Hubdub: Predicting tomorrow’s news is fun!


Photo by Brian Solis

According to their about page, “Hubdub is all about letting you actively participate in the day’s news, following today’s events, predicting their outcomes and competing with other users to find out who the real news junkies are!”

Hubdub harnesses the cognitive potential of collective intelligence, which truly sets it apart from the other social news sites such as Digg, Reddit, or Mixx. I was pretty excited when I started playing with it last month. Still I wonder, how accurate are the predictions? Will it still be useful if the forecasts end up different from real outcomes?


Types of Collective Intelligence

As seen in The Wisdom of Crowds (2004), James Surowiecki argued that the aggregation of information in groups resulted in decisions that are often better than could have been made by any single member of the group.

His book started with an anecdote relating Francis Galton’s surprise that the crowd at a county fair accurately guessed the weight of an ox when their individual guesses were averaged. In fact, the average was closer to the ox’s true butchered weight than the estimates of most crowd members, and also closer than any of the separate estimates made by cattle experts. Several more cases were shared, with the point being that even non-expert opinions would lead to better predictions or outcomes.

A few years ago, when the Singapore political blogosphere was just taking root, I remember discussing an idea with Elia Diodati about creating a web service which would aggregate and poll blog posts to determine which side bloggers leaned on regarding particular issues. During that time, memetrackers like Tech.Mememorandum and Megite (listen to theorycast.04) started taking off, and I saw this as a step up to that plate. I believed that unlike Digg, the process of getting opinions may have to be automated through aggregation and clever content analysis, otherwise it might be tedious for participants to consistently submit votes.

Perhaps this is why for Hubdub, how it gathers user predictions is mechanically delightful…

First, it pulls in news stories relevant to the posted topic (think Google News). Then, much like how the stock exchange works, you give your opinion by selecting the likely outcome (think Amazon’s Mechanical Turk), and finally bet on the level of certainty you have, where the more certain you are, the higher you “bet” your Hubdub dollars (i.e. risk element in games). When enough people throw in their votes (even non-experts), the aggregated result should give us a good sense of what would actually happen.

As you can guess by now, hot topics on Hubdub include the presidential elections, such as “Will Obama drop out of the race before 30th April?” (86% say “No”), or “Who will be the next President of the U.S.?” (39% say Barack Obama). Other than politics, Hubdub also features topics covering sports, entertainment, world news, business, technology, and science, which are quite active as well.

It would certainly be interesting to see over the course of time how much of the predictions actually fall through. A study wouldn’t be too difficult, though as Hubdub claims, it is still a young service building a track record of the accuracy of its forecasts. One thing I agree do agree with Hubdub founders is how in similar prediction markets (both play and real money), we’ve seen strong relationships between the forecast chances and reality.

For now, if you’re tired of simply voting up articles on regular social news sites, give Hubdub a run. Even if it’s a while before it’s give us accurate forecasts, it’s certainly a new breed of entertaining yet intellectual engaging news service.