HubDub and the failings of their model

I've been having a play with HubDub, a shiny new WEB2.0 prediction market-come-news-service. From the pre-beta hype, this promised to be an innovative news service with startling relevance that would blow one's socks away.

Unfortunately, it is a slightly jazzy Predictify with some fairly obvious flaws. From what I can tell, the model is that users subscribe to news stories they are interested in. They are kept up to date with news stories that they can score for relevance.

That's all well and good I guess for those who have time to wade through all the stories and assess them against one another. However, what most users will probably do is read the top story or two and then reinforce their dominance by scoring them as relevant. So early scorers (which for news is a factor of when the news happened and the time zone of the user) will have disproportionate influence. Worse still, those with a vested interest can invest time and resources in shaping the news, voting up favourable accounts of events and marking down critical news (for instance political parties and corporate interests). Just as search engine technology spawned the SEO industry, so prediction market news will create its own seedy world of Prediction PRs with huge news-skewing farms of prediction market users.

The consequence is that one or two views of a story will be assumed to be canonical with contradictory accounts suffering for this prejudice. The noble aim of wresting control of the news media from a handful of corporations would fast become a new element of the apparatus of their hegemony.

However, even if the system could be corrected to avoid this corruptibility, the other feature of the website incentivises dishonest tactics among its users. That feature is the ability to predict the outcome of news stories and benefit from correct predictions.

Now I'm going to take a little digression (although I won't go in to too much detail about human nature or we would be here all week). I want to talk about some competing conceptions of human behaviour. Firstly, there is a big group of people working in the IT world who are wedded to the OpenSource cause. In a nutshell, this lot get excited by the prospect of what can be achieved when we all work together. That's all well and good but it has to be viewed in the light of what can be achieved when people set out for themselves and make everyone else pay for the results. Egoism lies at the root of the IT revolution in the last 50 years and has contributed immeasurable to the economic prosperity that makes it possible for these hippyish sorts to indulge their idealism. Great, now I sound like a rabid free market economist - anyway, rather than me prattle on about the contradictory elements of my psyche, lets actually look at why this matters.

My point in bringing this up is that these OpenSource nuts who have their sticky mitts all over WEB2.0 stuff, think that people will be happy to share information, opinions and good advice in order to help one another. Laudable aims and there are lots of examples of it working on the internet. However, what a lot of people who are getting interested in the much vaunted 'power of the prediction market' fail to recognise is that 1) This is a market and 2) Its just another betting market - like those that have existed since the year dot.

And this is where HubDub falls down in trying to combine a prediction (betting) market with a news service. Prediction market theory assumes that people are self-interested rational agents - that is why one premise of prediction markets leading to accurate predictions is that those involved are incentivised to be right because they are rewarded with money (or something they value). However, the way to make money in a betting market is to bet on those prices that underestimate the true likelihood of an outcome. The upshot is that a bettor is incentivised to conceal information from others in the market, spread disinformation and over-emphasis the importance of information that makes the outcome he is backing appear less likely. If you get a good tip at the races, you keep it close to your chest and if you get a duff tip, you spread it round liberally. (I'll comment elsewhere about why betting markets might not be good prediction markets due to the nature of bettors.)

Therefore betting is antithetical to accurate news coverage. If you have an interest in a market, you have an interest in no news, bad news and lies. Accurate information is a precious resource and not one you are likely to disseminate freely. Ironically while attempting (presumably) to liberating news coverage from the grip of corporate vested interest, HubDub's model inadvertantly gives everyone their own vested interest to confuse, hoodwink and engage in whatever form of skullduggery necessary to prevent the truth seeing the light of day.