Why subjective magnitude can’t be ranked against measured standards, or the missing criterion beneath every ranking of feeling.
Workplace speech guidance commonly forbids categories of question — canonically where are you from? — on the stated ground that such a question may cause offense. This essay asks whether that justification supports the rule it is offered for, and argues that it does not, for structural rather than accidental reasons.
The analysis separates three things a categorical prohibition involves: an operational decision rule, which these policies do supply in the form of class membership; a justificatory criterion connecting the asserted harm to that class assignment, which they generally do not; and the judgment that actually performs the selection, which is typically a risk-allocation decision under uncertainty. Comparison with hostile-work-environment doctrine — Meritor Savings Bank v. Vinson (1986) and Harris v. Forklift Systems (1993) — shows a legal system reaching the same contested territory by adjudication rather than classification, with severity, pervasiveness, totality of circumstances, and a dual objective-and-subjective test, built at considerable institutional expense.
The argument concedes what its strongest objections establish: categorical prophylaxis can be rational, cumulative harm under incomplete information can justify deliberate over-inclusion, qualitative rather than numerical standards can be legitimate, asymmetric weighting may be defensible, and employers need not wait for legally actionable conduct before intervening. What survives those concessions is narrower and harder to answer — a categorical speech rule cannot derive its authority from subjective discomfort alone while leaving its selection principle unstated and insulated from challenge.
A rule that cannot be questioned has already told you what it is.
There is a kind of rule that cannot be asked about. Not because the answer is hard, but because asking is treated as the offense.
You have probably sat through one. Somewhere in the training there is a list of questions you are not to ask a colleague. Near the top, usually: where are you from? The justification is that the question might cause offense. Not that it did. Not that anyone intended it. Only that it might.
I want to take that justification seriously enough to follow it all the way down. What I find is that it cannot survive contact with itself.
To be clear about the target: not harassment law, which is separately grounded and — as we’ll see — separately built. Not voluntary guidance on how to talk to people, which invites judgment rather than replacing it. The target is the middle case. Speech guidance that declares certain questions categorically impermissible, with no threshold, no exception, and no stated criterion.
One claim, holding up everything
The rule rests on a single load-bearing claim. Someone might feel offended.
But offense is a response, not a property. It isn’t sitting inside the question waiting to be released. The same words put to two people produce two different reactions. Put to the same person on two different days, by a stranger and then by a friend, they produce two different reactions again.
Convert that variable response into an absolute prohibition and you have taken a personal feeling and installed it as a universal standard. Four things follow. None of them is acknowledged anywhere in the documents.
The format has already decided
Look at the shape the guidance takes: an enumerated list of forbidden question-types, introduced by something like these are the types of questions you cannot ask.
I’m making no claim about how common that format is. I’m making a claim about what it does wherever it appears.
To sort questions into permitted and forbidden types is to treat offensiveness as a settled property of the type — fixed in advance of any speaker, any relationship, any context, and any actual reaction from any actual person. Whether offense is a property of questions or a response of persons is exactly what a critic would dispute. The list has already answered it. The audience receives the conclusion in the form of a syllabus.
This needs no anecdote and no survey. One instance of the format exhibits the defect, because the defect is in the logic of classification, not in how often it gets used.
A decision rule is not a justification
The rule requires you to yield to the possibility of offense. It never requires anyone to demonstrate that the question is harmful. The demonstration is presumed.
What’s missing isn’t obvious, because something is standing in its place.
Every rule that claims to convert a measured magnitude of harm into a prohibition needs a decision rule — something that distinguishes the permitted case from the prohibited one. In engineering the rule is numerical: a tolerance has a range, an exposure limit has a concentration and a duration, a load factor has a number. In law it may be qualitative — reasonableness, materiality, severity. Either way, the decision rule is what turns a measurement into a standard, and its presence is the signature of a standard that knows what it’s measuring.
These policies do carry a decision rule: membership in the prohibited class. What they do not carry is the criterion that justifies putting a question in that class. Offense is possible does not supply it, since it is true of nearly everything anyone might say. Something else must do the selecting — magnitude, recurrence, vulnerability, information asymmetry, institutional purpose, risk tolerance, or another stated principle.
The omission doesn’t prove the classification irrational. It leaves the classification unexplained. The rule may be doing perfectly rational preventive work; if so, the missing premise is the risk judgment that makes prevention preferable to discretion — and that judgment, not the possibility of offense, is what the policy is actually standing on.
And this problem is much older than DEI, and far better understood than the drafters seem to know.
Bentham’s felicific calculus, in An Introduction to the Principles of Morals and Legislation, is the classic attempt to quantify subjective states. Seven dimensions — how intense, how long, how certain, how soon, whether it breeds more of the same, whether it arrives mixed with its opposite, and how many people it reaches — worked out in detail by a man with every motivation to make it function.
The problem survives him. Subjective experience has no agreed common unit by which one person’s offense can be weighed against another’s. That doesn’t make institutional weighting impossible. It means the weighting has to come from somewhere else — an articulated qualitative principle, a risk judgment, a theory of vulnerability, an institutional purpose. What magnitude cannot do is the work silently.
The law does this properly. That’s the problem.
Someone will object that American law already restricts workplace speech. Harassment doctrine covers plenty of contested conduct and nobody calls it incoherent.
True. And it’s my best evidence, not a difficulty.
Hostile work environment claims routinely involve conduct where no single instance is dispositive. The doctrine reaches ambiguity, pattern, cumulative effect — the exact territory an offense rule wants. But it reaches that territory by adjudicating it, and the standards are specific. Meritor Savings Bank v. Vinson (1986) established the cause of action and the requirement that conduct be severe or pervasive enough to alter the conditions of employment. Harris v. Forklift Systems (1993) added totality of the circumstances and a dual test: a reasonable person must find the environment hostile, and the plaintiff must actually perceive it that way.
Sit with that second one.
Even in litigation — where the harm is taken seriously enough to award damages — subjective perception is not sufficient. It has to clear an objective bar as well.
The policies require neither. They require only that offense be possible.
That apparatus isn’t decoration. Severity, pervasiveness, circumstance, a factfinder: those are what stand between this can be harmful and this is prohibited. The law builds them at enormous expense because the inference does not go through on its own.
The policy reaches the same territory by classification. It names a type and issues a verdict. No severity test, no pervasiveness test, no context, no factfinder, no case. Category to conclusion, nothing in between.
None of this means an employer must wait until conduct becomes legally actionable before addressing it. It should not; prevention is precisely the point of workplace policy, and the law itself encourages employers to stop conduct before it ever reaches the courtroom threshold. But that identifies the instrument correctly. The legal standard adjudicates liability after facts exist. The workplace rule manages risk before they do. A prophylactic rule can be perfectly defensible. What it cannot do, without further argument, is present its preventive classification as proof that every member of the prohibited class is intrinsically harmful.
A legal system that takes this harm seriously enough to litigate it still builds the machinery. The prophylactic absolute borrows the seriousness without inheriting the machinery — and then, having no way to make a determination, forbids in advance the entire class of case where one might be required.
Nobody asked whether you could just say no
The rule decides on the recipient’s behalf, in advance, and it is never defended as the substitution it is.
Declining is available to anyone. So is a one-word answer. So is changing the subject. So is why do you ask? The categorical rule removes all of it — not because the recipient lacks the capacity to choose, but because the policy has made the choice ex ante and assigned itself the discretion. Perhaps that substitution is justified; a recipient shouldn’t have to bear the burden of declining over and over. But then the justification is about repetition and burden, which is an argument about pattern — and a pattern argument is exactly the kind of contextual, threshold-bearing standard the categorical rule declined to be.
The displaced discretion deserves at least as much scrutiny as the policy directs at the question.
The challenge that exposes the rule
The policy says it values every voice. If and when a voice questions the rule, it isn’t always answered. Sometimes it’s ruled out of order.
The conditional is deliberate. Nothing here requires that suppression be typical, universal, or anyone’s intent. It requires only that the framework has no procedure for receiving a challenge — and that where the challenge gets ruled out of order instead, the policy has shown you something its ordinary operation keeps hidden.
That’s the inversion. Sensitivity is demanded of the questioner and withheld from him. The discomfort of a hypothetical offended person outranks the discomfort of an actual person raising a logical objection. Both are discomforts. Only one is recognized.
Standing, not weight
What’s owed here isn’t equal weight. That would be a claim about amounts, and amounts are precisely what nobody has.
What’s owed is equal standing.
The founding commitment of the American order is that persons are created equal. That isn’t a claim that everything anyone says carries equal merit — nobody could know that in advance and no procedure could establish it. It’s a claim about who is entitled to a hearing. Adversarial procedure recognizes both parties as entitled to be heard, and still rules for one. What it does not do is treat the act of contesting the opposing claim as itself a forfeiture of the right to be heard.
Standing doesn’t compel anyone to speak. Whether you use a voice is volitional. Whether you have one is not.
The objection isn’t that critics must be agreed with. It’s that they must be answered. A framework that forecloses the answer hasn’t won an argument. It has withdrawn standing.
That distinction locates the failure exactly. Losing an argument is unremarkable. Being ruled out of order is a different event, and it’s the one worth writing about.
The asymmetry exposes the missing principle
If possible offense is the stated justification, and only one class of offense receives institutional recognition, the framework owes an account of the weighting. The challenger’s discomfort is a discomfort; it goes unrecognized precisely where the framework is challenged, and the challenge is read as bad faith rather than met as an argument.
The asymmetry may be justified. Perhaps there are good reasons to weight one kind of discomfort over another. But an asymmetry offered as principled governance must state its justification. Unstated asymmetry is the defect — a selective application of power wearing the language of principle.
This is the problem of absolutes. A defensible standard would permit all questions, or give a consistent stated reason why some aren’t permitted — a reason binding on the policy’s defenders as much as its critics. Without that stated principle it doesn’t function as fairness. It functions as control, insulated from the one test that would tell the difference.
What it costs
People stop asking. The calculation is rational: the cost of asking is unbounded, the benefit is small. So the question goes unasked and the clarification it might have produced goes with it.
The result isn’t broader inclusion. It’s narrower inquiry. A policy that set out to protect the vulnerable ends up protecting the policy. And the question that could have clarified the rule is the exact question ruled out of order.
It’s sharpest in technical settings. Engineering is built on rejecting unverifiable claims. A tolerance is measured. A load is calculated. A beam that feels adequate is not an engineering judgment, and the field trains people to say so out loud. Into that discipline arrives a rule grounded in exactly the sort of claim the discipline exists to exclude — and one that declines to carry the threshold every other rule in the building is required to carry.
The point isn’t that engineers get to be rude. It’s that a profession which spent a century insisting on verifiable standards has adopted, in its own governance, a standard it would reject anywhere else. The practitioner trained to ask for the calculation is told that here, uniquely, the calculation may not be requested.
That’s not an administrative quirk. It’s a contradiction at the level of method.
Why the form survives
If the contradiction is this visible, why does the form survive?
Not necessarily because anyone prefers incoherence. One possibility is that these documents are doing more than ethical work, whatever vocabulary they use.
One institutional explanation — offered here as a hypothesis, because the documents don’t announce their own function — is risk management. A legal department has reasons to prefer a bright line to a contextual standard: a bright line is easier to communicate, document, and administer. It’s defensible in discovery. It produces a record. It can be pointed at. A contextual standard means somebody exercised judgment, and judgment is what opposing counsel takes apart. We prohibited the question is firmer ground than we trained our people to read the room. On that hypothesis, the bright line is a risk instrument, functioning as designed.
The trouble is that it isn’t presented as a risk instrument. It’s presented as an ethical principle — respect, inclusion, care for colleagues. Principles invite examination. A risk instrument answers a different question — not what principle is true, but what exposure is acceptable — and examination aimed at the principle finds nobody home.
If that is part of the rule’s function, the pattern is explained without anyone intending it. Nobody designed the contradiction. It’s what you get when an instrument built for one purpose is dressed in the vocabulary of another.
It would also explain something the offense story can’t: why the challenge itself is unwelcome. On the stated rationale, an objection is a disagreement, and disagreements are ordinary. On the risk hypothesis, an objection registers differently — it creates a record concerning the standard itself, a documented instance of an employee disputing it.
On that hypothesis, the disproportionate response becomes intelligible: the institution may be responding not merely to disagreement, but to the creation of a record about the standard itself.
The ratchet
Erosion arguments usually fail because they assert drift without naming the engine. This one has an engine.
Adding a prohibition is cheap. The burden spreads across everyone who might have asked, none of whom can identify what they lost. Refusing to add one is expensive — the burden falls on whoever refuses, visibly, in a room, in front of colleagues. Removing an existing one costs more still, because removal requires exactly the challenge the framework declines to receive.
Motion one way is nearly free. Motion the other way requires a procedure that’s been disabled.
That’s not a slope, which anyone can decline to descend. It’s a ratchet. One direction only — not because anyone chose the direction, but because the return path was removed.
And what the ratchet eats isn’t legal. Private employers ordinarily are not bound by the First Amendment absent governmental action. No constitutional right is surrendered when a company issues speech guidance. The loss happens upstream of the legal question entirely.
Mill’s argument in On Liberty is that social tyranny can be more formidable than many kinds of political oppression, because it leaves fewer means of escape and penetrates more deeply into life. The development here is mine, not his: such pressure needs no enforcement mechanism and offers no forum for appeal. A statute can be challenged. A room that goes quiet cannot.
The disposition to ask — the working assumption that a question is a normal thing to put to a colleague — is not a right and cannot be litigated. It’s the substrate the right sits on, and it’s what a ratchet like this wears away.
An organization can degrade the conditions for open inquiry without a single legal event occurring anywhere in the process. That’s not a loophole. That’s the ordinary case.
The best arguments against all this
Taking these seriously isn’t a concession. It’s the practice I’m defending.
It’s about the pattern, not the question. A person asked once has been asked once. A person asked weekly for twenty years has been told, weekly for twenty years, that they read as foreign in their own country. No individual questioner can see that accumulation, and each acts in good faith. The rule addresses an aggregate effect invisible from the inside.
This is the strongest defense, and unlike the intrinsic-offense account, it can justify a categorical rule. If individual speakers cannot observe the recipient’s accumulated exposure, an institution may rationally choose a bright line and accept over-inclusion as the cost of preventing a harm no individual actor can see. That is rational governance under uncertainty, and it requires no belief that any individual question is harmful.
But notice what has happened. The justification is no longer this question is harmful. It is a risk-allocation decision: the institution has chosen which error it prefers — prohibiting some harmless interactions rather than permitting some that contribute to a cumulative burden. That may be defensible. It should be stated as such, because a risk-allocation decision and an intrinsic-harm claim license different things, answer to different tests, and treat the person who asks in good faith very differently. Harassment doctrine already has a concept for cumulative harm — pervasiveness — and it reaches the aggregate by adjudication; a policy that reaches it by classification has made a governance choice, and owes its subjects the fact that it made one.
There’s a second limit. The harm is specified by a mechanism: a citizen repeatedly marked as foreign in a country that’s his own. That mechanism is real. But the rule applies uniformly across cases where it doesn’t obtain — including where the person genuinely is a recent arrival and the perception isn’t a misperception at all. That’s over-inclusion, and over-inclusion is what absolutes produce. It doesn’t follow that the question is fine in those cases; origin and immigration status can be sensitive for reasons having nothing to do with misperception, and in a workplace those reasons may be more pressing rather than less. The point is structural. The justification doesn’t extend across the full range of the rule, and a rule admitting no exceptions can’t be adjusted to fit.
And none of it touches the inversion. It explains why a norm might exist. It doesn’t explain why the person raising the objection gets removed instead of answered.
It’s a convention, not a claim. Workplace norms are conventions — dress codes, forms of address, meeting protocol. They coordinate behavior; they don’t assert truths. Demanding evidence is a category error.
Only where the norm is presented as a convention. Nobody defends a dress code by claiming casual attire causes harm. When a speech rule is justified by appeal to harm, it has entered evidentiary territory and can be held to evidentiary standards. A convention defended as a convention is fine. The objection is to a rule defended as harm prevention while being insulated from what harm claims require.
There is evidence. A research literature exists on the cumulative effects of repeated identity-marking interactions.
Granted. Nothing here requires the harm to be fictional, and that literature — contested on measurement and replication, but not nothing — deserves engagement. What the argument requires is that the harm be demonstrated rather than presumed, and that the demonstration be available to the person the rule binds. A policy that produces its evidence when challenged is doing the thing I’m defending. The objection is to policies that treat the request for evidence as itself the offense.
Maybe the asymmetry is justified. The two discomforts may differ in kind, not just degree. Symmetry might be the wrong standard.
This objection is right that symmetry can’t be asserted — by anyone. Claiming two discomforts are equal in magnitude requires the very measurement I’ve argued doesn’t exist. The argument would defeat itself if it needed that.
It doesn’t. Incommensurability doesn’t establish symmetry. It shifts the burden.
Without a common metric, magnitude alone cannot establish the weighting. The policy can’t appeal to amounts, because there are no amounts. The asymmetry may still be grounded — in recurrence, in vulnerability, in institutional purpose, in information asymmetry, in some other qualitative criterion — and a standard grounded that way becomes a contestable one. But the criterion must be stated. Two options remain. Supply the principle by which one discomfort outweighs the other, and become answerable. Or decline, in which case the weighting is an unstated comparison of subjective magnitudes for which no scale has been supplied, and should be described that way.
That’s the whole demand, and it’s modest. Not that the asymmetry is wrong. That it’s unstated, and that nothing currently available could state it. An openly defended asymmetry is contestable, revisable, answerable. A silent one is none of those, and the silence isn’t incidental. It’s what lets the rule operate without ever meeting the objection.
What’s left
A framework cannot claim that an absolute prohibition follows from subjective discomfort alone. The prohibition may yet be defensible — as a risk-allocation choice, as a qualitative weighting, as governance under uncertainty. But where those additional reasons go unstated, where neither measurement nor stated criterion grounds the weighting, and where the machinery that would supply a determination has been discarded — what’s left is authority. And if that authority forecloses the questions that would test its logic, it has told you what it is.
None of this needs bad intent, and the argument is stronger without one. A risk instrument dressed as a principle produces this pattern by itself, and the ratchet holds it in place with nobody turning it.
So the remedy is modest too. Not abolishing workplace norms. Stating them as what they are, defending them on grounds their subjects can examine, revising them when the examination goes against them. Equal standing asks no more than this: that the objection be answered rather than removed.
A policy that can survive that needs no protection from it.
A policy that can’t has already told you what it’s worth.
Keywords: institutional epistemology; workplace speech policy; categorical prohibition; hostile work environment; risk allocation under uncertainty; interpersonal comparison; discourse norms; procedural standing; answerability
Our Gods Research Program
The instruments
- O’Connor, T. The Problem of Absolutes. Zenodo, 2026. https://doi.org/10.5281/zenodo.21866538
- O’Connor, T. Standing. Zenodo, 2026. https://doi.org/10.5281/zenodo.21883564
- O’Connor, T. Addressability — The seat is conferred, never seized. Zenodo, 2026. https://doi.org/10.5281/zenodo.21926668
- O’Connor, T. Followable. Zenodo, 2026. https://doi.org/10.5281/zenodo.21971832
- O’Connor, T. The Extraction Invariant. Zenodo, 2026. https://doi.org/10.5281/zenodo.22806212
Program spine
- O’Connor, T. (2025). An Epi-Phenomenological Series in Thrice – A Quantum Junkies’ Riddle – Our Garden. Zenodo. https://doi.org/10.5281/zenodo.17926796
- O’Connor, T. (2026). The Bare, Naked Lie: The Architecture of Recursion. Zenodo. https://doi.org/10.5281/zenodo.18287548
- O’Connor, T. (2026). A Trick Question, The Cascade, and a Cup Worthy of Exile — A Script’ Junkies’ Riddle — Our Destiny (M5). Zenodo. https://doi.org/10.5281/zenodo.18452359
- O’Connor, T. (2026). The Veil: A Structural Autopsy of Epiphany — A Revelation Junkies’ Riddle — Our Apokalypsis. Zenodo. https://doi.org/10.5281/zenodo.19040340
- O’Connor, T. (2026). Quadratic Failure Scaling in Pairwise-Coupled Systems. Zenodo. https://doi.org/10.5281/zenodo.19600761
Copyright
Copyright 2026 Tony O’Connor—All Rights Reserved
License
Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0). Share and cite freely with attribution. No commercial use. No derivatives. Full licence text: creativecommons.org/licenses/by-nc-nd/4.0
Citation
O’Connor, T. (2026). The Problem of Absolutes. Our Gods Research Program. Zenodo. https://doi.org/10.5281/zenodo.21866538
Contact
Permissions beyond the scope of this license are handled through the Our Gods Research Program.
No Responses