Uncertainty makes predicting the future fundamentally difficult. Anyone with a true edge in anticipating what the future will hold would do well in starting a fund to profit from it. And yet, there is a growing army of professionals –often referring to themselves as futurists– trying to master the skill of forecasting. Not just asset price evolutions or trends in macroeconomics series, but just about anything. The point of their activity is not so much to guess the future, as it is to challenge current established assumptions to manage the risks of being caught off-guard as time passes.
Related to this new professional reality, a novel tool that is currently at our disposal in crafting a future-oriented education program is the up-to-date version of future telling. For whereas single individuals have historically been reviled for attempting to guess the future, collectives tell a completely different story. One currently imbued with prestige.
According to Wikipedia, prediction markets can be thought of as belonging to the more general concept of crowdsourcing — a sourcing model in which inputs provided by a large number of participants are aggregated. But a particular one, specially designed to aggregate information on future event probabilities. The main purpose of prediction markets is to elicit beliefs over an unknown future outcome. Participants with different beliefs trade on contracts whose payoffs are related to the unknown future outcome. The market prices of the contracts, or other synthetic outputs formed via collective aggregation, are then considered real-time consensus views on an uncertainty-ridden evolution of world events.
The prediction markets angle that I find particularly exciting connects developing judgement skills, a true skin-in-the-game experience, and real-world gamification. Not to speak of the amazing learning experience that building a market –any market– entails.
Say we showed to a group of students that a 2021 Metaculus prediction on humanity developing an Artificial General Intelligence system points to 2036. And then we opened up a session to dissect the different aspects at play. These would include a detailed definition of the particular market’s criteria. What does developing an Artificial General Intelligence system actually mean in practice? What are good and bad alternative definitions? What is the right tradeoff between actual relevance and lack of ambiguity in the criteria set? In the example at hand, students would see that the market makers landed on the following:
“A single unified software system that can satisfy the following criteria, all easily completable by a typical college-educated human:
- able to reliably pass a Turing test of the type that would win the Loebner Silver Prize;
- able to score 90% or more on a robust version of the Winograd Schema Challenge, e.g. the ”Winogrande” challenge or comparable data set for which human performance is at 90+%;
- able to score 75th percentile (as compared to the corresponding year's human students; this was a score of 600 in 2016) on all the full mathematics section of a circa-2015-2020 standard SAT exam, using just images of the exam pages and having less than ten SAT exams as part of the training data;
- and, able to learn the classic Atari game "Montezuma's revenge" (based on just visual inputs and standard controls) and explore all 24 rooms based on the equivalent of less than 100 hours of real-time play.”
The aspects worth delving into would also comprise an analysis of the shape of the market curve, where students would look at the cumulative probability distribution to extract from it a 10% chance of it happening within a few years, and another 10% chance of it taking more than a century, with 50% ranging between 2029 and 2059. What are the implications of that? How can they be mapped from broad trends, to concrete, multi-faceted strategic positionings, and then to actionable items today?
Continuing with prediction markets and their educational angle, what if public commitments took on a different shape? Think of a world where promises shifted from one-off, unilateral declarations hanging on thin air, to targeted, concrete prediction market quotes achieved over a given timeline. The benefits –and incalculable educational value– of such a change could spark a revolution. From public officials to corporate management or civic leaders, the shift from committing to try to achieve to committing to measurably persuade could transform our entire approach to collective action.
There are two sides to this precious coin. On one hand, from the outside looking in. Understanding who is a good and a bad predictor, dealing with potential distortions, such as difficulty levels, domain specialization, consistency over time, etc. And more importantly, how can prediction skills be improved? On the other hand, from the inside looking out. Learning about who can credibly persuade and who struggles in transitioning from stating bold intentions to delivering actual results. How can we measure performance in a way that fosters more results-oriented individual resolve?
