You used to be right. You still are. That is not a contradiction.
Five years ago, you told a mentee: "Stay at your first job for at least three years. Job-hopping signals unreliability." Last month, you told a different mentee: "If you are not growing after eighteen months, leave. Loyalty to a stagnant environment is self-sabotage." Both pieces of advice were sincere. Both were grounded in your experience. And if someone laid them side by side without dates, they would look like a flat contradiction.
But they are not contradicting each other. They are separated by time — and the variable that changed is the labor market. In the first era, employer loyalty was rewarded and switching costs were high. In the second, the market for skilled knowledge workers shifted, tenure stopped correlating with advancement, and staying too long in a role that was not developing you became a genuine career risk. The claim did not flip because you were confused. It flipped because the underlying conditions changed.
This is time disambiguation: the practice of resolving apparent contradictions by checking whether the conflicting claims were true at different times. It is the most common form of contradiction resolution, and the one people miss most often — because the human mind tends to treat beliefs as timeless. You believe X. You used to believe not-X. Something must have been wrong. Time disambiguation says: maybe nothing was wrong. Maybe the clock moved.
Tense logic: truth is time-dependent
The philosopher Arthur Prior formalized this insight in the 1950s with what he called tense logic — now known as temporal logic. Prior's fundamental innovation was simple but profound: he made truth values time-dependent. In classical logic, a proposition is either true or false, full stop. In Prior's system, a proposition can be true at one time and false at another, and this is not a defect in the logic. It is the logic working correctly.
Prior introduced temporal operators that capture what natural language already does with tenses. "It was the case that P" (P was true at some past time). "It will be the case that P" (P will be true at some future time). "It has always been the case that P." "It will always be the case that P." These operators let you express, with formal precision, what you already do informally when you say things like "that used to be true" or "it is not true anymore."
The motivation was partly philosophical: Prior believed that the precision of formal notation was essential for resolving philosophical problems about time. But the practical consequence matters more for your thinking. Most of the contradictions you encounter in daily reasoning are not timeless logical conflicts. They are propositions that were true at different times, mashed together as if time does not exist.
"The startup should not raise venture capital" and "the startup should raise venture capital" sound contradictory. Add timestamps — "before product-market fit" and "after product-market fit" — and the contradiction evaporates. Prior's tense logic provides the formal machinery, but the practical skill is the same: when two claims conflict, check the timestamps before assuming one is wrong.
Heraclitus and the river: nothing stays the same — including you
Twenty-five centuries before Prior wrote his formal logic, Heraclitus of Ephesus had already identified the problem at its deepest level. The fragment attributed to him — "You cannot step into the same river twice" — is usually read as a statement about the river. The water moves. The river you stepped into a minute ago is physically different from the river you step into now.
But the deeper reading, which scholars have increasingly emphasized, is that the statement applies equally to the person stepping. You are not the same person who stepped into the river the first time. Your cells have turned over. Your experience has changed. Your schemas have evolved. The river changed, and you changed, and the interaction between them is therefore doubly unrepeatable.
Heraclitus was not making a mystical claim. He was making a precise one about identity and time. When you say "I used to believe X and now I believe not-X," the word "I" is doing more work than you realize. The person who held belief X is not identical to the person who now holds not-X. Both the believer and the environment have changed. What looks like a contradiction in a single person is actually a comparison across two different person-environment configurations that happen to share a name.
This is why time disambiguation is not just about external conditions changing. It is about acknowledging that you are a different cognitive system than you were when you formed the original belief. The schemas you held in 2020 were appropriate for the 2020-version of you operating in a 2020 environment. Judging them by 2026 standards is like running a database query against a schema that has been migrated three times — the old queries were valid against the old schema, even though they return errors against the current one.
Kuhn: when the whole paradigm shifts, the past becomes a different country
Thomas Kuhn's The Structure of Scientific Revolutions (1962) describes time disambiguation at civilizational scale. When a scientific paradigm shifts — from Newtonian mechanics to relativistic physics, from phlogiston theory to oxygen chemistry, from Ptolemaic astronomy to Copernican — it is not simply that scientists learned new facts. The entire framework for interpreting facts changed. What counted as a good explanation, what qualified as evidence, even what the key terms meant — all of it shifted.
Kuhn called this incommensurability: the idea that paradigms before and after a revolution cannot be straightforwardly compared, because the standards of comparison themselves changed. Newtonian physics treats mass as an absolute property of an object. Relativistic physics treats mass as a function of velocity. The word "mass" survived the paradigm shift, but its meaning did not. A statement like "mass is constant" was true under Newton and false under Einstein — not because Newton was sloppy, but because the concept of mass itself was redefined.
Three types of incommensurability make this concrete. Methodological incommensurability means the methods for evaluating claims change across paradigms. Observational incommensurability means perception itself is theory-laden — scientists literally see different things depending on which paradigm they inhabit. Semantic incommensurability means the languages of different paradigms are not fully translatable.
For your personal epistemology, Kuhn's insight translates directly. When your fundamental mental models shift — not a minor patch, but a major-version rewrite in the sense of L-0305 (Version your schemas explicitly) — the beliefs you held under the old model are not "wrong" in any simple sense. They were coherent within a framework that you no longer inhabit. Judging your 2020 self by your 2026 framework is like judging Ptolemy by Copernican standards. The question is not who was right. The question is which framework was operating at which time.
Schema evolution revisited: the time axis of belief change
Phase 16 introduced schema evolution — the idea that every mental model has a shelf life and must be deliberately updated as conditions change. L-0301 (Schemas must evolve or become obsolete) established that the models which made you effective last year will make you rigid this year if you do not evolve them. L-0305 (Version your schemas explicitly) taught you to label those versions with dates and diffs.
Time disambiguation is the contradiction-resolution technique that follows directly from schema evolution. When you find two beliefs that conflict, and both carry version numbers with different dates, the contradiction is often an artifact of comparing across schema versions. Schema v1.0 said X. Schema v2.0 says not-X. That is not a contradiction — it is an upgrade changelog.
The danger is in collapsing the timeline. If you treat all your beliefs as existing in a single timeless present — "I believe X" without the implicit "as of today, given current conditions" — then every update looks like an error, and every past belief looks like a mistake. This is why people resist changing their minds. Without temporal indexing, updating a belief feels like admitting the old version was stupid. With temporal indexing, it feels like shipping a patch.
This connects to a specific failure pattern from schema evolution: silent overwriting. When you change a belief without dating the old version, you lose the ability to see that the contradiction was temporal. The current belief simply replaces the previous one, and you lose both the history and the lesson the transition contained. Schema versioning protects against this by preserving the old version as an explicit artifact. Time disambiguation uses those preserved versions to resolve contradictions that would otherwise seem irreconcilable.
Concept drift: when the world moves and your model does not know
Machine learning provides a mathematically precise model of what happens when truth changes over time and the system fails to keep up.
Every deployed ML model is a schema — a learned mapping from inputs to predictions, trained on historical data. The moment it goes live, the world begins drifting away from the training distribution. In ML, this is called concept drift: the statistical relationship between inputs and outputs changes, causing the model's predictions to degrade even though the model itself has not changed.
Concept drift comes in several forms. Sudden drift is a paradigm shift: the relationship flips overnight, like a fraud detection model facing an entirely new class of attack. Gradual drift is the slow erosion that makes last year's recommendation engine slightly worse each month. Recurring drift captures seasonal patterns — holiday shopping behavior is "true" in December and "false" in March, year after year.
The critical lesson from concept drift is that a model does not know when it has become wrong. A fraud detection model trained on 2023 patterns will still make confident predictions about 2026 transactions — it just makes them incorrectly. There is no internal alarm. The model's confidence does not decrease as the world drifts away. It keeps producing outputs with the same certainty it had when the outputs were valid.
Your cognitive schemas behave identically. A belief you formed based on solid evidence five years ago may now be producing confident but wrong predictions about current reality. You do not feel the drift. Your sense of conviction does not come with an expiration date. The belief feels just as true as it did when it was formed — because the feeling of truth is generated by the internal coherence of the schema, not by its ongoing correspondence with external reality.
ML teams solve this with monitoring pipelines that track prediction accuracy against ground truth, trigger alerts when drift exceeds thresholds, and initiate retraining. The equivalent practice for your own cognition is the schema review protocol from L-0301 — regularly checking your models against current reality rather than relying on the feeling that they are still accurate.
Developmental context: the same behavior, different meanings
Developmental psychology provides one of the most intuitive demonstrations of time disambiguation. The same behavior — the exact same observable action — can be healthy, neutral, or pathological depending on when in a person's development it occurs.
A two-year-old who throws a tantrum when told to share is exhibiting age-appropriate behavior. The child's prefrontal cortex is not yet developed enough for impulse regulation. The tantrum is not a character flaw — it is the default output of a neurological system that cannot yet do better. The same tantrum in a thirty-five-year-old signals something very different. The behavior is identical. The developmental context has changed completely.
Erik Erikson's stages of psychosocial development formalize this. Each stage presents a central conflict — trust versus mistrust in infancy, identity versus role confusion in adolescence, generativity versus stagnation in middle adulthood. The same drive (to explore, to test boundaries, to define oneself) manifests differently at each stage, and evaluating it requires knowing which stage the person is in. An adolescent trying on different identities is doing necessary developmental work. An adult in their forties doing the same thing might be responding to an unresolved earlier conflict.
The epistemic parallel is exact. A belief that was appropriate for you at one stage of your intellectual development may be inappropriate now — not because it was wrong then, but because you have grown beyond it. "Follow the experts" is sound advice for a novice. For someone who has developed genuine expertise, it can be a trap that prevents original thinking. The belief did not become false. You outgrew the developmental stage where it was the right operating principle.
Time disambiguation in personal epistemology requires you to ask not just "when was this true in the world?" but "when was this true for me, given where I was in my development?"
Time disambiguation in your knowledge graph
If you have been building a knowledge graph through Phase 18, time disambiguation gives you a new structural tool: the validity window.
A validity window is a temporal annotation on a belief node that specifies when the belief was true or applicable. Instead of a node that says "Monolithic architecture is the right choice," you create a node that says "Monolithic architecture is the right choice [2020-2023, team size <10, pre-product-market fit]." The belief becomes explicitly time-bound.
When a contradiction appears in your graph — Node A says "X is true" and Node B says "not-X is true" — the first diagnostic question is whether the nodes have overlapping validity windows. If Node A was valid from 2020-2023 and Node B is valid from 2024-present, there is no contradiction. There is a schema version transition with a clear boundary.
The "contradicts" edge in your knowledge graph should be reserved for claims that genuinely conflict within the same time period, the same context, the same level of abstraction, and the same perspective. Time disambiguation, scope disambiguation (L-0367), level disambiguation (L-0368), and perspective disambiguation (L-0370) are the four filters you run before declaring a genuine contradiction. Most apparent contradictions dissolve when you apply one of these filters. The ones that survive all four are the deep contradictions that carry real epistemic information — the productive tensions that L-0361 (Contradictions are valuable data) taught you to mine.
The AI connection: temporal context windows
Large language models provide an unintentional demonstration of why temporal context matters. An LLM trained on data through a cutoff date will answer questions about the world as it was at that point. Ask it about a technology landscape, a geopolitical situation, or a company's strategy, and its answer reflects the training window — not the present moment. The model is not wrong. It is temporally displaced.
This is concept drift made interactive. When you work with an AI and notice that its answer contradicts your current understanding, the first diagnostic should be temporal: is the AI operating from a different time window? Is the contradiction real, or is it an artifact of different temporal reference frames?
The same diagnostic applies to your own thinking. When you discover that your current view contradicts something you wrote in your notes two years ago, the first question is not "which is right?" It is: "are these claims from the same time period?" If they are not, you likely do not have a contradiction. You have a changelog entry.
A practical protocol for temporal disambiguation
When you encounter an apparent contradiction between two beliefs, run this four-step temporal check before concluding that you have a genuine conflict:
Step 1: Timestamp both claims. When did you form each belief? What evidence or experience produced it? If you are comparing a belief from 2021 with a belief from 2026, you are comparing across schema versions. This is expected. This is evolution working.
Step 2: Identify what changed. Between the two timestamps, what shifted in the environment? What shifted in you? If the labor market changed, the technology matured, the team grew, your skills developed, or your understanding deepened — these are all legitimate reasons for a belief to flip without either version being wrong.
Step 3: Check for ongoing validity. Is the old belief still true in some contexts? "Stay at your job for three years" might still be good advice for someone in a specific industry or career stage, even if it is wrong for your current situation. If the old belief retains validity in a narrower scope, it is not obsolete — it needs a scope qualifier rather than deletion.
Step 4: Update with temporal metadata. If the contradiction resolves under time disambiguation, update both beliefs in your knowledge system with explicit validity windows. The old belief gets a closing date. The new belief gets an opening date. The schema changelog records the transition. Future-you, encountering these beliefs, will see a trajectory rather than a contradiction.
Not all change is time-dependent
A critical caveat: time disambiguation is a powerful tool, but it is not a universal solvent. Some contradictions are genuinely atemporal. "Honesty is the most important value" and "kindness is the most important value" do not resolve by adding timestamps. They are a values conflict that exists within the same moment, and they require the tools of perspective disambiguation (L-0370) or dialectical synthesis (L-0365) rather than temporal separation.
The skill is in knowing when to apply each disambiguation tool. If both beliefs were formed in the same period, under the same conditions, by the same version of you — timestamps will not help. If they were formed in different periods, under different conditions, by a you that had since evolved — timestamps will often dissolve the contradiction entirely.
Phase 19 is building your disambiguation toolkit one axis at a time. L-0367 gave you scope. L-0368 gave you level. This lesson gives you time. L-0370 will give you perspective. Together, these four axes resolve the majority of surface contradictions, leaving you with a clean set of deep contradictions that deserve the serious dialectical work the rest of the phase teaches.
The bridge to perspective
You now have three disambiguation axes: scope, level, and time. Each resolves contradictions by showing that apparently conflicting claims were operating in different contexts. But there is a fourth axis, and it is the one most relevant to interpersonal and collaborative thinking.
Two people can observe the same event, at the same time, at the same level of abstraction, within the same scope — and reach opposite conclusions. Not because one of them is wrong, but because they are standing in different places. Their vantage points, their priors, their roles, and their experiences produce genuinely different — and genuinely valid — observations of the same phenomenon.
L-0370 (Perspective disambiguation) takes up this problem. It is the axis that matters most when contradictions arise not within your own thinking, but between your thinking and someone else's. And it is the key to turning disagreement from a threat into a source of information.