Community Challenge Highlights Fun Possibilities of Forecasting

Community Challenge Highlights Fun Possibilities of Forecasting

What are the chances that this summer will be among the ten warmest on record in the Gulf of Maine? (Surprisingly, summer 2025 didn't even crack the top 10.)

Would you have guessed the ice-out date for the Damariscotta River this spring would be after April 1? (Probably yes, if you suffered through this record-breaking winter in New England.)

Do you think there will be a Steller's sea eagle sighted in Maine this year? (It's among the world's rarest raptors and native to Asia, so you'd be safe to assume no, but an individual affectionately named "Stella" has been spotted several times in recent years.)

If you get all three of those questions right, you might be a superforecaster, someone who consistently and statistically outperforms experts when it comes to predicting the future.

For the last three years, the Tandy Center for Ocean Forecasting at Bigelow Laboratory for Ocean Sciences has organized the "Superforecaster Challenge," a friendly competition to help foster superforecasters among the institution's community and partners (with outside "experts" chiming in, including a middle schooler, several AI assistants, a professional forecaster, and, new for this year, a forecasting cat named Galaxy). Since it began in 2024, the annual challenge has grown in the number of participants and the number of questions. But it continues to offer participants room to grapple with the challenges of forecasting when it comes to Maine's unpredictable, rapidly changing environment (and the chance to win bragging rights and a specially designed t-shirt).

"Superforecasting is a captivating idea, not because of the specific predictions people make but because of what it highlights in terms of the learnable skills superforecasters have to make good predictions, like a clear-eyed view of their past mistakes and an openness to different kinds of information," said Senior Research Scientist Nick Record, who directs the Tandy Center and organizes the annual competition. "The challenge seemed like a fun way to engage with my colleagues on those ideas - and if it helps any of us make sense of the wild world we're living in these days, that's a bonus!"

At the beginning of each year, Record circulates a series of questions, and participants must estimate the likelihood that each will happen: 100% means absolutely certainty an event will occur; 0% means absolute certainty that it won't. A guess of 50% means total uncertainty.

For 2026, questions cover a wide variety of local climate and environmental phenomena, from white shark sightings to fire risk to lobster landings. Predictions, so far, are all over the place. The least likely event on the docket, according to predictions, is a sighting of a rare tufted puffin in Maine. The average guess was just 40% likelihood (actually pretty optimistic considering the bird was only sighted on the East Coast for the first time in 2022). The most likely event? A high tick count: participants predicted an almost 80% chance on average that the University of Maine Tick Surveillance program would count over 3,500 deer ticks (here's hoping that prediction doesn't come true!).

Through the experience, participants have learned numerous fun facts about Maine. They've also come to appreciate just how hard forecasting can be!

In the first year, most participants did better than the random 50-50 guess. Last year, though, only nine landed above that threshold. For everyone else, they would have been better off guessing 50% across the board. Interestingly, several of those top performers also did very well the first year; one of the hallmarks of superforecasters is their ability to reflect critically and improve, suggesting that a few participants may be ahead of the curve on that front already.

Despite last year's struggles, though, human participants still, on average, did better than the AI assistants. The best human forecasters also far outperformed the best AI predictions - a nice reminder, Record says, of the value of human experience when it comes to critical thinking. (Though, he points out, the top entrant so far this year is actually a randomly flipped 10-dirham coin from Morocco, which is a bit more discouraging for Team Human).

The superforecaster challenge is meant to be a community-oriented, fun introduction to forecasting. The actual forecasting the Tandy Center does is not so low-stakes. The team of data scientists and computational oceanographers are working to develop real-time forecasting tools that are useful and usable for a variety of users, predicting an array of events from the movements of endangered right whales to the presence of algae toxins that threaten shellfish production.

What that work has in common with the challenge, though, is that both require people to grapple with the future, which, when it comes to the ocean, is both deeply mysterious and hugely important.

"One of my core questions as an ocean forecaster is what do we do with all of this information? If a forecast predicts a future that's untenable, how does that change what we do here and now?" Record asked. "Forecasting is about more than data and algorithms, it's about imagination. This challenge may be an abstract exercise, but my hope is that it gets people asking very real questions and thinking creatively about the future and, more importantly, how we shape it."

Photo: A Steller's sea eagle perched on ice (Credit: Klub Boks).