The primal brain created beliefs as a short cut to survive in the natural world. Broadly speaking, beliefs reside in our subconscious and are formed by our environment, social conditioning and epigenetics. Research has shown that a single belief can produce many complex variables of emotional triggers and behaviors.
HOW BELIEFS ARE FORMED
Belief3 IS A PROPRIETARY LONGITUDINAL BELIEF TECHNOLOGY
Belief3 uses an iterative group surveying technique (similar to the Delphi Method for forecasting) to measure and track a population’s beliefs. Our questionnaire format gathers both quantitative and qualitative data using the important elements of successful collective decision making systems such as those listed below. We surface the nudges (messaging) that will shift the population’s beliefs in a desired direction. We then test and measure the effectiveness of the messaging. We map out demographic and psychometric analysis to reveal which population segments are most, and least, susceptible to influence campaigns. These techniques work for commercial and government applications.
COLLECTIVE INTELLIGENCE SYSTEMS WE EMPLOY
THE WISDOM OF THE CROWDS
Imagine if everyone in your organization submitted a guess about the number of jellybeans in a jar. The wisdom of the crowds says that the average of all the guesses will be close to the actual number of jellybeans, even if no single person guessed correctly. This is because individual error tends to cancel each other out. Some will overestimate and some will underestimate. This collective intelligence can be a powerful tool for solving problems or making predictions, and it's behind familiar successes like the Netflix movie recommender system and Wikipedia. Our CTO, Jen Watkins, studies collective intelligence and the systems that produce the phenomenon. Refer to her chapter Collectively Intelligent Systems to learn the principles that make crowds plus the systems that enable them smart.
Our CEO, John Fuisz, studies prediction error and used his knowledge to become the most prescient forecasting in IARPA’s forecasting competition for the intelligence community.
LAW OF LARGE NUMBERS
The most basic systems that utilize wisdom of crowds rely solely on the law of large numbers.
The law of large numbers says that the more random events you average out, the closer you get to the expected outcome. If you flip a coin, after several flips, you may have mostly heads or mostly tails. But the law of large numbers tells us that the more you flip, the closer the average number of heads will get to 50%, which is the expected outcome. This is why the wisdom of crowds works - with a large enough group, even if everyone's guess is a bit off, by averaging them out, you get closer to the true answer. To make this work as a forecasting technique, the participants need to be independent (guessing without knowledge of other’s guesses) and have some access to relevant information.
The classic example of the law of large numbers comes from a contest at a British fair in 1906. Sir Francis Galton, a pioneer in statistics, observed a competition where 800 fairgoers guessed the weight of an ox. No single guess was accurate and was on average 37 pounds off. However, the collective guess was only off by a fraction of a percent! Even though individual guesses were wrong, the collective guess gave an accurate result.
DELPHI METHOD
In the 1950s, a team at the RAND Corporation developed the Delphi method. The Delphi method is another way to harness the power of collective intelligence. Participants in this system answer a survey and then review the answers of the other anonymous participants. They then answer the same survey again. The process goes through several rounds, with participants revising their answers based on the group’s consensus. This iterative technique allows them to consider different viewpoints and refine their own. It avoids one pitfall of collective decision making: groupthink. Because answers are anonymous and presented as data rather than discussion, it is impossible for participants to be swayed by the louder or more esteemed members of the group.
PREDICTION MARKETS
Prediction markets offer a robust interplay of design elements to harness the power of crowds to predict the future. Imagine a stock market, but instead of companies, one invests in the outcome of future events e.g., "Will inflation be above 5% by next year?" The futures contracts pay out depending on whether the event happens or not.
As people buy and sell contracts, the price goes up or down and this price reflects the collective guess about how likely the traders think the event is to happen. For example, a high price for a "yes" contract suggests the traders believe inflation will be high.
Prediction markets incentivize people to put their money where their mouth is, leading to a more honest reflection of what they truly believe will happen rather than what they want to see happen. This can be valuable for businesses, governments, or anyone who wants to make informed decisions about the future.
To learn more about the mechanisms that enable prediction markets to reveal the future, read Jen’s paper Prediction Markets as an Aggregation Mechanism for Collective Intelligence.
B3 PROCESS DETAIL
Belief3 uses panels of 50 -100 people who engage in a forecasting exercise. With 25 responses every week, we anticipate 95% accuracy rising to 99.5% accuracy when 50+ consumers participate. By way of comparison, IARPA’s ACE program engages in a similar group forecasting system and uses panels of 25 and achieved 95% accuracy in their forecasts.[1]
The longitudinal tracking of prediction error (change in forecasts) linearly predicts belief update.[2] Belief3 uses numeric input to track emotion because a single feelings integer (i.e., my pain 2 out of 6) has greater predictive power than does a combined set of economic and social variables.[3]
[1] See Norma C. Dalkey, The Delphi Method: An Experimental Study of Group Opinion (June 1969).
[1] Goldstein, McAfee and Suri, The Wisdom of Smaller, Smarter Crowds, Proceedings of the 15th ACM Conference on Economics and Computation, (June 1, 2014).
[2] Vlasceanu, Morais and Coman, The Effect of Prediction Error on Belief Update Across the Political Spectrum (Manuscript in press at Psychological Science)(2022).
[3] Kaiser and Oswald, The Scientific Value of Numerical Measures of Human Feelings, PNAS Vol. 119, No. 42 (2022).
