What Viruses Want
Viruses don't want anything, of course. But thinking about viral evolution as if they did — as entities with strategies and trade-offs — turns out to be one of the most productive frameworks in modern biology.
In the spring of 2020, as SARS-CoV-2 was spreading across continents and public health officials were still arguing about whether it spread through droplets or aerosols, a small group of evolutionary biologists was asking a different kind of question. Not what the virus was doing now, but what it was likely to do next. Not where it was, but where it was going.
Their framework was not epidemiology or molecular biology, though they used both. It was Darwinian logic applied to viral behavior — a set of ideas developed over decades that offered something the more standard modeling approaches couldn't: a theory of why viruses evolve in the directions they do, not just how fast they evolve. The framework goes by the technical name "evolutionary virology," and it is, as the pandemic demonstrated repeatedly, both genuinely predictive and systematically underutilized in the way scientists communicate with the public.
To understand why, it helps to start with a thought experiment that evolutionary virologists actually use in their teaching, even though it violates everything they believe about the intentionality of natural processes. Imagine, they say, that a virus has goals. What would it want?
The Trade-Off Problem
A virus wants, in this metaphorical frame, two things simultaneously, and these two things are in constant tension with each other: to replicate as efficiently as possible inside its current host, and to spread to as many new hosts as possible. These are not the same objective, and optimizing for one typically compromises the other.
The virulence-transmissibility trade-off is one of the foundational concepts in evolutionary epidemiology. It was formalized mathematically in the 1980s by Roy Anderson and Robert May, whose work on the evolutionary dynamics of parasites showed that virulence — the degree to which a pathogen harms its host — is itself subject to natural selection and tends toward a stable equilibrium that is not zero. This seems counterintuitive. Shouldn't natural selection favor viruses that don't kill their hosts? A dead host cannot transmit. The answer is that the relationship between virulence and transmissibility is not simple. Often, the same biological mechanisms that cause pathology also drive transmission: a virus that replicates rapidly, causing severe symptoms, may shed more viral particles before the host dies or is incapacitated than one that replicates modestly and causes no symptoms at all.
The equilibrium that selection drives toward is not minimum virulence but optimal virulence — the level that maximizes transmission success given the specific biological and epidemiological context. And that context matters enormously. A pathogen transmitted by a vector (like malaria, carried by mosquitoes) faces very different selection pressures than one transmitted by respiratory droplets. The vector-borne pathogen can evolve high virulence freely, because the incapacitation of the host doesn't reduce transmission — the mosquito finds you whether you're bedridden or not. The respiratory pathogen faces a much tighter constraint: it needs mobile hosts who are socializing, which is why respiratory viruses tend to evolve toward a middle ground of severity that keeps hosts functional long enough to cough on someone new.
SARS-CoV-2 followed this logic with a fidelity that evolutionary biologists found both instructive and predictable. The original Wuhan strain was highly virulent relative to subsequent variants. Delta was significantly more transmissible but also more virulent — a combination that generated ferocious pandemic waves. Omicron, arriving in late 2021, represented a dramatic shift toward a different equilibrium: highly transmissible, substantially less virulent per infection, optimized (if we must use the word) for a world in which a large proportion of the population had acquired some immunity. Omicron's evolutionary strategy, if we allow the metaphor, was breadth over depth: spread to everyone rather than killing many.
This trajectory — high initial virulence, evolving toward higher transmissibility and lower per-infection severity — was predicted in broad terms by evolutionary virologists before it happened. The predictions did not derive from genomic surveillance or case fatality rate calculations. They derived from theory: a model of what the selection pressures facing a novel respiratory pathogen in a naive population would tend to produce over time.
What the Framework Explains
The power of evolutionary thinking about pathogens extends well beyond the virulence-transmissibility trade-off. It illuminates a series of phenomena in pandemic biology that look puzzling without it and relatively coherent with it.
Consider immune escape. As population immunity to SARS-CoV-2 built up — through vaccination and prior infection — the virus faced new selection pressures. Variants that could partially evade existing immune responses gained a selective advantage over variants that could not. This is straightforwardly predictable from evolutionary theory: any time you create strong selection pressure (immunity) that disfavors some variants over others, you create conditions for the evolution of escape. The influenza virus manages this cycle annually, which is why flu vaccines require updating every year. The evolutionary virologist's view of COVID-19 vaccine policy was always more complicated than "get vaccinated and the virus goes away" — not because vaccines don't work, but because blanket immunity creates the precise selection environment that drives immune escape.
Consider also long COVID. One of the more counterintuitive findings of the post-acute COVID literature is that Omicron, while causing less severe acute illness than earlier variants, appears to cause long COVID at rates that are not dramatically lower in proportion to its infection spread. This is, from an evolutionary standpoint, entirely unsurprising: selection does not operate on long COVID outcomes, because these emerge weeks or months after the acute infectious period has ended. Whatever mechanisms produce long COVID are effectively invisible to natural selection — the virus has already spread or failed to spread before their consequences manifest. This means we should expect viral evolution to be indifferent to long COVID outcomes, and the evidence so far is consistent with that prediction.
The evolutionary framework is also clarifying about what it does not predict. It does not predict that all respiratory viruses inevitably evolve to become benign. This is a popular piece of folk evolutionary wisdom — often stated as "viruses evolve to become less deadly over time" — and it is not supported by evolutionary theory. Viruses evolve toward the level of virulence that maximizes their fitness in a given ecological context. In some contexts, that level is low (rhinoviruses, which cause common colds, are mild because they've been circulating in human populations long enough to reach an evolved equilibrium). In others, it is high (rabies, which is nearly 100% fatal in unvaccinated humans, has no evolutionary pressure toward attenuation because it is transmitted before symptoms appear). SARS-CoV-2 happened to be moving toward a less virulent equilibrium; there was no guarantee of this, and the dynamics of future novel pathogens will not follow the same script.
The Communication Gap
Given the explanatory power of evolutionary virology, it is striking how little of it appeared in mainstream public health communication during the COVID-19 pandemic. Official guidance consistently treated the virus as a static entity — here are its properties now, here is what it does to you — rather than as an evolving population of organisms responding dynamically to changes in the epidemiological environment.
This created specific failures. The early emphasis on containment was predicated on an implicit assumption that the virus was fixed, that its properties could be characterized and its spread prevented. When it became clear that the virus was mutating and that the mutations were not random noise but directional evolution under selection pressure, the public health establishment was slow to integrate this into its messaging. The appearance of Omicron caught officials by surprise partly because they were not consistently applying the evolutionary prediction that partial immunity would create strong selective pressure for immune escape.
The communication gap has structural origins. Evolutionary biology is taught in secondary and undergraduate curricula but rarely with the depth or precision that would make Darwinian reasoning about pathogens intuitive. Epidemiology, which does inform public health communication, has until recently treated pathogen evolution as a background factor rather than a central dynamic — this is changing rapidly, but the change is not yet complete. And there are sociological pressures on public health communication that favor the simple message ("this variant is more dangerous"; "get your vaccine") over the complex one ("here is the selection logic that will shape how this virus evolves over the coming months, with these qualifications and these uncertainties").
The cost of the simple message is not just intellectual incompleteness. It is prediction failure. When public health authorities consistently communicated as though each new variant would necessarily be worse than the last, they generated either panic or complacency depending on whether the variant turned out to be more or less severe. When they communicated as though vaccines would end the pandemic by achieving herd immunity, they failed to prepare the public for the reality that immune escape is a predictable consequence of widespread immunity, not a surprise development. The evolutionary frame would not have provided certainty — evolutionary predictions are probabilistic, not deterministic — but it would have prepared people for the range of possible developments rather than continually presenting new developments as unexpected.
Pandemic Futures
The evolutionary framework has clear implications for how we should think about future pandemic risk — implications that are neither reassuring nor catastrophist but are, I think, more honest than either extreme.
The greatest risk, by most evolutionary epidemiologists' reckoning, is not the evolution of existing pathogens but the emergence of novel ones from animal reservoirs. The conditions that enable zoonotic spillover — the interface between human populations, domesticated animals, and wildlife habitats — have expanded dramatically as human land use has encroached on previously isolated ecosystems. Evolutionary theory predicts that novel pathogens entering a naive human population face very different selection pressures than pathogens that have co-evolved with humans for generations. They have not had time to reach the equilibrium virulence level. They may be highly virulent by default, not because they have been selected for virulence but because they have been selected for fitness in a non-human host and are simply mismatched to ours.
The historical record supports this: the 1918 influenza, HIV, Ebola, SARS, MERS, and SARS-CoV-2 all emerged from animal reservoirs and were, in their initial forms, significantly more dangerous than the pathogens that have circulated in human populations for centuries. This is the evolutionary argument for pandemic preparedness investment that is rarely made explicitly: it is not that we can prevent novel pathogen emergence, but that the first months after emergence, before evolutionary adaptation and before vaccine development, are when the selection pressures are most unpredictable and the consequences potentially most severe.
There is also an argument, derived from evolutionary theory, about the long-term consequences of the specific immune landscape we are building through COVID-19 vaccination and infection history. The global population now has an extraordinarily complex immunity profile — different variants encountered in different orders, different vaccines with different mechanisms, different levels of waning — and this complex landscape creates a correspondingly complex selection environment for the virus. Predicting how SARS-CoV-2 will evolve in this environment is, as evolutionary virologist Trevor Bedford and his colleagues have noted, substantially harder than predicting influenza evolution, partly because the immune landscape is so heterogeneous. The appropriate response to this uncertainty is not fatalism but investment in surveillance capacity — the ability to detect evolutionary signals early, before new variants have spread globally.
The Anthropomorphism Defense
Evolutionary virologists are sometimes challenged, by students and by more fastidious colleagues, on the anthropomorphism embedded in their framework. Viruses don't want things. They don't have strategies. They replicate according to chemical rules, and the ones whose replication is favored by the environment proliferate. The language of "viral goals" is a metaphor, not a description.
This is true, and the best practitioners of evolutionary virology know it well. But it is also true that the metaphor works — that thinking about viruses as entities with evolutionary interests, navigating trade-offs, responding to selection pressures, produces accurate predictions in a way that thinking about them as static chemical assemblages does not. The heuristic earns its place by its track record. When Andrew Read at Penn State and colleagues developed evolutionary theory predicting that imperfect vaccines — ones that reduce disease severity without sterilizing infection — can, under some conditions, select for more virulent pathogen strains, they were using exactly this framework. The prediction has been empirically validated in Marek's disease in chickens and is actively studied as a potential risk in other vaccination contexts.
The deeper point is that evolutionary thinking is uncomfortable because it is the opposite of the control narrative that public health communication naturally gravitates toward. Evolutionary biology does not promise that we can eliminate pathogens; it describes the dynamics of an arms race without a finish line. It does not promise that vaccines will make viruses go away; it describes the selection pressures that vaccines create and their likely consequences. It is not pessimistic — evolutionary insights have guided remarkably effective interventions, from myxomatosis control in Australian rabbits to HIV treatment strategies that minimize the evolution of resistance — but it is unsentimental.
A pandemic communication strategy that routinely incorporated evolutionary framing would ask the public to hold a more complicated picture in mind: not a war that will end in victory, but an ongoing relationship between human populations and microbial life, shaped by selection pressures that we partially control and partially don't. The viruses don't want anything. But understanding what they would want, if they did, is still the best guide we have to where they are going.