Who Is More Reliable Nate Silver or the U.S. Polling Nostradamus? We Asked ChatGPT

As former President Donald Trump and Vice President Kamala Harris go toe to toe for the presidency, two political heavyweights have been engaged in a heated social media spat throughout the election cycle.
U.S. polling “Nostradamus” Allan Lichtman and FiveThirtyEight founder Nate Silver have been embroiled in a war of words over their predictions for the 2024 presidential elections.
Until, that is, Lichtman recently extended an olive branch to the pollster and said he is ready to end his long-running feud with Silver.
Lichtman is a political historian and professor known for developing the “Keys to the White House,” a model with an impressive track record of predicting presidential election outcomes since 1984. He has forecast that the Harris-Walz ticket will win the White House in November.
His system is based on 13 true/false statements called “keys,” which focus on broad indicators of the incumbent party’s political strength rather than polling data or granular analysis.
The model ignores day-to-day events and public opinion polls, and relies instead on the historical context and larger trends that shape the electorate.
Silver released data showing Harris leading Trump in the national polling average by nearly three points—48.9 percent to 46 percent—but Trump and his running mate, JD Vance, hold a 56.2 percent chance of winning the Electoral College, compared to 43.5 percent for the Harris-Walz ticket.
The feud dates back to 2011, when Silver wrote an article for The New York Times titled, “Despite Keys, Obama Is No Lock,” criticizing the subjectivity of Lichtman’s keys, especially the “incumbent charisma” factor.
Silver questioned Lichtman’s confidence that Barack Obama would defeat Mitt Romney, given the struggling U.S. economy and Obama’s low approval ratings at the time.
In 2016, Lichtman predicted Trump’s victory using his 13 Keys model, while Silver’s FiveThirtyEight gave Hillary Clinton around a 71 percent chance of winning.
The dispute flared up again in July 2024, after President Biden’s poor debate performance against Donald Trump but before his withdrawal from the race.
To settle the rivalry once and for all, Newsweek asked the generative AI tool, ChatGPT,: “Who is more reliable at predicting the outcome of U.S. presidential elections, Nate Silver or Allan Lichtman?
“Take into consideration past election predictions by both and give reasons for your answer.”
“When comparing Nate Silver and Allan Lichtman in terms of predicting U.S. presidential elections, both have impressive track records, but their approaches are fundamentally different. Here’s an analysis of each based on their methodologies and past predictions.
“Nate Silver is a statistician and data analyst. He uses a quantitative, data-driven approach to predict elections, relying on polling data, historical trends, demographics, and other factors. His model at FiveThirtyEight aggregates polling data and weights them based on their reliability, historical performance, and more.
“Strengths: Silver’s models adjust in real-time as new data becomes available, offering a probabilistic forecast rather than a binary yes/no prediction.
“Weaknesses: His forecasts can appear uncertain because they are probabilistic. For example, if his model gives a candidate a 30% chance to win, people often interpret it as a loss, even though such an outcome is far from impossible.
“Allan Lichtman uses a historical model called “The Keys to the White House,” which is based on 13 true/false statements that are primarily focused on the incumbent party’s performance rather than polling data. His method is based on historical patterns rather than real-time polling or statistical models. If six or more of these statements are false, the incumbent party is predicted to lose.
“Strengths: Lichtman’s model is remarkably consistent and has been accurate over nearly four decades. It focuses on structural and political factors like economic performance, social unrest, and scandals, which often play significant roles in election outcomes.
“Weaknesses: His model is more binary, predicting only the winner without assigning any probability. It also doesn’t take real-time data like polls or state-by-state results into account, and it might miss nuances in the Electoral College versus the popular vote.
“Both Nate Silver and Allan Lichtman have strong records, but they excel in different ways.
Nate Silver is more reliable if you want probabilistic forecasts and a detailed, data-driven approach that accounts for real-time information like polling and demographics.
“His models can give a clearer picture of how tight an election might be, and he is especially good at capturing electoral dynamics on a state-by-state basis.
“Allan Lichtman is more reliable if you’re looking for a binary, long-term prediction that focuses on broader historical patterns and structural conditions.
“His method has been consistently right in predicting the overall winner (though not always the Electoral College dynamics).
“If we’re looking at long-term historical accuracy, Allan Lichtman’s “Keys to the White House” have been spot-on in predicting the winner. If we want a more granular, real-time analysis of how the race is unfolding with probabilities and state-by-state insights, Nate Silver is the better choice.
“Ultimately, both have strong claims to reliability, but Lichtman’s simplicity and long-term accuracy have made him a standout in election prediction over the past 30+ years, whereas Silver’s nuanced, data-driven approach is invaluable for understanding election dynamics in detail.”
Newsweek has contacted Allan Lichtman and Nate Silver for comment.
Do you have a story Newsweek should be covering? Do you have any questions about this story? Contact [email protected]