think prediction markets are still very interesting however I think their most valuable application is within narrow fields not horizontal markets like Hubdub.
