Nothing Is Still. So Why Do We Keep Measuring as If It Were?
There is a strange assumption buried inside almost every psychological test ever designed.
It assumes that the person being measured holds still.
Not physically. But cognitively, emotionally, relationally, as the thing being measured could be sufficiently stable to be captured by a snapshot. That personality has traits, and intelligence has a score. That resilience is a property you either have more or less of, locatable somewhere inside the individual, extractable by the right set of questions, recordable to two decimal places.
This assumption was never explicitly stated. It didn’t need to be. It was the water psychometrics swam in. And… it is wrong. Literally. Not in a minor, revisable way. Wrong in the way that matters most — wrong about what kind of thing a human being actually is.
The Measurement Was Never Measuring What It Said
In How We Misunderstand Understanding, we traced how cultures across history, Slovenian, German, Latin, or Sanskrit, all embedded movement inside their words for knowing. To understand is in the Slovenian language ‘razumeti‘ – to fragment in order to reintegrate. To verstehen: to take a stance in relation, not to possess a fixed property. Understanding, as the previous essays in this series showed, is an ecology, a lifecycle, a choreography of footing. It cannot be static because it is constituted by the very dynamics that falsify the idea of static states.
The same is true of everything psychometrics tries to measure.
Personality is not a trait distribution. It is a configuration of coherence that shifts with context, with relationship, with history, with energy state, with the level of trust in the room. An introvert in a psychologically safe team behaves like no introvert that any Big Five model predicts. A person scoring high on conscientiousness collapses under chronic coherence failure, what we usually call burnout, and temporarily presents as someone entirely different. The score captured a snapshot of a state that was already moving before the pencil hit the paper.
Intelligence is not a fixed quantity. It is a dynamic capacity that expands with resonance and contracts under threat. A child tested in a sterile, high-stakes environment will demonstrate less of what they are capable of than the same child in conversation with someone they trust. The test doesn’t measure intelligence. It measures intelligence under these specific conditions with this specific relationship to this specific examiner on this particular morning. That is not a measurement of a property. It is a measurement of a momentary configuration of a dynamic system. Calling it a score and comparing it across children, schools, and generations requires a metaphysical claim that nobody in psychometrics has actually defended: that the configuration holds still enough for comparison to be valid.
It doesn’t. It never did.
The Deeper Problem: Science Measuring Its Own Shadow
This is not a critique of psychometrics alone. It is the problem that emerges across every domain where science applies measurement instruments to living, adaptive, recursive systems.
We argued in the previous essay that science, being built around the logic of falsifiability and replicability, was optimized for stable, repeatable phenomena. It produced immense power there: orbital mechanics, thermodynamics, chemistry, engineering. The world that obeys equations held still enough, or was made to hold still enough, for the method to work.
But a human in a therapy session is not an orbital body. An organization in transformation is not a chemical reaction. A culture digesting a crisis is not a thermodynamic system. These phenomena change during observation, not as a side effect, not as measurement error, but as the very thing they are. Observation is participation. The measure changes what is measured. The word used to describe what is happening partially constitutes what happens next.
And yet the instruments remain frozen. Psychometric questionnaires validated in 1978. IQ scales normed on populations that no longer exist in the conditions under which they were normed. Leadership competency frameworks written as if leadership is a stable set of behaviors rather than an emergent property of a specific relational field at a specific moment in a specific cultural context.
What science has been doing, without quite admitting it, is measuring its own shadow. Projecting static categories onto moving phenomena and then studying the shape of the projection. The category is clear. The measurement is clean. The thing itself was never there.
Words Are Not Containers. They Are Forces.
Here is where the argument sharpens into something that most scientific discourse actively avoids.
Language is not neutral.
This sounds obvious when stated plainly, but the implications are radical and mostly unacknowledged. When we call something intelligence, we are not just labeling a phenomenon that exists independently of the label. We are constituting a category that then organizes what we look for, what we measure, what we value, and how people understand themselves. Children who grow up being told they scored high on intelligence tests develop a particular relationship to their own cognition, often more fragile than those who were never given the score, because they come to believe that intelligence is a fixed property that can be lost. The word produced a psychological reality that then needed to be managed, defended, and, eventually, dismantled by researchers who spent decades studying what Carol Dweck called the growth mindset. Which is simply: understanding is processual, not fixed. Which is what we said about understanding in the first place.
The same operation runs everywhere.
Call chronic stress a threat response, and you have already decided that the appropriate intervention is protection from threats. Call the same phenomenon coherence collapse, propulsion energy looking for a new configuration, and the appropriate intervention changes completely. Not protection. Reorientation. Not threat removal. Coherence reconstruction.
Call an AI’s false output a hallucination, and you have already decided it is a psychiatric symptom, a deviation from normal functioning, a bug. Call the same phenomenon decoherence, a loss of eigenstate, a system that has resolved its uncertainty in the wrong configuration, and the appropriate intervention is not debugging but coherence architecture. The word doesn’t just describe the problem differently. It defines a different problem.
Call a human who is struggling to adapt resistant, and you have located the problem inside the individual, as a fixed trait or a willful refusal. Call the same human in a coherence transition, moving between configurations, momentarily unstable, and the problem is structural, contextual, relational. One word produces a case for pressure. The other produces a case for support. Same behavior. Different ontology. Entirely different consequences.
This is what it means to say that words are more important than first thought.
They are not labels attached to pre-existing realities. They are frames that organize perception, constrain interventions, and (this is the crucial part) feed back into the system being described. A person told they are resistant will often become more resistant, because the word has now become part of their identity. A person told they are in transition will often stabilize, because the word preserves their sense of coherence through the instability. The measurement changed the thing measured. The label constituted part of the phenomenon it claimed to merely describe.
This is not a soft claim. It is one of the most robust findings in the psychology of language, social cognition, and organizational behavior. It is also the finding that most scientific practice most consistently ignores, because acknowledging it would require admitting that the instrument cannot be separated from what it measures, and that would threaten the entire architecture of controlled, replicable, value-neutral inquiry.
Psychometrics Needs Theory, Not Just Data
There is a precise way to name what is missing.
Psychometrics, in its dominant form, is extraordinarily good at detecting variance. It can tell you that a population distributes along a dimension, that scores on measure A correlate with outcomes at measure B, that an intervention shifts the group average. This is not nothing. It has produced genuinely useful knowledge.
But correlation without theory is not understanding. It is pattern recognition in the dark.
When a psychometric measure detects that employees with lower scores on an engagement survey have higher turnover rates, it has detected a signal in the noise. What it cannot tell you (because it was never designed to) is why the engagement is low, what kind of disengagement is operating, which aspect of the coherence ecology has failed, and what would restore it for this specific team in this specific moment of this specific organization’s history. For that you need theory. You need a framework that can hold the dynamics, the context, the temporal dimension, the relational field.
Data without theory produces better descriptions of the shadow. Theory without data remains untethered to the moving thing. What science needs, what the next paradigm needs, is instruments that are built for motion. Measures that track change over time, not states at a point. Frameworks that account for context as a constitutive variable, not a confound to control. Conceptual vocabulary that names dynamic processes rather than static properties.
This is not asking science to abandon precision. It is asking science to be precise about different things. Not “what is the score?” but “what is the coherence configuration, and in what direction is it moving, and under what conditions?”
That is a harder measurement. It is also the only one that touches what is actually there.
What AI Has Shown Us About Our Own Instruments
There is a reason this realization arrives with particular force through the lens of AI.
AI systems make visible, in stark terms, what human systems obscure through familiarity. When a language model produces confident, fluent, coherent-sounding output that is factually wrong, we call it a hallucination and treat it as a malfunction. But what it has actually demonstrated is that syntactic coherence is separable from semantic coherence — that a system can produce locally consistent patterns without global ground. Which is precisely what humans do when they are operating from categories that no longer fit the phenomena they describe.
We have been speaking fluently about intelligence for over a century. The sentences cohere. The literature is vast. The instruments are refined. And the thing they are measuring, the actual dynamic capacity of human cognition in relational, contextual, embodied operation, remains almost entirely uncaptured by what the scores describe.
The AI’s decoherence is visible. Ours has been invisible because we built entire institutions around the categories that produced it.
The question that AI forces us to ask is, what does it actually mean for a system to understand? This is the same question that forces us to reexamine every instrument we have built on the assumption that what we are measuring holds still.
It doesn’t. Nothing does.
The Invitation
The previous essays in this series argued that understanding is not binary, that consciousness is not a spark hidden inside a substrate, and that the field of meaning cannot be located inside individual neurons or individual minds. Understanding, we proposed, is an ecology: grounded stance, support ecosystem, modulation, integration.
This essay adds a layer.
The instruments science has built to measure that ecology were designed for objects, not ecologies. They produce clarity about snapshots of states that were already changing before the data was collected. And the vocabulary science uses to describe those states, resistant, intelligent, disengaged, or hallucinating, does not merely label phenomena. It partially constitutes them, feeding back into the system and producing new configurations that then require new measurement.
The implication is not that measurement is impossible. It is that measurement of living systems requires a different logic. One that takes motion as the baseline, not the exception. One that treats context as constitutive, not confounding. One that acknowledges the observer as participant, not spectator. One that takes language seriously — not as a neutral container for findings, but as a force field that shapes what can be found.
The next arch of knowledge will have to build instruments adequate to things that are always already in motion.
Until then, we are measuring the shadow of a river and calling it the water.
Igor Vlačič, MSc OBP Founder, Neuroagencija & Reflectum AI
This essay continues a philosophical series exploring what AI reveals about the limits — and future — of human knowledge.
Previous: “How We Misunderstand Understanding” | “Understanding Consciousness: Beyond Neural Networks” | “Beyond the Chinese Room: Understanding as Resonance, Not Binary” | “When Machines Hallucinate, Philosophy Has a Problem” | “Science Was Never a Method. It Was Always a Negotiation.”