The strongest case for realism begins from the success of modern science. Scientific theories actively generate new predictions, explain phenomena, and allow humans to intervene in laws of nature. For example, atomic theory explains chemical bonding, electricity, radiation, and the structure of matter, and it enabled technologies such as semiconductors, lasers, and x-rays. This practical and predictive success would be difficult to understand if the theory were not, in some important sense, tracking reality, echoing Hilary Putnam’s “no-miracles argument”. The reasoning is an inference to the best explanation: if a theory is successful in predicting phenomena and useful in technological applications, then the best explanation is that it is at least approximately true. For instance, germ theory explained infection and help developed antiseptic surgery, vaccination, and antibiotics, so a realist would say that even if we can’t see them, it should be plausible to assume that bacteria and viruses really exist and really play the causal roles.
Anti-realism offers a different interpretation of the same scientific success. It denies that scientific achievements require us to believe that scientific theories describe unobservable reality exactly as it is. The central anti-realist claim is that science should aim at empirical adequacy rather than metaphysical truth. This position is especially associated with Bas van Fraassen’s constructive empiricism, which argues that accepting a scientific theory means believing that it saves the observable phenomena, not necessarily that its claims about electrons, fields, or spacetime reveal reality in itself.
The realist argument fails firstly at the move from success to truth. Anti-realists argue that a theory can be scientifically useful without being real. For example, a subway map can be extremely effective for navigation even though it abstracts and distorts distances, directions, and geographical proportions. It is not a literal picture of the city, but it presents relations in a simplified and purpose-specific way. Anti-realists argue that scientific theories function similarly, being successful in prediction and application without mirroring reality exactly.
A second difficulty for realism is that science often advances idealized models. Scientific models tend to isolate variables and ignore interfering factors; by simplifying the world, scientists make complex systems mathematically tractable and experimentally controllable, but this contradicts the claim that science describes the world as it really is. For example, the ideal gas law treats gas particles as if they have no volume and no intermolecular force, but real gas particles do have volume and do interact with each other, which is why the model breaks down under high pressure or low temperature. Similarly, frictionless planes do not exist in reality, but physics models them to reveal relationships between force, acceleration, and motion. Nancy Cartwright’s model-based anti-realism argues that science is valuable specifically because of its artificial assumptions, idealized systems, and controlled models. Thus, science is better understood as tools for representing selected aspects of reality than describing the world simply “as it really is.”
The strongest objection to realism comes from the history of science. Larry Laudan’s pessimistic meta-induction argues that many past scientific theories were successful, yet are now considered false. Phlogiston theory, for example, once explained combustion by claiming that burning objects released a substance called phlogiston, before being replaced by Lavoisier’s oxygen theory. Similarly, nineteenth-century physicists believed that light waves required a medium through which to travel, just as sound requires air, so they posited that luminiferous ether filled all space. Ether was later abandoned after developments associated with special relativity. The geocentric model provides another example. The Ptolemaic system predicted planetary positions with considerable accuracy through epicycles. In reality, phlogiston and ether are not real entities, and the Earth orbits the sun, showing that empirical success can coexist with deep theoretical error. Therefore, the success of current science does not prove that it describes reality as it really is.
A further anti-realist argument comes from Kyle Stanford’s problem of unconceived alternatives. Scientists often infer that a theory is true through eliminative inference. Stanford points out that scientists have often failed to imagine alternative theories that later turned out to explain the evidence better than the accepted theory. Thus, scientific success cannot guarantee that current theories describe reality as it really is; they may simply be successful within the limited range of alternatives scientists have so far considered.
Finally, anti-realism is strengthened by the problem of underdetermination. Scientific evidence rarely tests a single theory in isolation. As the Duhem-Quine thesis argues, theories are tested alongside background assumptions, auxiliary hypotheses, instruments, and mathematical methods. Underdetermination matters because two incompatible theories may fit the same observable data while giving different accounts of unobservable reality. In quantum physics, the Copenhagen interpretation states that particles do not have a definite position until they are observed, and their behavior is ruled entirely by probability, while Bohmian mechanics states that particles do have definite positions guided by a pilot wave. Both theories are mathematically valid and yield the exact same physical predictions for every possible macroscopic experiment, so observation alone cannot prove which theory describes the world as it really is. Anti-realism therefore offers a more cautious interpretation: scientific theories are plausible interpretations of reality, but they are not simply reality itself.
Scientific realism argues that the universe described by science exists independently of human thought, and that our best scientific theories are true descriptions of that reality.
Scientific anti-realism argues that the goal of science is to predict and explain observable phenomena, rather than to uncover absolute truth about reality.
The claim that “modern science describes the world as it really is” is about scientific realism. Scientific realists hold that the aim of science is truth, and that mature scientific theories are at least approximately true. Against this, anti-realists argue that science need not be interpreted as a literal description of reality, but instead as empirically adequate. This essay argues that anti-realism is more convincing because scientific models depend on idealization, the history of science contains successful but false theories, and the same evidence can often be explained by incompatible theoretical frameworks.
In conclusion, scientific realism gives an impressive explanation of the success of modern science. The no-miracles argument rightly emphasizes that science’s predictive power and technological applications would be surprising if scientific theories had no connection to the world. However, the claim that science describes the world as it really is goes beyond what this success proves. Scientific models often depend on idealization, successful theories in the past have later turned out to be false, and empirical evidence may underdetermine theory choice. For these reasons, anti-realism is ultimately more persuasive. Modern science should be understood as producing empirically adequate and instrumentally powerful models of the world, rather than final descriptions of reality as it exists independently of all theoretical frameworks. Science may reveal aspects of reality, but it does not justify the stronger claim that it tells us exactly what the world is really like.