Farage on top, and Musk fired

Will Kemi Badenoch still be the leader of the Conservatives by the end of 2025? Will Rachel Reeves be in Number 11? Will a deal over the Chagos Islands ever happen, and what chance a ceasefire between Russia and Ukraine?

Predicting politics and current affairs is not an easy business. In the autumn of 2021, Boris Johnson was infamously described by one distinguished journalist as squatting over British politics like a “giant toad”. Stories were written about him eyeing “another decade in power”. In less than a year he was gone.

But what if there was a way to boost our powers of political prediction? And what if it lay in the intuitions, hunches and guesswork of ordinary men and women

Iain Mansfield is a former government special adviser who for the last three years has run an online forecasting contest, challenging people to make predictions about UK politics and the economy, world events, science, the arts and other topics.

This year, nearly 400 people have taken part, with about 100 holding down jobs in public policy, politics or journalism (including a few reporters who work for The i Paper – your author included).

Given the number of people involved, intriguingly, Mansfield says it should have some predictive power for how 2025 will unfold.

The Wisdom of Crowds

The claim rests on the “Wisdom of Crowds” – a phenomenon whereby the collective opinion of a group of people often provides a better judgement than that of a single expert.

In its most famous example, the statistician Frances Galton recorded how in a competition to guess the weight of an ox at a country fair in 1907, the average of all the 787 entries was astonishingly accurate (1,207 lbs against an actual weight of 1,198 lbs) beating everyone in the contest, including apparent cattle experts. The Wisdom of Crowds happens because when a large group of individuals with diverse perspectives provide independent estimates on a topic, the aggregate of the estimates tends to cancel out the individual biases and errors.

Mansfield’s competition invites people to predict the probability of 45 different things happening in 2025, with contestants putting the likelihood as a percentage between 0 and 100. For example, what is the chance of Sir Keir Starmer still being Prime Minister by the end of the year, Reeves being Chancellor and Badenoch being Tory leader?

No expertise necessary

Crucially with the Wisdom of Crowds, those taking part do not have to be experts. “As long as they’re averagely informed people, then just averaging can make something that is really hard for people to beat,” Mansfield says. In his 2024 prediction contest, the Wisdom of Crowds beat 87 per cent of the individual forecasters taking part in the exercise.

So what does the crowd think will happen in 2025? Mansfield’s hivemind thinks that there is an 88 per cent chance that Starmer will be PM by the end of the year. Reeves and Badenoch might have more reason to feel uncomfortable – Reeves was given a 65 per cent chance of being Chancellor, and Badenoch a 78 per cent chance of being Tory leader.

The crowd’s prediction will make gloomy reading for Tory supporters. Contestants collectively thought there was a 62 per cent chance that Reform UK would be polling ahead of the Conservatives by the end of the year (and a 51 per cent chance Nigel Farage’s party would be polling ahead of Labour).

On the policy front, the crowd thought there was a 67 per cent chance that net migration to the UK in the year to June 2025 would be over 500,000, but only a 26 per cent chance that the total number of houses built in the UK in 2024-205 would be over 300,000 – the figure Labour will need to hit to keep up on its 1.5m new homes by 2029 target. The crowd also had a dim view of the likelihood of Starmer’s deal to hand the Chagos Islands to Mauritius materialising, giving it a 46 per cent probability.

Trump taking Greenland seen as unlikely

On the international front, the crowd thought there was a 51 per cent chance that Elon Musk would be fired from his role at DOGE and a 16 per cent chance that either the Panama Canal or Greenland would be ceded to the US.

The chance of a lasting ceasefire or truce between Ukraine and Russia before the end of the year was put at 50 per cent, a Chinese blockade of Taiwain at 32 per cent and the exit of 500,000 or more Palestinians from Gaza (whether forcibly or voluntarily) at 36 per cent. (The Gaza question was posed presciently before Donald Trump’s recent statements about “buying and owning” the territory).

Of course, the Wisdom of Crowds is not an oracle. Mansfield says that “it falls down slightly” when there are “black swans” – high impact events that come as a surprise. “Last year the one big thing where the wisdom of crowds was off was the success of Reform in the general election, so this is not a surefire thing,” he says.

It is also possible for particularly prescient individuals to beat the crowd. For the last two years, Mansfield’s competition has been won by a health economist. “He’s not a politics expert but he follows the news and is clearly really into statistics and probability.” And how did those who work in politics for a living perform? The reputation of “experts” has taken a bit of a battering of late. But Mansfield says that “on average, last year, those who were in politics did do better than those who weren’t by a noticeable margin”.

Dominic Cummings and superforecasting

Mansfield’s competition is mainly a bit of fun, but he thinks there is a serious point too. SW1’s interest in prediction waxes and wanes. It hit its highpoint during Dominic Cummings’ reign in Number 10, when the controversial strategist ordered special advisers to read books on “superforecasting”, but it fell out of vogue when he left the scene.

Mansfield thinks the quality of government would be better if Westminster and Whitehall took a more systematic approach to forecasting (journalists should also hold themselves accountable for their predictions, he adds). He argues “it would be very helpful” if there were more assessments done of the accuracy of forecasts that come out of institutions like the Office for Budget Responsibility so that “we could judge more if they consistently got it wrong”.

To take one example, whenever the Government receives legal advice, its lawyers put a forecast on how likely there is to be legal challenge, and how likely it is to succeed. The problem is, Manfield says, “no one ever goes back and checks are they correct?”. Improving the accuracy of forecasts (or using lawyers with a better grasp of likely outcomes) could save the taxpayer considerable amounts by not pursuing doomed litigation, he thinks. “Getting more used to predictions with probabilities, predictions with error bars, this should be something which we try and normalise in government and the media,” he says.

How to predict the political future

So if you are an armchair political pundit, how can you hone your own powers of prediction? Mansfield has some advice. Firstly, start with a “base rate” probability. “So if you’re looking at the chance a leader will change or something, or a natural disaster will happen, look at the base rates for just how often it happens. That gives you a grounding to adjust around.”

Next, try to consider all the different things which could happen for a given scenario. When Mansfield asked people to set the likelihood of Joe Biden being re-elected last year, lots of contestants plumped for 50 per cent. On the face of it, it was a reasonable call given the polls were about 50/50. But deeper thought might have yielded a more tailored response. “If you think more, well, one, Biden could die because he’s quite old. What are the odds of an 82 year-old-man dying?” It could happen, Mansfield points out, and secondly there was a chance that he could pull out of the contest, as indeed happened.

“So if you’re answering whether Biden could win, you should be downgrading that 50 per cent prediction because of all the small possibilities that could happen.” Contestants who regularly make these small adjustments with their probabilities can pick up “material advantages” in the game, he says.

Lastly, try to put your emotions to one side. “People end up betting on what they hope will happen or are afraid will happen,” Mansfield says. Last year, a lot of contestants attached “very high probabilities” to the idea that large numbers of immigrants would be deported to Rwanda, “probably because they were worried about it and they let that affect their probabilities”.

“The people who do well are the people who can separate what they want to happen or fear will happen from what the data is telling them.”