Procedure Turns Difference Into Deviance

We have a deep instinct to punish what doesn’t fit, and much of modern management and statistics is built on it.

From the statistical worship of the average to the managerial obsession with consistency, we took variance, which is just human difference, and recoded it as a vice. Grief is “disruptive” and calm is “professional”; unpredictability gets pathologized while serenity gets held up as a moral ideal.

The logic runs deep and it isn’t an accident. We learned to moralize the mean. When statisticians invented the average man (l’homme moyen) they didn’t just describe a tendency; they made an ideal, and deviation became defect.

Then Taylorism trained bosses to treat inconsistency as inefficiency. Workers got disciplined until spontaneity itself looked like misbehavior.

Our emotional lives followed. Bit by bit, the right to be variable, to move, emote, or err, got withdrawn. Institutions learned to enforce calm.

Bureaucracies standardize judgment, algorithms flag deviation as risk, schools reward the compliant mind over the curious one, and predictive policing literally criminalizes variance by treating statistical difference as a threat.

The language of “quality assurance” and “risk management” makes deviation the implied enemy, the work to be done.

What you get is a fragile order, smooth on top and brittle underneath. The more a system suppresses what’s left over, the more easily it shatters when reality shows up anyway.

Variance is feedback rather than dysfunction.

Some of it should still be corrected. A surgeon’s rising complication rate and a dosing pump that drifts are variance too. The line runs between variance in people and situations, which a good system learns from, and variance in the system’s own performance that harms the people it serves, which it should reduce. Disciplining a grieving worker for an uneven week mistakes the first for the second.

Where employers use the Bradford Factor, the step from an uneven week to a write-up is arithmetic. The score is the number of separate absences in a rolling year, squared, times the total days absent, on the theory that frequent short absences disrupt work more than a few long ones. When the British Medical Association surveyed NHS trusts and health boards in 2024, 17 still used it, with trigger points ranging from 80 to 300.1

One academy trust’s attendance policy shows each step.2 Absences related to maternity, pregnancy, disability, critical illness or an accident at work are normally discounted, and every other sickness absence counts. At a score of 150 the manager checks the record for those discounts, and if the score still stands, calls a formal meeting with at least 14 days’ written notice. A Stage 1 warning stays live for six months. Two more absences while it’s live lead to Stage 2, and two more after that to Stage 3, where dismissal is considered. Bereavement isn’t on the list of discounts.

Say a teaching assistant at one of the trust’s schools loses her father in the autumn. Over the next months she calls in sick five times, six days in all, most of them single mornings when she can’t face the classroom. Her score is 5 × 5 × 6 = 150. A colleague off for two weeks after knee surgery loses more days and scores 1 × 1 × 10 = 10. She gets a letter calling her to a formal attendance meeting, and he doesn’t.

The letter keeps two numbers: how many times, and how many days. It drops why she was off, that the absences cluster after a death and will likely stop, and whether any work went undone. The trust holds a return-to-work discussion after every absence, so her reason is already on file, and the score has nowhere to put it. The BMA’s objection to the method is that it “cannot take into account reasons for absence.”1

Living systems need friction, noise and unpredictability to stay adaptive: evolution runs on mutation, ecosystems run on imbalance, and thought itself runs on contradiction. An organization that treats every deviation as a fault loses the information deviation carries, until ordinary disturbances look like crises and ordinary emotions look like threats.

In a workplace, what we call “stability” often means grief and anger have been pushed out of sight.

Decriminalizing variance means treating difference as a signal rather than a threat, and in an attendance policy the mechanism for that already exists. The discount list is where the policy decides which variance it will learn from instead of punish. Disability got onto this one through the Equality Act 2010, and adjusting trigger points for disability is a recognized reasonable adjustment.1 Bereavement could go on the same list. Or the trigger could open a conversation about the reasons the return-to-work discussions already recorded, before anything counts toward a warning. In emotional life, the same change means letting anger, confusion and uncertainty coexist with reason instead of being disciplined out of it.

Systems last longer when they repair what breaks than when they repress what varies.

In practice, decriminalizing variance means rules that change when they keep colliding with ordinary human difference, instead of punishing the people who collide with them.

The smoothness, the calm and the averages we take for order can hide a lot of suppressed variance. When a workplace or a classroom looks perfectly calm, ask what was disciplined out of it to get there.