Thinking Historically about the Future

The present is a proto-past to multiple possible futures — and anticipating which futures we might get is best done using classical historiographical methods, with the temporal telescope reversed

History is an inherently retrospective discipline. The historian’s particular epistemic privilege — the thing that makes historical explanation possible at all — is knowing how things turned out. Armed with that knowledge the historian works backwards, sorting the evidence into a narrative (be it tragic, comic, ironic, or romantic¹): these structural forces were operating, these actors made these choices, and these contingent events intervened to produce the outcome we now observe. For historians, the outcomes often feel, to use the jargon, “over-determined”: with the benefit of hindsight, the structural forces seem to trump the agency of the actors or the contingency of specific events. Even the most astonishing events in history — the French Revolution, say — are sometimes depicted as inevitable.

But that’s bad history. In fact, the way it turned out was always already indeterminate. For actors living through a particular era, the outcomes rarely feel predetermined at all, but rather feel deeply uncertain and dependent very much on agency and fortune. Just think about how you feel today about the future — it all seems highly uncertain, right? But I promise you that some future historian will look back on 2026 and, knowing how it all turned out, will tell us it more or less had to have turned out that way. And that again will be bad history.

The point of today’s post is that this sort of false retrospective privilege offers a lesson for thinking about the future: the assumption that because the past can be explained, the future can be extrapolated. Much conventional and folk futurology makes exactly this mistake. It identifies trends, projects them forward, and presents the result as a prediction about “the” future. But this misunderstands the basic structure of historical causation. History is merely a series of trend-lines. It is the interaction of structure, agency, and contingency — and that interaction cannot simply be run in reverse. As with the evolution of species, if you ran the timeline again, the outcomes would be different.

What I want to propose instead — and what I have long tried to practice — is an approach that takes the historian’s analytical framework seriously while being honest about the radically different epistemic conditions under which we must apply it when looking forward rather than backward. I refer to this method as thinking historically about the future.

Structure/Agency/Contingency

The framework rests on a proposition about historical causation that most good contemporary academic historians would accept, even if it is under-appreciated outside the discipline: every historical outcome is the product of three distinct types of causal factor. First, there are deep structural forces — demographic trends, resource constraints, technological paradigms, long-run shifts in economic and geopolitical power, the accumulated weight of institutional arrangements and cultural dispositions, and so on. These are the slow-moving tectonic plates of history; they do not “determine” outcomes, but they shape the terrain on which outcomes are contested. Second, there is the agency of individual and collective actors — leaders decide, movements organize, institutions adapt or fail to adapt, and these choices matter, sometimes enormously. Third, there is contingency — the random, the accidental, the unforeseeable.

The historian’s craft lies in adjudicating the relative weight of these three factors for any given outcome. Did the French Revolution happen because of fiscal weakness of the French state and class hatreds in the ancien régime that made some kind of rupture nearly inevitable? Or because of the specific choices of specific actors at specific moments of crisis, like Necker’s decision to call the Estates General or the King’s decision to flee Paris? Or because of contingent events — like the terrible harvest of 1788? The answer is always all three — but the interesting question is the relative mix. Furthermore, no historian can write about what happened in Paris in 1789 without knowing that the Terror is coming in 1793-1794. That retrospective knowledge doesn’t just inform the historian’s causal weights — it determines which aspects of the past she even bothers to look at.

It is worth noting that for non-historians — for most people, in fact — this analytical trinity rarely feels balanced. Agency tends to dominate popular historical imagination. The Great Man theory of history, the idea that extraordinary individuals are the primary drivers of historical change, might well be described as the most common form of folk historiography. It is intuitively compelling because it maps onto how we experience our own lives, where our choices feel consequential and structural forces are largely invisible background radiation. What professional historians do is to defamiliarize this intuition — to show how often the Great Man was surfing a wave he did not create and could not have stopped. (As some wag once remarked, Ronald Reagan’s “Tear Down This Wall” speech caused the collapse of the Soviet Union in the same sense that a rooster crowing causes the sun to rise.)

Academic social science has spent the better part of the last two centuries trying to move beyond the folk historiographical understanding of historical causality, and the result has been a rich, if often inconclusive, set of debates about the relationship between structure and agency. Marx gave us the most influential structuralist framework: the material conditions of production shape political institutions, and individuals, however willful, are ultimately expressions of class forces larger than themselves. Weber pushed back, insisting on the independent causal power of ideas and the choices of charismatic individuals. In the twentieth century, the debate was reformulated many times over — by Parsonian functionalism, by the Annales school’s emphasis on longue durée geographical and economic structures, by rational choice theory’s reconstruction of agency on micro-economic foundations, by Giddens’s structuration theory, which tried to dissolve the dichotomy by arguing that structure and agency are mutually constitutive rather than opposed. Bourdieu offered perhaps the most sophisticated synthesis, with his concepts of field and habitus attempting to show how structural positions are internalized by actors and reproduced through their choices — structure becoming agency becoming structure again.

These are genuine intellectual achievements, all serious students of history or social science engage with them. But here I want to register a frustration of mine with all of them: the near-total neglect of contingency. In their eagerness to adjudicate between structure and agency as the primary drivers of historical outcomes, these frameworks have tended to treat contingency as a residual category — the noise left over after the signal has been extracted, or a placeholder for causes not yet identified. This seems to me a serious error, both analytically and practically.

Consider a single example. It was not structurally determined, nor the result of any actor’s deliberate choice (conspiracy theorists’ claims to the contrary), that a deadly novel coronavirus would emerge in Wuhan in late 2019 — and that it did so at precisely the moment when Donald Trump was seeking reelection, when the accumulated failures of his administration’s public health infrastructure were about to be exposed at maximum political cost, and when the decades-long development of mRNA vaccine technology had just matured to the point where a effective vaccine could be produced in months rather than years. Each of those conjunctions was, in a meaningful sense, random. Even if it walked out of the Wuhan Institute of Virology on the bottom of some researcher’s shoe, the virus did not consult the American electoral calendar. The mRNA platform did not time its maturation to the outbreak. And yet the interaction of that contingent event with those structural conditions and those specific actors produced consequences — for American politics, for global health governance, for the legitimacy of public institutions — that are still reverberating. A structuralist account that ignores the timing misses something essential. An agency-focused account that focuses on Trump’s (or Fauci’s) decisions misses the structural context that made those decisions so consequential. And both miss the sheer contingency of the triggering event itself.

This is the key insight I want to press when it comes to thinking about the future. The present is not just a proto-past awaiting historical explanation. It is multiple proto-pasts — because which aspects of the present will matter historically depends on which future obtains. Take influencer culture. It might be the dominant cultural logic of the next half-century, in which case future historians will trace its origins in the smartphone economy and the attention markets of the 2010s and 2020s. Or it could turn out to be a passing peculiarity, remembered the way we now remember hipsters from the early 2000s — an emblematic cultural tic of a specific moment rather than a structurally significant feature of what came next. We cannot know in advance which of these proto-pasts we are living in.

This relates to why I have long been skeptical of the tendency, ubiquitous in political commentary and irresistible to the popular imagination, to focus analysis mainly on the choices of visible actors. Whether Musk or Xi or Netanyahu or Mamdani makes this decision or that one matters at the level of specific outcomes — but it matters against a background of structural conditions that none of them fully controls. The historian of 2075 will foreground a different set of causal factors than the ones dominating our daily headlines: the long-run relative economic and geopolitical decline of the United States; the broad erosion of institutional legitimacy across the democratic world; the deepening of social distrust, both horizontal and vertical; the inexorable march of climate change; the emergence of a new eco-ideological Cold War; and the disruptive emergence of large language models, which are restructuring the economics of cognition in ways whose full implications we have barely begun to absorb. These are the deep tidal forces of our moment. Actors surf the waves above them, and occasional rogue waves of contingency appear — but the tidal forces are inexorable.

This line of thinking comes out of a long a deep critical engagement with the cautionary tale of modernization theory — the subject of my book Mandarins of the Future. The modernization theorists who dominated American social science in the 1950s and 1960s were in fact engaged in a kind of futurology (sometimes explicitly): trying to think historically about the future by identifying the structural forces that had produced the historical outcomes they could observe in the West, and then projecting those forces onto the rest of the world — assuming that everyone would therefore eventually converge on the same political-economic outcome. Their fatal error was overconfidence about the repeatability of contingently-informed structural forces, the assumption that the structural forces they had identified as having “caused” modernity in the West were more-or-less present everywhere in embryonic form and immune to the slings and arrows of agency and contingency. They wrote as though they already knew which proto-past the developing world was living in: “We” were living at the end of history, and “our” present was “their” future.²

The method I am proposing is more modest, aiming for epistemic humility in thinking about the deeply uncertain future without succumbing to the epistemic despair of thinking we can anticipate nothing at all. The method begins with the most rigorous assessment we can make of the deep structural forces operating on the present — and of our genuine uncertainties about them. This is the work that distinguishes serious futures thinking from mere speculation. Then, rather than extrapolating a single trajectory, we use scenarios to model the space of plausible interactions between those structural forces and the agency and contingency that will shape how they play out. This is what I tried to do in my earlier writing on the Schwartz Window — the concept, drawn from the great scenario planner Peter Schwartz, that describes the range of futures a given society considers genuinely plausible at any given moment. The opening or closing of that window is itself a structural datum, one that tells us something important about which historical moment we are in.

Structured Anticipation

The goal of all this is what I might call structured anticipation — an historical mode of thinking, directed at the future. Just as the historian asks which structural forces, which acts of agency, and which contingent events combined to produce the world we now inhabit, the practitioner of structured anticipation asks which structural forces are most consequential in the present, how different combinations of agency and contingency might interact with them, and what range of futures those interactions make plausible. The method is the same but the temporal telescope is reversed: the historian reconstructs a particular past based on a known outcome, while we must construct possible futures in which the outcomes are presently unknown. Thinking historically about the future means embracing that asymmetry with intellectual humility — neither pretending to certainties we do not have, nor surrendering to the paralysis of pure contingency — and using the historian’s tools, adapted to our condition, to navigate the uncertainty of the present moment.

This is where scenario thinking becomes indispensable, and here I’d like to highlight an observation that Stewart Brand has often made, namely that in order to be useful scenarios about the future must be plausible in the present: if a scenario cannot be connected to present structural forces by a chain of reasoning that thoughtful people can follow, it will not engage the imagination seriously enough to do the cognitive work scenarios are meant to do. That is a real constraint, and it disciplines the exercise. But Brand’s deeper point — and it is easy to miss — is that the actual future that arrives need not have been plausible in the present. The future is not required to pass the credibility tests we set for it in advance. Consider: how much of the world we inhabit today would have seemed remotely plausible twenty-five years ago, in those last blissfully complacent weeks before 9/11? Many of the structural forces that produced our present were already operating then — the currents were running beneath the surface — but the specific shape of what arrived, the precise combination of agency and contingency that crystallized those forces into this world rather than some other, would have struck almost any reasonable observer as somewhere between far-fetched and unthinkable. I mean, Donald Trump? As President of the United States?!

This is, I think, the deepest point of thinking historically about the future. Structured anticipation is not only, or even primarily, a predictive exercise. It is a tool for shaking us out of our dogmatic slumbers about what is possible. The Schwartz Window, at any given moment, defines the boundaries of what a society considers plausible; but those boundaries are themselves historically contingent, products of habits of imagination rather than hard limits set by reality. Part of what rigorous futures thinking should aim to do is precisely to expand that window — to make vivid and thinkable futures that our present assumptions render invisible. The goal shouldn’t be to predict which future will arrive, but to ensure that as it does — in whatever form, through whatever combination of structure, agency, and contingency — we are not simply left staring at it in disbelief, but are instead ready with a plan of action.

1. It’s interesting to consider Hayden White’s work on historical emplotment in light of the agency/structure/contingency triad I am proposing here: I think White would argue that even historians who think they are judiciously adjudicating between those three causal registers are already, prior to that adjudication, choosing a narrative form that will govern which causal factors feel satisfying as explanations (e.g. whereas an historian working in a tragic mode will find structural determinism more emotionally and rhetorically coherent, one working in a satirical mode will be drawn to contingency and irony). The choice of emplotment in other words, is upstream of the causal analysis itself. I’m not personally sure this is true: it’s just as likely, it seems to me, that historians look at the historical episode they are trying to explain and only after weighing the causal factors do they choose which mode of employment to deploy. And over time, a tragedy may even become a comedy — to quote a famous line: “Comedy is tragedy plus time… The night Lincoln was shot you couldn’t joke about it. You just couldn’t do it. Now time has gone by, and now it’s fair game. See what I mean?”

2. Go ahead: buy it!