The nature of scientific progress is a perennial question in the philosophy of science. Are we gradually converging on a complete, true description of reality through scientific inquiry, or is the evolution of scientific knowledge more akin to an unguided evolutionary process? In this post, I’ll argue that the answer lies somewhere in between: science produces models that are increasingly accurate representations of reality and pragmatically powerful, but these models are always partial, approximate, and provisional constructions rather than direct mirrors of an absolute, mind-independent truth.

The case for convergence is compelling. Over time, scientific theories tend to become more unified, parsimonious, and broad in their explanatory scope. Newton’s laws subsumed Kepler’s, while relativity and quantum mechanics in turn subsumed Newtonian mechanics. The rate of radical paradigm shifts and upheavals decreases as fields mature. Core elements of theories become entrenched, and disagreements, while never disappearing entirely, become more bounded. The stunning predictive successes of our best theories, from the confirmation of relativistic effects to the technological marvels enabled by quantum mechanics, suggest that our models are latching onto real patterns in the world.

However, even our most successful scientific theories are known approximations. Newtonian mechanics, despite being strictly ‘false’ in light of relativity, is still used in many contexts because it is enormously useful. The entities postulated by scientific theories, from caloric to the ether to phlogiston, often get revised or eliminated over time. And the applicability of even our best theories breaks down in some domains, as with general relativity and quantum mechanics at the Planck scale. Scientific models, it seems, can be immensely powerful without being complete or perfectly true.

Moreover, if we take seriously the idea that all knowledge is mediated by the semiotics of consciousness—that the ontology of the world is not independent of the symbolic systems through which we construct and interpret it—then treating even our best scientific models as direct representations of a mind-independent reality may be a category error. Scientific models, in this view, are always abstracted from and conditioned by modes of human experience.

So what are we to make of the general trend in scientific knowledge? I believe it is one of convergence toward increasingly accurate and comprehensive models of reality that are pragmatically sufficient for understanding, predicting, and controlling natural phenomena. Science produces models that progressively better account for our observations and enable us to successfully navigate the world. But these models are always maps, not the territory itself—they are partial, approximate, and provisionally true rather than perfect mirrors of an absolute, mind-independent truth.

The evolution of scientific knowledge, then, may be best understood as a pragmatic march toward predictive and explanatory adequacy, not a direct unveiling of a singular, timeless reality. It is a process of constructing increasingly powerful lenses for making sense of the world as it is disclosed to us through our distinctly human forms of experience and symbolic cognition.

This is a perspective informed by the history of science, as well as by instrumentalist and constructivist views in the philosophy of science. It seeks to affirm both the progressive success of science and the irreducibly constructed nature of scientific knowledge. Absolute truth may forever remain elusive, but through science, we can create knowledge that is robust, reliable, and immensely potent in helping us understand and navigate the world. The map may never be the territory, but it can become an ever more powerful tool for finding our way.