We say: “this is good”, “this is weak”, “she’s an 8/10”, “he’s a loser”, “the student is a C-grade kid”, “the employee is average”. One of the quiet victories of bad education is that it taught us to confuse evaluation with reality. Language gives us convenient shortcuts, but those shortcuts can be intellectually brutal. They reduce a person, a situation, a work, or a behavior to a single axis. And then we wonder why social life looks like a ranking, a spreadsheet, a selection process, and a stock exchange of humiliations. Let’s name it. This is rankizm: the cult of a single measure, served as common sense.

Rankizm: the cult of a single measure

Rankizm is the belief that a person, a situation, a work, or a behavior can meaningfully be reduced to a single number, a single place in a hierarchy, a single evaluation on a “better–worse” axis. It sounds neutral, and that is precisely why it is dangerous. It doesn’t show up as an ideology. It shows up as common sense. “I just want to know who’s the best.” “I’m just being objective.” “The market just decides.”

Underneath that “just” sits a whole machinery of assumptions: that the axis exists, that there is only one, that comparison along it is fair, and that whoever stands higher is simply… higher. Rankizm eats school, the job market, social media, dating, sport, art, politics, and our conversations about ourselves. It is the fuel of the stock exchange of humiliations.

In short: rankizm is not measurement. It is measurement without humility.

Monoaxiology, or the world reduced to a single axis

The problem isn’t just that people evaluate badly. The problem is deeper: we have been trained to evaluate one-dimensionally. Philosophy will call it monoaxiology: the belief that something can meaningfully be placed on a single axis from “bad” to “good”, from “stupid” to “smart”, from “ugly” to “beautiful”, from “1” to “6”. Everyday language has a shorter name: rankizm. These are two names for the same phenomenon: the academic and the colloquial. Neither is the natural state of reason. Both are cultural, administrative, and educational practices that were pressed into us. And neither can really be “found” in the world, because the world doesn’t look one-dimensional. It’s the human who tries to make it so, in order to manage, classify, and account for it faster.

The critique isn’t a grudge against axes as such. Some phenomena are one-dimensional: temperature, mass, reaction time, shot accuracy, reading speed. There a 1D scale is adequate to the thing. Rankizm begins only when the one-dimensional axis is stretched onto a phenomenon that does not fit on it: intelligence, an employee, a partner, a work of art, a life. The critique of monoaxiology doesn’t attack the thermometer. It attacks the use of a thermometer where a whole panel of measurements is needed.

Evaluations are not facts

Evaluations are not facts. This sentence should hang over the entrance to every school, newsroom, company, and family home. A fact might read: the student solved 7 out of 10 problems. The evaluation reads: the student is “good” or “weak”. A fact might read: someone was late three times this month. The evaluation reads: they are irresponsible. A fact might read: a person has certain features of appearance, behavior, or speech. The evaluation reads: they are “attractive” or “unattractive”. Between fact and evaluation there is always a measure, a criterion, a perspective, and a purpose.

This doesn’t mean everything is arbitrary. That would be the second mistake: to go from naive objectivism to lazy relativism. Measures may be relative, but they don’t have to be arbitrary. Temperature depends on the scale, but that doesn’t mean water freezes according to our mood. A test score depends on how the test is constructed, but that doesn’t mean every score says the same thing. A project evaluation depends on the criteria, but those criteria can be transparent, sensible, useful, and fair. Maturity isn’t about pretending there are no evaluations. It’s about knowing what we are actually measuring, and what we are not.

One digit instead of a map

For years, school trained us in a different way. One digit. One grade. One place in the hierarchy. The 1–6 scale is administratively convenient, but psychologically and cognitively it tends to be primitive. It says less than it pretends. A student who gets a “3” doesn’t learn whether they have trouble understanding concepts, focusing, parsing the instructions, working at speed, dealing with test anxiety, holding things in working memory, motivating themselves, finding a learning style, or relating to the teacher. They only learn they are “a three”. An efficient tool for the electronic grade book. A poor tool for understanding a human being. And a flawless school of rankizm, because it teaches that at the end there is always one number.

Descriptive feedback is not a soft indulgence. It’s a denser form of information. Instead of saying “4”, you can say: you grasp the main idea but confuse the concepts; you have good intuition but jump to conclusions too quickly; you argue in interesting ways but don’t distinguish example from proof; you write boldly but need structure. Such feedback doesn’t lock a person inside a number. It shows a map of development. A map works differently from a stamp. It shows direction rather than handing down a verdict.

Give someone a hammer, they see nails everywhere

The most dangerous part is that the school logic of scales doesn’t stay in school. It migrates everywhere. Twelve years of training in one-dimensional evaluation then works like a hammer in the hand. To someone with a hammer, everything looks like a nail. This is how a rankist grows: a person trained on a 1–6 scale who starts evaluating everything on a 1–10 scale: coworkers, friends, their own body, partners, children, themselves. Not because they are bad. Because it’s the only tool they were given.

This is where 1–10 scales in relationships come from, attractiveness rankings, “type” tables for partners, the stock exchange of humiliations on social media. These aren’t innocent jokes, nor are they the problem of one gender. They’re the transfer of administrative logic into a language that should never have been administrative. When both sides reach for the same hammer, the discussion turns into a contest over who reduced whom to a score first. And the source isn’t biology or some “corruption”. The source is a tool we were never taught to put down.

So the error isn’t in who evaluates whom. The error is that we treat one-dimensional evaluation as the default mode of looking at a person at all. Instead of teaching how to recognize value, we teach labeling. Instead of cultivating attention, we drill reaction. Instead of saying: “look at what this situation is made of”, we say: “evaluate quickly”. And whoever evaluates quickly usually evaluates shallowly. This is what rankizm looks like in practice, not as an open ideology, but as a reflex.

Education as a factory of default settings

The crux is in education, but not only in teachers. It’s easy to say: the educators failed. It’s harder to add: the system failed, the one that turned educators into clerks of an overloaded institution. If a society pays teachers as though their work were an add-on to the real economy, it shouldn’t be surprised that education often comes down to mere survival. A person who is chronically undervalued, poorly paid, and administratively squeezed rarely has the space to build nuanced, multidimensional feedback systems. They give as much as someone can give when the system tells them: “you have a mission, but we don’t have a budget for that mission”.

And yet education is exactly where we have to start, because education produces the default settings of culture. If for twelve years you hear that the grade matters most, you then start evaluating everything. Yourself, others, your body, status, intelligence, money, attractiveness, success. One axis. One ranking. One place. This is rankizm in its purest form. Intellectual poverty dressed up as order.

Language models on an axis: rankizm in the age of AI

The most honest illustration of rankizm today is the language model. “Which model is the best?”. You hear this question everywhere, from conferences to Discord groups. We try to answer with benchmark tables: HumanEval, MMLU, MATH, GPQA. Numbers from 0 to 100, columns, sorted by average. This is fresh rankizm, produced by people who consider themselves technically sophisticated, while reproducing exactly the same structure as the electronic grade book from primary school.

Meanwhile a language model has more than a dozen dimensions that no single benchmark captures together: precision, long context, conversational usefulness, calibration of uncertainty, safety, cost, latency, quality in a specific language, behavior under pressure, long-term stylistic coherence, hallucination density, code quality in a specific domain. The question “which is best” assumes an axis that isn’t there. Just like the 1–10 scale in relationships, it reduces a multidimensional object to a score. Memes like “model X told a user to go to sleep” gather hundreds of reactions because they fall into the same groove: a stamp instead of a map.

Benchmarks themselves are not the problem. They are diagnostics. The problem begins when a score table starts pretending to be a full theory of intelligence, usefulness, or safety.

The Polish public debate about AI offers two recent examples of rankizm in action. Olga Tokarczuk tries to draw a distinction between AI as research-assistant and AI as ghostwriter; the internet ignores the distinction and compresses the whole thing into the headline “Nobel laureate writes with AI”. Anthony Aguirre, in conversation with Tomasz Kawecki, talks about the configuration space of states and the alignment problem; the title of the conversation reads “AI is evil by nature”, literally contradicting his thesis. A map turned into a stamp. A multidimensional position turned into a label. The same thing that happens to a student getting a “3”, only performed by editors who should know better. (I develop this case in a separate post on the Aguirre–Kawecki interview.)

Multidimensional axiology

The antidote is not “no evaluation”. The antidote is multidimensional axiology, the opposite of rankizm. We don’t live in a 1D world where the only measurement is temperature. We don’t live in a 2D world where human intelligence has the surface area of a square meter. A human is not a line segment, a chart, or a KPI. Even in business, the simplest financial result doesn’t say everything: you can have revenue without margin, growth without stability, efficiency without trust, speed without quality, success without meaning. Let alone in a human life.

Multidimensional axiology means we ask not only “is this good?”, but: good in what respect? For whom? On what timescale? At what cost? By what measure? Does it strengthen agency, or only obedience? Does it deepen understanding, or only improve the score? Does it build a relationship, or only a hierarchy? Does it describe reality, or brutally simplify it?

Binary decisions and the aggregation trap

A fair objection: multidimensionality sounds beautiful, but life forces discrete decisions. A company has to hire one of two candidates. A court has to rule guilty or not guilty. Procurement has to pick one supplier. These are binary decisions. There is no “we hire him at 73%”. So isn’t aggregating many dimensions into a single decision simply necessary?

It is necessary. It is also mathematically constrained. Kenneth Arrow proved in 1951 that no fair procedure exists for aggregating several orderings of preference into a single decision-ordering while simultaneously satisfying a handful of very modest reasonableness conditions (Arrow’s impossibility theorem). In other words: every method of compressing multiple dimensions into one decision loses something. There is no perfect ranking. There is only a ranking we have chosen to accept, aware of its losses. In plain language: every final ranking is also a sacrifice of information.

The conclusion isn’t that aggregation should be abandoned. The conclusion is that aggregation is secondary to understanding, not primary. First I see the candidate multidimensionally. Only afterwards, if I must choose, do I consciously compress, knowing what I lose. Rankizm reverses this order: it starts with compression and pretends the compression is a fact. A mature decision starts with understanding and treats compression as a tool to be set down again afterwards, so that the next decision can again be made out of many dimensions, not from an inherited score. If a scale is meant to support decisions, it should also be judged by the quality of the decisions it produces.

The narrow dignity of a good scale

There is, however, a narrow dignity in a well-built scale. The critique of rankizm should not be confused with a critique of measurement. A good scale can say something meaningful, but only if it knows its own limits.

A scale becomes intellectually legitimate when it is attached to a clear construct, a defined purpose, a transparent method, and a known margin of error. It must say: this is what I measure, this is why I measure it, this is how the score was produced, this is how reliable the score is, and this is what the score does not mean.

In that sense, a strong methodology can make a single scale useful. A reading test may tell us something about reading performance under specified conditions. A reaction-time measure may tell us something about speed. A credit score may tell a lender something about repayment risk. A benchmark may tell us something about a model’s performance on a particular task.

But even then, the score remains an instrument, not an ontology. It describes performance in relation to a purpose; it does not reveal the essence of a person, a mind, a work, or a life. Good methodology does not remove reduction. It makes reduction explicit, controlled, and honest.

A one-dimensional scale is not wrong because it is one-dimensional. It becomes wrong when it forgets that one-dimensionality is a reduction. Measurement can be humble. Ranking rarely is. Rankizm begins when a scale forgets that it is a tool and starts behaving like a verdict.

Evaluation as a map, evaluation as a label

The point isn’t to abandon evaluation. The point is for evaluations to stop pretending to be absolute facts. An evaluation can work like a map or like a stamp. It can describe a relation between goal, criterion, and observation, or it can stick to a person like a label. The first says: “here is where you are, here is where you could go, this is worth recognizing”. The second says: “this is what you are”. The difference is not that one is “good” and the other is “bad”, because that is precisely the trap we would fall back into, rankizm under another name. The difference is that the first opens several dimensions, while the second closes a person inside a single point.

The cult of a single measure is convenient for institutions but costly for people. It simplifies management but complicates life. It eases selection but destroys understanding. It offers an appearance of objectivity but often hides cognitive laziness. Because it’s easiest to say: “a three”, “a seven”, “average”, “top”, “bottom”. It’s harder to say: “I see several dimensions at once”.

A scale can be a map. The problem begins when we treat the map as the territory, and the human being as a point on an axis.

Seeing several dimensions at once

And that is precisely what we should be learning: to see several dimensions at once. In a person. In work. In love. In education. In public debate. In language models. In ourselves. Leaving rankizm behind doesn’t mean we stop evaluating. It means we stop pretending that one measure is enough, and where aggregation is necessary, we do it consciously and as a step that follows understanding, not one that precedes it. Because perhaps the greatest mistake is not that we evaluate. The greatest mistake is that we evaluate as if reality were simpler than it is.

FAQ: rankizm in questions

What is rankizm?
Rankizm is the belief that a person, situation, work, or behavior can meaningfully be reduced to a single number or position in a hierarchy. It is the cult of a single measure served as common sense. A particular ranking can be a tool; rankizm is an ideology. It holds that one axis suffices to describe phenomena that are inherently multidimensional.

Is rankizm the same as monoaxiology?
They are two names for the same phenomenon. Monoaxiology is the academic term (mono- = one, axiology = theory of value); rankizm is its colloquial, memetic counterpart. In academic citation: monoaxiology. In everyday speech: rankizm.

Are all rankings bad?
No. Some phenomena are genuinely one-dimensional: temperature, mass, reaction time, shot accuracy. A 1D scale is adequate there. Rankizm begins only when a one-dimensional axis is stretched onto a multidimensional phenomenon: intelligence, a human being, a work of art, a partner, a film.

What about decisions that force a single choice (a candidate, a supplier, a verdict)?
Aggregation is necessary but mathematically constrained. Arrow’s impossibility theorem (1951) shows that every compression of many dimensions into one decision loses something. Aggregation is secondary to understanding, not primary. First we see multidimensionally; only afterwards, if we must, do we consciously compress.

What is the alternative to rankizm?
Multidimensional axiology: describing a phenomenon along several axes at once, without pretending one is more important than the rest. Instead of asking “is this good?”, we ask: good in what respect, for whom, on what timescale, at what cost, by what measure.

Cite this article