When nearly 150 million US voters cast their ballots on 5 November, opinion polls were neck and neck, and commentators were forced into the awkward position of hazarding a guess. The story for 24/7 news became the pollsters themselves, their hourly new insights and their potential influence on the campaigns and the result. Questions about sizes of poll samples, whether people are truthful when polled, whether pollsters themselves are seeking to influence the result, whether the betting sites projected into America from third countries were seeking to subvert democracy, the cost of polling, whether pollsters are any good, all mask the fact that the important poll is the secret ballot. The urge to know the outcome before the event is another symptom of our anxiety about the idea of living with uncertainty. It may help to think about it differently.

It has become customary to make a distinction, when making assessments of world events, between a secret and a mystery. The former Cabinet Secretary, Lord Butler of Brockwell, made this observation in the Review of Intelligence on Weapons of Mass Destruction, published in 2004. At the end of chapter one, under the heading ‘The Limitations of Intelligence’, the review noted that: ‘A hidden limitation of intelligence is its inability to transform a mystery into a secret. In principle, intelligence can be expected to uncover secrets. The enemy’s order of battle may not be known, but it is knowable. The enemy’s intentions may not be known but they too are knowable. But mysteries are essentially unknowable: what a leader truly believes, or what his reaction would be in certain circumstances, cannot be known, but can only be judged.’

In a similar vein, US Secretary of Defense Donald Rumsfeld’s ‘known unknowns and unknown unknowns’ answer to a question at a US Department of Defense news briefing on 12 February 2002 has become a shorthand for the things we don’t understand or can’t anticipate. The question was about evidence linking the government of Iraq with the supply of Weapons of Mass Destruction to terrorist groups. The mystery becomes a mix of the known and unknown unknowns.

In fact, neither of these approaches are helpful, because they obscure the real distinction, which is between the potentially knowable past and present, and the unknowable and endlessly mutable future. This is a question between our ability to ascertain and interpret established fact, and facts yet to be created. The difference is a temporal one, and it would be clearer to think about it in these terms.

When making an assessment of a geopolitical adversary it is common to break the challenge of understanding into two parts: their capability and their intent. Russia’s invasion of Ukraine has reminded us that capability can also be separated into ‘capability on paper’ and ‘real capability’; the analyst’s job is to understand as fully as possible from imperfect data what the adversary could do in theory, and then to refine that by an understanding of what they could do in practice. Is the military’s inventory properly representative of the number of serviceable vehicles, for example? The answer to Rumsfeld’s question was essentially knowable, because it was established in the past and present. To the question ‘does Saddam Hussein have weapons of mass destruction and is he giving these to terrorist groups?’, the answer could be either: ‘yes’ or ‘no’; he either does or he doesn’t. The assessment could be right or wrong. This doesn’t mean it is easy, but it is theoretically possible to answer.

Intent, however, equals the future. This is a completely different kind of analytical challenge to questions of fact, which cover the past and the present, and where the answer could be right or wrong. In January 2022 there could not have been a right or wrong answer to the question: ‘Will Vladimir Putin invade Ukraine?’, because it had not happened yet. The question should have prompted the answer: ‘It depends on what happens to make him change his mind.’ Thinking about the future as a mystery to be divined, rather than an outcome to be shaped has become hard-wired into the way we talk about what will happen next in foreign policy. It has got to stop, because it is wasteful of time and energy, and prevents us from having the right conversations about how we might shape what happens next to our advantage. There is a risk that the quest to predict the future correctly becomes an end in itself, absorbing bureaucratic and intellectual effort which could better be spent in preparation.

Professional analysts use the language of probability as a means of communicating uncertainty. This is often done through percentages of likelihood, which confer a falsely scientific impression. And we have sort-of learned to contend with this way of thinking. The publication since Russia’s invasion of Ukraine of regular updates from the UK’s defence intelligence community has revealed to a wider audience the existence of the ‘Probability Yardstick’, which describes a scale of probabilities split into seven distinct numerical ranges with terms assigned to each probability, ranging from ‘remote chance’ at one end to ‘almost certain’ at the other. Possibly confusingly, these are qualified by a statement of high, medium or low confidence in analytical judgements.

The problem is that the distinction has still not been made between a potentially knowable fact (past and present) or a yet-to-be-created future fact. The use of a common probability yardstick to describe assessments both of knowable already-existing things and not knowable not-yet-existing things is confusing, not least because in assessing whether something is likely to happen in the future, a completely different set of factors come into play. It doesn’t make sense to apply the same methods and language because the number of variables is so much greater. The analysis needs to include not only that of the enemy’s intent, but the likely impact on the enemy’s choices of the actions of ourselves or our allies. That requires an understanding of the likely future course of the actions of allies as well, and what might influence their choices. Naturally assessments of future possibility may also serve to provide a warning, to ensure that action can be taken to avoid it coming to pass, which would itself make the predicted future incorrect. We are tied in knots.

Instead, we should require absolute clarity, when using probability in assessments, as to whether these assessments relate to the past and present or to the future. For those seeking to produce analysis which looks at future events, the more important question for decision-makers is usually the 50 per cent mark: has this now become more likely than not? And if so, how does that change our decision-making process about what to do? If we don’t want it to happen, what action do we need to take to reduce the likelihood of it happening, or to prepare for it so that we are ready?

This brings us neatly back to the 50:50 US poll split. In the case of elections, where it would be impossible and improper to influence the outcome, energies should flow instead into robust provisioning. The significant fact was not who was topping the polls from one day to the next but that it was seemingly too close to call for months on end. As with other large democracies that voted this year, such as India, the presidential election was an enormous exercise in collective decision making. Of course, the pollsters would not be quite right. There is an entrancing unpredictability about devolving the nation’s biggest strategic decision to a crowdsourced model in which every adult, with varying degrees of information and insight, has an equal say. Polls are entertaining but diverting, and can become a psychologically corrupting, hugely distracting end in themselves. In anticipation of a secret ballot of 150 million people which is too close to call the only possible course of action should be – yes – to monitor and seek to understand, but most importantly to prepare for either outcome.