Jordan Despanie is a technology and security policy fellow at RAND with interests spanning biosecurity, artificial intelligence, open-source intelligence, and foreign policy. Before joining RAND, he was part of an elite group of professional crowdsourced forecasters recognized for accurate track records. The RAND Forecasting Initiative asked Jordan to share insights on how he achieved top-tier forecaster status to inform its approach toward recruiting and training forecasters across RAND.
How did you get started with crowdsourced forecasting and how did you get so good at it?
My forecasting odyssey began with PredictIt, a prediction market where up to $800 can be deposited to buy or sell contracts revolving around domestic and international outcomes. Although small in scale, this tangible capital investment had the net effect of aligning the level of risk I'd willingly take on, especially if my preferred outcome proved wrong, with my confidence in the research (e.g., polling data, news flow) I was able to uncover.
Following the announcement of PredictIt's legal challenges by the CFTC, however, I sought out crowdsourced forecasting platforms that had questions focused on international relations, given my longstanding passion for foreign policy issues across the Middle East and Southeast and East Asia which never had the opportunity to be expressed through my academic training as a life scientist.
This led me to INFER Public in November 2022, which I saw as a superb platform for achieving various aims: exercising and expanding my knowledge of foreign policy in an evidence-based manner; practicing policy writing by formulating rationales with a government stakeholder audience in mind; and building rapport with other forecasters to stress test my own thinking via dialectical dialogues.
What were some of your favorite forecasting projects to work on?
The forecasting projects I found most impactful included:
- Will the People's Republic of China (PRC) announce an end to its zero COVID policy by 31 December 2022?
This query merged my scientific training in virology and the pharmaceutical sciences with a focus on domestic PRC policies to successfully forecast that: PRC citizens would encounter three waves of the BF.7 Omicron variant between December 2022 and March 2023; lagging vaccination rates among 60+ year old citizens, coupled with PRC pharmaceutical policy eschewing the use of available mRNA vaccines from Western companies in favor of developing domestic equivalents, could result in an onslaught of 2.1 million or more cumulative deaths; and protests seen in Guangzhou's Haizhu district might presage an event like the A4 Revolution, which would subsequently prompt the scrapping of zero COVID and dovetail into a high cumulative death rate scenario. - Will Hu Chunhua be a member of the Standing Committee of the Politburo of the Chinese Communist Party's (CCP) Central Committee on 31 December 2022?
This query parlayed my interest in CCP factionalism towards tracking the de facto downfall of the Tuanpai (Communist Youth League) faction at the 20thNational Congress, as exemplified by the sidelining of a vice premier once considered to be a future leader of the 'post-60s' generation - From 1 January 2023 to 31 December 2023, will ACLED record 1,000 or more fatalities in Syria from remote violence and battles involving Turkish security forces?
Alongside highlighting differing policy stances involving Turkish troops stationed in northwestern Syria against the PKK during the 2023 Turkish presidential election cycle, this question leveraged my deep knowledge of the Syrian sociopolitical landscape to anticipate that President Bashar al-Assad would forgo rapprochement during clandestine negotiations in favor of setting maximalist preconditions to regain all lost territory. - Will the Pheu Thai Party be part of a governing coalition in Thailand after the next election and before 1 January 2024?
My poor score on this query, despite my prior knowledge of Thai political history dating back to the era of populist, pro–Shinawatra Red Shirts versus pro-royalist Yellow Shirts, prompted me to revisit my cognitive framework after the electoral success of the progressive Move Forward Party presented a confounding factor. This forced me to reckon with the fact that I neither envisioned a royalist prime minister stepping down nor these two opposing factions ever allying to sideline the more progressive Move Forward Party.
How does forecasting as a capability relate to your current work helping find solutions to public policy problems in technology and security?
Forecasting plays an outsized role in my work as a TASP Fellow within the Biological Evaluations Group building threat models to determine whether and how increasingly capable AI models might facilitate the misuse of biology. My colleagues and I employ these scenario-based cognitive frameworks to envision how differing types of threat actors could leverage dual-use AIxBio technologies. Throughout this process, I monitor my biases regarding how I think any given threat actor should act, given my background as a pharmaceutical scientist who thinks in terms of designing and commercializing therapeutic interventions. The assessments arising from those anticipatory frameworks, in turn, guide policy writing outputs aimed at informing government stakeholders of these emerging risks while also providing timely recommendations.
What is one piece of advice that you would give to those looking to improve their forecasting skills?
The most crucial advice I can give entails crafting rationales directly informed by hypothesis generation to tailor one's forecasting for maximal impact.
On any given topic, this involves identifying underlying drivers; potential catalyst events; and signals, in the form of qualitative and/or quantitative data, obtained via news flow and gray literature resources.
This approach ultimately serves to focus a forecaster's signals research by directly mapping those findings onto the drivers and catalysts key to answering the forecasting query.
Any other thoughts for potential forecasters at RAND?
Be prepared to spend time tracking news flow on topics pertaining to one's forecasts across multiple sources, like major outlets for breadth; local or specialized outlets for depth, early and often in order to integrate the latest signals into updated forecasting rationales. A good analogue for this forecasting approach would be the process by which a life scientist takes aliquots from a liquid solution where a biological process is occurring to track this process in a time-dependent manner until the experiment has run its course.
