When physicists predicted the existence of the Higgs boson decades before its detection, they demonstrated something remarkable: a mature science can forecast unseen reality with stunning precision. Meanwhile, economists struggle to predict next quarter's growth, and sociologists debate whether their theories even count as scientific.

This contrast invites a question that has puzzled philosophers of science for over a century. Are the so-called hard sciences harder because their methods are more rigorous, or because their subject matter happens to be more tractable? The answer reveals something important about what science can and cannot do, and why some questions resist neat experimental answers.

Physics studies systems that can often be described with a handful of variables. A planet orbiting the sun obeys equations involving mass, distance, and velocity. The system is closed enough, and the relevant factors few enough, that mathematics captures it elegantly.

A human society, by contrast, involves billions of interacting agents, each carrying memories, intentions, cultural histories, and the capacity to change their behaviour in response to being studied. The number of relevant variables is not just larger by degree but different in kind. Each variable interacts with others nonlinearly, producing emergent patterns that resist reduction to underlying laws.

This is why economists and sociologists often work with statistical tendencies rather than precise predictions. It isn't a failure of method. It reflects a genuine feature of their subject matter. Complex systems with many degrees of freedom may simply not admit the kind of mathematical compression that makes physics so powerful.

Takeaway

The precision of a science often reflects the simplicity of its subject matter more than the cleverness of its practitioners. Some realities resist equations not because we lack insight, but because they are genuinely intricate.

Modern physics owes much of its success to a particular trick: isolating phenomena from their surroundings. In a particle accelerator, researchers can strip away interference and observe a single interaction under controlled conditions. Repeat the experiment, vary one factor, and causal relationships become visible.

Many sciences cannot do this. You cannot place a society in a laboratory and rerun history with one variable changed. You cannot ethically expose children to different parenting styles to measure outcomes. Even ecology struggles, because removing a species from an ecosystem to study its role often destroys the very thing being studied.

Researchers in these fields develop clever workarounds: natural experiments, statistical controls, longitudinal studies. But these methods always carry more inferential risk than a controlled laboratory test. The inability to isolate is not laziness or imprecision. It is a structural feature of certain subject matters that demands different evidential standards.

Takeaway

Experimental control is not a universal scientific virtue but a luxury that some subject matters permit and others refuse. A science should be judged by how well it adapts to what its phenomena allow.

Some sciences study events that happened only once. Cosmology investigates the origin of the universe. Evolutionary biology reconstructs lineages that branched billions of years ago. Geology pieces together continental movements across deep time. These disciplines cannot rerun their experiments because their subject is the past itself.

This does not make them unscientific. Historical sciences develop their own rigorous methods: inferring causes from preserved traces, testing hypotheses by predicting what new evidence should be found if the theory is correct. When a paleontologist predicts that a transitional fossil should appear in rocks of a certain age, and it does, that is genuine scientific confirmation.

Still, the epistemic situation differs from experimental physics. Historical sciences must work with the evidence that survives, and they cannot generate new data by manipulating conditions. Their explanations are typically narrative as well as nomological, weaving together general principles with particular contingencies. This is a legitimate form of science, just one with different strengths and constraints.

Takeaway

Science is not a single method but a family of methods adapted to different relationships with evidence. Studying the unrepeatable past is no less scientific than studying the repeatable present, only differently scientific.

The hierarchy of hard and soft sciences is partly real and partly illusory. Real, because some subject matters genuinely admit more precise treatment than others. Illusory, because precision is not the only mark of scientific success.

Recognising this changes how we evaluate disciplines. We should not demand that sociology mimic physics, nor dismiss historical sciences for failing to run experiments. Each field develops methods suited to its phenomena, and judging them all by a single template misunderstands what science actually is.