A Truth Seal via Predictive Markets

A recent inquiry had me looking at the detection and analysis of 'disinformation'.  Much in and of the news.  That led me to work by Rick Hayes-Roth, who was the CEO of an AI company called Teknowledge that I have written about here before (See tag) .  Teknowledge was one of the AI giants in the late 80s. We used their capabilities, even bought a major equity position with them.  Did some great things, but alas, they are no more.

Around 2011 Hayes-Roth and his colleagues came up with an idea called 'Truth Seals', related to the and driven by Predictive Markets, where you could deliver some measure of the validity of information.  Now needed more than ever.    The startup existed until at 2014 and then folded.  But in my research I noted that Rick, now Prof Emeritus at the Naval Postgraduate School, had written a document that did a post mortem, ala After-Action-Review (AAR) that covered what was done.  Very informative.  Also points to some of the intellectual property developed.    Useful for anyone thinking about the topic.    Reviewing.

I also notice that the idea of 'Predictive Markets' is far less talked about recently, any pointers to work still going on there?

Forecasting Tournaments

Interesting term, similarity to predictive markets and wisdom of the crowds.  We ran some models in this space.  Government reactions to use of related methods,  by Mark J Perry

US Moves to Become Top Global Energy Producer

From MJ Perry:  A visualization and view of data about real markets in action.  US Moving quickly to be top global energy producer.  Not aided by, but often in despite of government investments, lack of investments, and over regulation.

Organ Donor Compensation

MJ Perry examines the case of the US, where organ donor compensation is illegal, and Australia, where it is not.    For the case of human kidneys.  The difference produced by the efficiency of the market is remarkable.

Thinking Through the Forecast

I just caught this article in the Insiders Group blog and thought I would comment about our demand forecasting experiences and how they can be applied to the midsize business.

About the sales forecast.  Or for that matter any forecast.  In my own experience this has yet to be solved generally.   We worked for years aiming to perfect the methods used.  It is often not so much about the analytical technology you can apply, but the number of contextual influences you can include.   Can you include the influence of promotion, of the economy, of competitor activity?  Depending on the industry cyclical and the changes in fashion are also important.  The article states:

" ... Instead of relying on gut feelings and hope when forecasting , top-performing companies in Aberdeen's research are 46% more likely than all others to perform regular sales pipeline modeling and simulation exercises.

On a tactical level, predictive analytics can cut down on the end-of-cycle demands for C-level support to close deals that their reps claim are "THIS close to the goal line!" In reality, there are only so many opportunities that merit high-level help, volume discounting, and the other forms of late-stage motivators.

Accurate forecasts have benefits across the business. For example, the folks who run purchasing, inventory, logistics, supply chain, operations, and even human capital management, can dramatically benefit from realistic sales forecasts that helps them more efficiently plan for their own activities post-sale. .. ." 

Good thoughts,  and I will add a few.  First is that you need a single set of forecasts, so that all the company is being driven from the same numbers.  You need to frequently calibrate the numbers, as the context of your markets change.  You also need either a corporate economist, or access to good econometric models.  This last point has changed radically in the last few decades.  When I arrived at the enterprise we had a room full of corporate economists.  When I left we had none, and had outsourced the entire econometric process.  This created several problems in our ability to deal with  changes proactively.

What does this mean for the small to midsize company?  The forecast is very important to the Midsize, even more important because minor changes in forecasts can severely hurt the small business.  A close linking between true business process and forecast is also important.  A forecast should be a key part of the business process model, so it can help direct next steps and cautions.  Make sure there is a business model, and you know where it links to sales and demand forecasts. Make the forecasts count, and continually re calibrate them.  Any technology choices should support this process and be clear to executive using the results.

This post was written as part of the IBM for Midsize Business program, which provides midsize businesses with the tools, expertise and solutions they need to become engines of a smarter planet. I’ve been compensated to contribute to this program, but the opinions expressed in this post are my own and don't necessarily represent IBM's positions, strategies or opinions.

Consensus Point Prediction Markets

Robin Hanson's Prediction markets blog.  He has a post on the GE Healthymagination Challenge crowdsourcing effort.  Plus lots more of interest that is worth following.  Hanson is chief scientist of Consensus Point.  We examined prediction markets in the enterprise, but not their offerings, worth a look.

Unilever Exits BrainJuicer

Unilever has sold off its position in online research agency BrainJuicer, after an usually lengthy eight year investment.  A quick look at BrainJuicer seems to make them be an agency using  alternative methods like neuroscience, sociology and crowd sourcing.  In particular see their predictive markets paper.

On Prediction Without Markets

Good overview and link to the original Cornell paper, here an abstract of that:

' ... Though theoretical and empirical evidence suggests that markets do often outperform alternative mechanisms, less attention has been paid to the magnitude of improvement. Here we compare the performance of prediction markets to conventional methods of prediction, namely polls and statistical models. Examining thousands of sporting and movie events, we find that the relative advantage of prediction markets is surprisingly small, as measured by squared error, calibration, and discrimination. Moreover, these domains also exhibit remarkably steep diminishing returns to information, with nearly all the predictive power captured by only two or three parameters ... '