Approximate Truth Theory

There is a move that gets made in arguments when someone's preferred interpretation is challenged. They say: well, no one can really know, can they? Everything is subjective. You have your truth, I have mine. The move presents itself as philosophical humility. It is usually something else: a mechanism for escaping accountability by eliminating the category of better and worse accounts of what happened.

This move is wrong. Not because absolute truth is available. It is not. But because the absence of absolute truth does not make all interpretations equally valid.

The problem with certainty-or-nothing

The demand for absolute certainty before accepting a claim is not epistemically sophisticated. It is epistemically paralyzing. If certainty is the only acceptable standard, almost nothing qualifies. The history of science is a history of provisional accounts, each better than the last, none final. Newtonian mechanics was not proven wrong by the arrival of Einstein. It was superseded: shown to be an excellent approximation within a certain range of conditions, and inadequate outside that range. The better theory explained more, predicted more, held up under more adversarial testing. It was not perfect. It was better.

Peirce, the founder of American pragmatism, described truth as the limit of inquiry: the position that a community of investigators, applying good methods with sufficient time and resources, would eventually converge on [1]. This is not a definition of truth as whatever someone currently believes. It is a definition that builds in the requirement of evidence, method, and convergence across multiple perspectives. Truth, in this framework, is approached asymptotically. We never fully arrive. But some positions are much closer than others.

The implication is that better and worse apply to interpretations of reality, even in the absence of absolute proof. Some maps explain more of the terrain. Some accounts accommodate more of the data. Some interpretations make better predictions about what will happen next. These are real differences that matter.

Popper's test

Popper's contribution was the concept of falsifiability as a criterion for the quality of a claim [2]. A claim that can be tested against evidence and potentially disconfirmed is stronger than a claim that can accommodate any evidence. The strength of a scientific theory is not its resistance to counterexample but its willingness to be tested by them and its consistent performance when tested.

This criterion applies to everyday interpretations of people and situations. An interpretation that predicts behavior accurately, that holds up when the situation changes, that can accommodate new information without requiring constant revision, is a better interpretation than one that requires an ever-expanding list of special conditions to survive contact with reality. The person who interprets their partner as genuinely committed to the relationship has a theory. If behavior consistently disconfirms that theory, the theory should be revised. Refusal to revise is not loyalty. It is poor epistemology.

Most people do not think of their beliefs about people and situations as theories to be tested. But that is what they are. And applying Popper's criterion, even informally, makes it possible to distinguish the person who has updated their interpretation in response to evidence from the person who has insulated their interpretation from evidence by reframing every disconfirmation as an exception.

The absence of absolute certainty does not require a collapse into relativism. Some maps are better than others. The work is to get closer.

James and the pragmatic test

William James added a different criterion: the pragmatic one [3]. A belief is better if it makes a difference, if acting on it produces different outcomes than acting on an alternative, and if those outcomes are better by some standard. This is not the same as saying that whatever works for you is true. It is saying that the value of a belief is partly demonstrated by what it enables.

An interpretation of a relationship that allows you to see the pattern clearly and respond effectively is better, in the pragmatic sense, than an interpretation that leaves you confused and immobilized. Not because clarity of action is the only criterion of truth, but because an interpretation so disconnected from reality that it cannot guide effective action is probably not a good map.

The pragmatic test does not replace the evidential test. It supplements it. Together they produce a framework: a good interpretation is one that explains the available data, holds up under adversarial testing, accommodates new information without requiring endless special pleading, and enables effective navigation of the reality it describes.

What approximate truth asks of us

It asks two things that are in tension. First, it asks us to hold our interpretations seriously enough to act on them, to commit to them as the best available account rather than retreating into permanent agnosticism. The person who says "I can never really know anything" is not being humble. They are avoiding the risk of being wrong, which is also the risk of being right.

Second, it asks genuine openness to revision. An interpretation that explains sixty percent of the data should give way, after sufficient investigation, to one that explains eighty percent. This requires the ability to tolerate the discomfort of being wrong, to release a framework that has organized your understanding, and to rebuild around a better one.

Most people fail in one of two directions. Some are too attached to their interpretations and cannot revise even when the data clearly warrants it. Others are so uncertain of their interpretive capacity that they abandon all interpretations at the first challenge and can never build the stable understanding that real navigation requires. The discipline of approximate truth occupies the space between: hold the best current account firmly enough to act on, loosely enough to release when something better emerges.

The relativist escape hatch and why it fails

The relativist move, "your truth, my truth," is not humble. It is convenient. It removes the cost of being wrong by eliminating the category of wrong entirely. It removes the cost of being right by making rightness arbitrary. And it allows any interpretation, no matter how motivated by self-protection, no matter how inconsistent with the evidence, to occupy the same epistemic standing as a carefully developed, evidence-tested, self-revised account.

This is not a minor convenience. It is a strategic operation, usually performed by the person whose preferred interpretation cannot survive scrutiny. "Who can say what really happened" is the defense of someone who knows what happened and would prefer not to be accountable to it.

The truth-seeker should not accept this frame. Not because they are certain of absolute truth. They are not. But because approximate truth is real, is better than other approximations, and can be demonstrated to be so by the criteria of evidence, coherence, and predictive accuracy. The demand for epistemological humility applies to both parties. It is not a free pass for one.

The point

We are not in a position of knowing nothing simply because we lack omniscience. The gradient from worse to better accounts of reality is real and navigable. Truth approached asymptotically is still truth approached. The map that is more accurate is worth more than the map that is less accurate, even though neither is perfect. The work is not to find the final account. The work is to get closer, to test what you hold, to revise when revision is warranted, and to resist the collapse into the comfortable fiction that no account is better than any other.

Sources

1. Peirce, C. S. (1878). How to make our ideas clear. Popular Science Monthly, 12, 286-302.
2. Popper, K. R. (1963). Conjectures and Refutations: The Growth of Scientific Knowledge. Routledge.
3. James, W. (1907). Pragmatism: A New Name for Some Old Ways of Thinking. Longmans, Green.