The Future Perfect team at Vox revisited 25 forecasts made at the beginning of 2025, determining that 19 predictions came true while four did not. The annual exercise, conducted every December 31, assesses the accuracy of predictions made with assigned probabilities, indicating the team's confidence level in each forecast.

The team considered a prediction a "correct call" if an event with a greater than 50 percent probability occurred, or if an event with a probability below 50 percent did not occur. Conversely, a prediction was deemed incorrect if an event with a greater than 50 percent probability failed to materialize, or if an event with a probability below 50 percent did occur. Unresolvable forecasts, due to unforeseen circumstances such as governmental delays in data release, were excluded from the final tally.

Bryan Walsh, a member of the Future Perfect team, explained the methodology: "Our goal is to hold ourselves accountable and provide a transparent assessment of our forecasting abilities. By assigning probabilities, we aim to quantify our uncertainty and track our performance over time."

The exercise highlights the inherent challenges in predicting future events, particularly in a rapidly evolving technological and sociopolitical landscape. The team's forecasts spanned a range of topics, including advancements in artificial intelligence, shifts in global political dynamics, and developments in public health. The specific predictions and their outcomes were not detailed in the initial report.

Dylan Matthews, another member of the team, noted the importance of such exercises in improving forecasting accuracy. "By analyzing our successes and failures, we can refine our forecasting models and better understand the factors that influence future outcomes," Matthews said. The team plans to publish a detailed analysis of each prediction, including the reasoning behind the initial forecast and the factors that contributed to its success or failure, in the coming weeks. This analysis will provide valuable insights into the complexities of forecasting and the limitations of predictive models.