Facebook announced last week that it is shutting down Forecast, an experimental mobile app that lets users make predictions about world events. The app, which launched in June 2020, will disappear on Oct. 15., marking the end of one of Facebook’s most interesting experiments.
Forecast lets users make predictions about everything from Covid-19 vaccination rates to who will win the Great British Baking Show. The app operated like a market, but without any real money exchanged. Users spent virtual points to “buy” or “sell” outcomes they thought were more and less likely to occur.
The app was an experiment in making discourse online more factual and reasoned, by a company that has struggled to police an ocean of misinformation. Forecasting platforms force users to turn their beliefs about the world into public predictions. By combining users’ forecasts, these prediction markets, researchers suggest, can be more accurate than experts at forecasting the future. And by practicing and tracking their predictions over time, researchers found people can overcome biases and make better forecasts about real-world events.
That was the vision for Facebook’s Forecast app. “We became interested in prediction markets because, when they work, they help push participants to be rational,” argued Rebecca Kossnick, the project lead in an Oct. 2020 blog post. Framing the discussion as a forecast, she argued, could incentivize rational conversation around divisive topics. “In a period of increasing uncertainty and division,” she wrote, “what we want most is to create a space in the world where people, especially those with divergent views, can find common ground.”
Facebook wasn’t the first to launch an app like Forecast, but it was the one best positioned to see if prediction markets could really work on a global scale to make social media better. Facebook’s decision to sunset Forecast signals that for now, at least, that’s unlikely.
How Forecast works
Forecast came out of Facebook’s New Product Experimentation group, an R&D operation where teams of three to seven people work on creating new projects outside of the core Facebook apps. Any user can submit a question for others to predict, though the Forecast team vets the questions. For example, a question about vaccination allows users to “buy” the outcome they think is most likely to happen, or that the market is underrating.
The app has a leaderboard showing which users have earned the most “profit” by accurately buying or selling, and users get points for posting explanations of their trades that other users upvote.
But many reviewers seemed confused about whether the point was to make accurate predictions or to discuss current events. Forecast was also only available on iOS, a fact multiple people active on other forecasting platforms whom Quartz spoke to cited as a reason they hadn’t used it.
Kossnick, the project lead and an avid forecaster, left Facebook in March for a startup. The Forecast app’s Twitter account has not posted since January. A Facebook spokesperson declined to comment beyond the statement released on the Forecast website. “Forecast began as an experimental app to help us understand how crowdsourced wisdom can arrive at truth,” the statement reads. “We’ve learned so much with and from you, and will be sunsetting Forecast on October 15, 2021.”
The case for forecasting platforms
Facebook’s app was one of several online forecasting platforms. Some allow users to wager real money on predictions, like Kalshi, a Y Combinator-backed startup, and PredictIt, a project of the Victoria University of Wellington. Others don’t involve risking money and are aimed at hobbyists, like Good Judgment Open, CSET Foretell, and Metaculus.
But Forecast was different because of the resources of the company behind it, and its experience building apps for billions of users. Most forecasting sites tend to attract people who already think probabilistically or enjoy online betting. (Forecasting platforms can be fun, but almost no one would call them addictive.)
Facebook’s entry raised the possibility that forecasting platforms could grow beyond this narrow audience, turning them into a new form of social media. In the Forecast launch post, Kossnick hints at how forecasting might fit into the rest of Facebook:
“At scale, we think that the points system in Forecast could serve as a new form of expert credentialing, both for forecasters and for the information itself. If, for example, someone makes consistently great forecasts on economic news, maybe that could give them access to extra responsibilities or distribution on economic topics. Similarly, the frequency with which a source is cited by the best forecasters could be used as one signal to determine the quality of its content.”
In other words, forecasting accuracy could be one way for Facebook to decide who’s worth paying attention to. But as Forecast shuts down, the hope that forecasting platforms will become the next frontier for social media appears to be fading. For now, it will left to other, more specialized forecasting platforms to see if predicting the future can encourage critical thinking and more reasoned and empirical discussion online.