“That which is altogether just shalt thou follow, that thou mayest live, and inherit the land which the LORD thy God giveth thee.”
Pursue that which is wholly just, that you may live and possess the land which the LORD your God gives you.
Deut 16:20
I. An examination no one announces
A state does not sit its examination when everything is working. Then there is no examination — there is routine, and routine tells us nothing about ourselves. The examination begins on the day something goes wrong: when a document proves defective, a decision misguided, money spent without cover, and the person responsible for it stands in the light.
No one announces this examination. There is no board, no date, no question paper. There is only a sequence of events in which institutions reveal what they are truly capable of — and they do so within hours, before anyone has had time to prepare a position. That is why the result of this examination is far more credible than any declaration of values. A declaration is written at leisure. A reflex reveals itself under pressure.
This text is about the criteria by which that examination is marked. Not about whether the state makes mistakes — it does, and it will continue to, because it is made of people. It is about how one can tell that, after a mistake, it has acted justly.
The claim I put forward here is a single one, and it can be stated in one sentence: the maturity of a state is decided not by the severity with which it responds to error, but by the order in which it does so.
II. The doubled word
The verse from the Book of Deuteronomy reads smoothly in Polish, and in doing so loses its edge. In the Hebrew original the key word falls twice: tzedek, tzedek tirdof — “justice, justice shalt thou seek”. The English translation renders this doubling as an intensification: that which is altogether just — that which is just to the very end, without remainder.
For centuries the interpretive tradition has read the repetition as pointing to two things at once: the end must be just, and the way must be just. It is not enough to want justice. One must seek it by a means that itself fulfils justice. The first tzedek speaks of the outcome. The second — of the method.
It is worth noting the company in which this verse stands. Two sentences earlier there is talk of appointing judges and officers “in all thy gates”, and immediately afterwards of the prohibition on wresting judgement, on partiality and on taking gifts, for “a gift doth blind the eyes of the wise”. This is not a passage about moral sentiments. It is a passage about institutions: about who judges, by what procedure, and what that procedure is meant to guarantee. Justice is described here as a construction, not as an intention.
And from this follows the whole of the rest of this essay. A state that reaches the right result by the wrong way does not half-pass the examination. It fails it, because the way is named in that sentence as a second, separate condition — not as the means of fulfilling the first.
III. Order is an institution
There is an order that distinguishes accountability from a hunt, and it is simpler than its gravity would suggest. First, the facts are established. Then, responsibility is assigned. Finally, consequences are drawn.
This order looks self-evident, and that is precisely why it is so often reversed. The reversed version runs: first we point to a person, then we look for guilt to fit them, and the facts are established along the way, as the narrative requires. Each of these three steps is defensible in itself — after all, someone was at fault; after all, things must be explained; after all, the facts are being established. Only the order is defective. And that is enough for the result to cease to be just, even if by chance it proves accurate.
The reason is cognitive, not legal. When a person is pointed to first, everything that comes afterwards begins to be read as confirmation. An ambiguous document becomes evidence. A clumsy explanation becomes an evasion. Silence becomes an admission. Not because anyone is acting in bad faith — but because a mind that already has its answer stops asking questions and starts collecting confirmations. Cognitive psychology described this mechanism long ago; it was named confirmation bias. To it is added a second, still stronger one: once we know the outcome, the past seems obvious, and every error that led to it looks like an error that “could have been seen”.
Institutions were invented precisely against this. Not because the people in them are better, but because procedure enforces an order that reflex will not hold of its own accord. The presumption of innocence, the right to be heard, the separation of the one who gathers evidence from the one who judges, the two-tier system of appeal — all of these are devices for holding the order in place. They are neither civilisational ornament nor a privilege for the guilty. They are the only known method of ensuring that the result does not depend on who spoke first.
The Book of Proverbs puts it bluntly: he that answereth a matter before he heareth it, it is folly and shame unto him. Let us note that the shame falls here on the one who answers, not on the one being judged. That is an exact description of what happens to a state that passes sentence before it has heard.
IV. Four things that must be distinguished
Justice begins with distinctions, and a crowd does not make distinctions. This sentence is not a judgement on the crowd but a description of its nature: a collective acts quickly, and distinguishing is slow.
Systems in which error costs lives — aviation, medicine, energy — have learned to separate four things that public opinion fuses into a single word: “guilt”.
The first is honest error: a person acted as best they could, in the conditions they found, and erred as any competent person in their place would have erred. The response is to repair the conditions, not to punish the person.
The second is at-risk behaviour: someone departed from the standard without perceiving any danger in doing so, because departure had become the norm in their surroundings. The response is correction, training and the restoration of the standard — together with the question of why the norm had been allowed to drift.
The third is negligence: the duty was known, it was possible to perform, and it was not performed. Here personal responsibility is appropriate, and its absence would be an injustice towards those who do perform their duties.
The fourth is wilful action: conscious disregard of an obvious risk, or the deliberate misleading of others. Here a sanction is not merely permissible but necessary, because its absence destroys the meaning of all the other distinctions.
These four categories are not a softening of responsibility. They are its precondition. A state that treats them as one is not stricter — it is less precise, and imprecise severity always strikes whoever happens to be standing nearest, not whoever was most at fault. European law, indeed, knows this logic and applies it where human life is at stake: in the system for reporting occurrences in civil aviation, an employee who reports an incident is protected from consequences, and that protection ceases in exactly two cases — wilful misconduct and gross negligence, understood as a manifest and serious disregard of an obvious risk. The distinction was written into the regulation because without it no one would report anything.
V. Proportion, or the choice of instrument
Proportionality is sometimes understood as leniency. That is a misunderstanding. Proportionality is a matter of matching the instrument to the weight of the case — and the well-chosen instrument is sometimes more severe than the one reached for by reflex.
Every mature legal order has at its disposal a ladder of instruments. At its foot stand correction and repair: an audit, a demand for the return of funds, the amendment of a document, the completion of a procedure. Higher up — administrative and disciplinary liability, with sanctions that are real yet commensurate. At the very top stands criminal law, which the continental legal culture treats as ultima ratio: the last resort, reached for when instruments of a lower order are not sufficient.
This ladder does not exist for the convenience of wrongdoers. It exists because every instrument has a side cost, and the side cost of criminal law is the highest of all: the proceedings alone, regardless of the verdict, take from a person their work, their trust and their peace of mind, and from an institution its capacity to explain its own error calmly. To reach for it first is not severity. It is the skipping of every rung on which the matter could truly have been put right.
The Book of Leviticus speaks of this in language that sounds surprisingly technical: ye shall do no unrighteousness in judgment, in meteyard, in weight, or in measure; just balances, just weights shall ye have. Justice is described here through instruments of measurement. Not through anger, not through compassion, not through a sense of being in the right — but through a balance that shows what is. The state has no other task: it is to be a balance that lies neither one way nor the other.
This must be said plainly, for otherwise the whole argument would be one-sided. A balance that weighs short is as false as one that weighs heavy. Indulgence towards genuine abuse is not mercy — it is a wrong done to all those who did their work honestly. Public money is always subject to account; asking about it is a citizen’s duty, not an assault. The whole difficulty lies in making the accounting thorough rather than loud.
VI. A truth one must want to hear
There is a resource without which none of the above works, and which can be neither bought nor commanded: truth brought forward voluntarily.
A state learns of its own errors in two ways. Either from a catastrophe, or from a person who speaks of them beforehand. The first source is infallible and very costly. The second is cheap and depends entirely on what befalls those who make use of it.
And here a law operates that is worth remembering, for it knows no exceptions: the amount of truth reaching an institution is inversely proportional to the cost borne by the one who brings it. This is not a moral observation but a description of how reasonable people behave. If reporting one’s own error ends worse than keeping silent about it, silence becomes a rational strategy — and ceases to be a question of character.
That is why the most mature systems build protected candour as an institution, not as a virtue. Civil aviation did so decades ago, recognising that a pilot who is afraid to report an incident is a greater danger than a pilot who caused one. Whistleblower protection entered the Polish legal order in September 2024 and rests on the same assumption: a person who discloses an irregularity is a good to be protected, not a risk to be silenced. Even criminal law knows this logic — it provides for impunity for one who repairs the damage before proceedings are set in motion. That is the statutory expression of the principle that repair is worth more than punishment.
The practical consequence is simple and applies to every organisation, not only to the state. Whoever, of their own accord and before detection, reports their own error and presents a plan of repair may not suffer consequences more severe than one whose error was detected. The principle fits into a single sentence of an internal order. Yet it changes everything, because only it makes honesty pay — and a value that costs more than it yields will not survive in any institution for longer than one generation of staff.
VII. The one who told the truth
The best illustration of this examination is a story that Europe knows, though it usually remembers a different hero in it.
In the autumn of 1894 the French army accused Captain Alfred Dreyfus of treason. He was convicted in a closed trial in December, publicly degraded in January of the following year, and in April transported to Devil’s Island. The case seemed closed. The state had first pointed to a man and only afterwards fitted the evidence to him — including evidence that had been forged.
In 1896 the head of intelligence, Lieutenant Colonel Georges Picquart, established that the document on the basis of which Dreyfus had been convicted was the work of another officer. He was not a man free of the prejudices of his age, nor had he been appointed to rescue anyone. He simply saw what he saw, and judged that he was not permitted to conceal it. His superiors ordered him to keep silent. He persisted. In December 1896 he was removed from intelligence and transferred to a regiment in Tunisia. In 1898 he himself was accused — of forging the document that had convinced him of the convicted man’s innocence — and arrested.
This is the purest description of the cost of truth in an institution that does not wish to hear it: the man who was right went to prison for being right.
What followed, however, is equally important, because it shows what a passed examination truly is. In January 1898 Émile Zola published in “L’Aurore” the open letter “J’accuse…!”. In August of that same year the officer who had forged the evidence confessed, and the next day took his own life. In 1899 the Court of Cassation quashed the verdict of 1894 and ordered a retrial; the retrial convicted Dreyfus once more, whereupon the President pardoned him — a pardon that ends a punishment but does not restore innocence. Full rehabilitation had to wait until 11 July 1906, when the Court of Cassation annulled that verdict too. Two days later Dreyfus returned to the army. Picquart was promoted to brigadier general, and in October 1906 took office as Minister of War.
Twelve years from accusation to rehabilitation. France did not pass this examination in 1894, and did not pass it in 1899. It passed it in the end in 1906 — not because it had never gone astray, but because it proved capable of annulling its own verdict and of restoring to office the man it had earlier punished for the truth.
The same pattern repeated itself a hundred years later in the United Kingdom. Between 1999 and 2015, on the basis of a defective IT system, more than nine hundred sub-postmasters were prosecuted; some two hundred and thirty-six went to prison, and at least thirteen suicides have been documented. For years the institution maintained that the system was beyond reproach, and told each successive accused person that they were the only one to have experienced a problem. The judicial finding that the system contained errors came in 2019, the independent public inquiry later still, and in May 2024 Parliament passed an Act quashing en bloc the convictions handed down in these cases and launched a compensation scheme.
I do not recall these two stories to show that states go astray. I recall them to show what passing the examination consists of: the existence of institutions of reversibility. A court of cassation that annuls its own verdict from years before. A parliament that by statute wipes out a thousand convictions. Promotion for the one who was punished for telling the truth. None of these things happens by itself. Each requires someone to acknowledge publicly that the state was wrong.
VIII. Reversibility instead of infallibility
From this follows a criterion that I propose as the principal measure of this examination.
A mature state is not an infallible state. There has never been one and there never will be; the promise of infallibility is, moreover, the beginning of every tyranny, for an institution that cannot be wrong needs no oversight. A mature state is a reversible state: one that has built mechanisms for undoing its own decisions and is able to use them against itself.
Reversibility has three layers, and they are worth distinguishing, because confusing them is a source of illusions.
The first is legal reversibility: the possibility of setting aside a judgment, reopening proceedings, making good the damage. It exists in every European legal order, ours included, and in itself guarantees nothing — because the mechanism may stand unused.
The second is institutional reversibility: the capacity to examine one’s own process, to describe its causes and to change the procedure. Here the measure is not a dismissal but a recommendation with a deadline and the name of the person responsible for implementing it. A dismissal without a change of procedure leaves the system exactly as it was, and is in essence the cheapest way of avoiding reform.
The third is moral reversibility: the readiness to say “we were wrong” in the first person plural, without looking for someone onto whom it can be shifted. It is the hardest, because it cannot be written into a regulation, and the most important, because without it the two preceding layers remain a dead letter.
A state that has all three can make a very grave mistake and remain worthy of trust. A state that has only the first is sometimes law-abiding on paper and unbearable in experience.
IX. Seven marks of mature accountability
The examination can be passed deliberately, because its criteria are known. The seven marks below can be applied to any institution — a ministry, a hospital, a university, a company, an editorial office, a foundation — and none of them requires a new statute.
First: facts before judgement. The first question is “what happened”, not “who is to blame”. The second is “why did it happen”. Only the third concerns the person.
Second: the injured before the perpetrator. Before any decision on consequences is taken, the question must be asked who has been harmed and what they need. An accounting that forgets this is concerned only with itself.
Third: the instrument matched to the weight. Correction, then administrative liability, and criminal law last of all. Skipping the rungs does not speed up justice — it skips the places where something could have been put right.
Fourth: a day’s delay in decisions about people. No personnel decisions in the first twenty-four hours after emotions erupt, with one exception: when people, evidence or continuity of operations must be protected, one acts at once — but by means of a protective measure, not a verdict.
Fifth: the separation of fact-finding from the imposition of consequences. The one who investigates does not adjudicate. This is one of the oldest principles in the history of law, and one of the most frequently abandoned once a case becomes high-profile.
Sixth: informational restraint while proceedings are under way. An institution that reports on its own proceedings before it has established anything inflicts a hardship for which it does not yet have any legal title. The presumption of innocence binds not only the court.
Seventh: recommendations instead of closure. A case does not end with a personnel decision, but with a list of changes with deadlines and a public account of their implementation a year later. It is the only mark by which it can be seen from the outside that an institution has really learned something.
To these seven is added one test for each of us, because no society that does not want these things will sustain any of them. Three questions before pressing “publish”: am I writing about a deed or about a person; have I checked, or am I repeating; do I want repair or a spectacle. Thirty seconds. That is the price of the difference between a citizen and an audience.
X. The measure
Let us return to the doubled word. Tzedek, tzedek — justice and justice once more: what we wish to achieve and how we intend to achieve it. The verse does not allow these two things to be separated, and it does not envisage a situation in which a righteous end justifies the way.
This is the whole examination of the state, expressed in a single ancient sentence. It does not ask whether we managed to avoid error. It asks whether we sought justice by a means that was itself just.
Everything I have written above is an unfolding of this one condition. Order is a just means. The distinction between the four forms of fault is a just means. The ladder of instruments is a just means. Protecting the one who brings the truth is a just means. The capacity to annul one’s own verdict is a just means. Each of these things is slower and less spectacular than reflex. Each is also the only thing that distinguishes accountability from revenge.
We do not need a state without errors. We need a state that, after an error, keeps to the order, matches the measure to the weight, protects the one who told the truth, and is able to reverse its own decision when it proves unjust.
Such an examination is always sat at the same moment: when the case is high-profile, opinion has already formed, and doing the right thing will bring no one an immediate reward. The state is not then judged by what it proclaims. It is judged by the order in which it acts.
Justice — and justice once more — shalt thou seek.
An essay in the Just Council series: on the conditions of just judgement.
Sources referred to in the text: Deut 16:18–20; Lev 19:15 and 19:35–36; Prov 18:13. Protection of those reporting occurrences in civil aviation and the limits of that protection — Regulation (EU) No 376/2014, Article 16. Whistleblower protection in Polish law — the Act in force since 25 September 2024. Impunity where damage is repaired before proceedings are initiated — Article 297 § 3 of the Polish Criminal Code. Chronology of the Dreyfus affair after the timeline of the Musée d’art et d’histoire du Judaïsme; biography of Georges Picquart — encyclopaedic sources. The Horizon scandal and the Act of May 2024 quashing the convictions — documentation of the British public inquiry and the text of the Act.
