Who Guards the Guardians?

A security institution is a special problem of power

Security, policing, intelligence and defence are not ordinary services. Their agents may, in certain circumstances, restrict movement, collect sensitive information, use coercion, intrude on privacy or control resources that ordinary members of a community do not possess. The same institution that reduces one danger can therefore create another.

The OSCE consequently treats democratic political control of military, paramilitary, internal-security, intelligence and police forces as an indispensable element of security. That insight matters for THY-REALITY as well: more power requires more verifiability, not more blind trust.

This does not mean every guard, police officer, commander or investigator is suspect. It means the system must not depend on personal virtue alone. A sound institution can expect most people to act competently while still containing safeguards for error, conflicts of interest, pressure, corruption and deliberate abuse.

The worst time to invent oversight is after the first major abuse. If a security function is created with rules only about what it may do, but not who checks the use of power, how it is recorded, where complaints can go and how a mandate can be revoked, the institution begins with an open governance debt.

“From Self-Defence to Federative Collective Defence” already required common command to have a predefined scope, duration and route of revocation. This article extends the same logic to the coercive function as a whole. Oversight is not decorative machinery added to an otherwise effective institution; it is part of the design.

Emergency conditions should therefore not automatically erase every safeguard. Some procedures may need to move faster during immediate danger and some operational information may temporarily remain restricted, but there should still be a record, later review and a route to determine responsibility.

There is no single guardian of the guardians: oversight must be layered

If the answer to concentrated coercive power is one even more powerful overseer, the problem has merely moved. DCAF therefore distinguishes among internal control, executive or managerial control, legislative or representative oversight, courts and legal mechanisms, independent expert bodies and public oversight of the security and intelligence sector.

THY-REALITY does not copy a state institutional diagram from this. It retains the principle of overlapping safeguards: no single point should be the sole source of legitimacy, information and sanction. Where possible, different overseers should see different aspects of the problem and provide checks on one another.

A small community may implement these functions with fewer bodies than a large state, but the functions remain similar: internal accountability, independent complaint, scrutiny of money and equipment, legal or procedural review and enough public information for the community to know whether the system works.

The organisation itself can often detect a problem fastest. Clear lines of responsibility, written powers, access logs, dual approval for sensitive acts, recusal where interests conflict and supervisory review can prevent many abuses before they require an external investigation.

UNODC anti-corruption work also identifies four-eyes controls and rotation in especially sensitive posts. These should not become dogma. Rotation is most useful where long-standing personal relationships create a serious risk of corruption or capture; excessive rotation elsewhere can destroy expertise and institutional memory.

Internal control also has a structural weakness: the organisation is reviewing itself. It is necessary but insufficient. If every complaint ends inside the same chain of command that is being complained about, formal oversight is not yet independent oversight.

Independent oversight needs real powers, not an impressive title

UNODC guidance on police accountability emphasises an independent external body with a stable mandate, sufficient funding, transparent appointment, capacity to handle complaints and regular public reporting. DCAF likewise describes ombuds institutions as channels for receiving complaints, investigating maladministration and strengthening accountability in the security sector.

For THY-REALITY the decisive functional question is: can the overseer actually reach the information, people and records needed to do the job? A commission that can only ask the supervised institution to choose what it will reveal remains dependent on the goodwill of the body it oversees.

The oversight power also needs a boundary. An overseer should not take over operational command merely because it checks legality or integrity. Its role is to discover, require explanations, trigger procedures, recommend or order what its defined mandate permits, and enable correction — not to become a parallel secret commander.

Defence and intelligence sometimes legitimately require that information not be public in real time: a source's identity, an operational plan, the location of vulnerable infrastructure or an ongoing investigation can lose value or endanger a person if disclosed too early.

It does not follow that secret activity may remain unreviewed. DCAF specifically highlights the tension between secrecy and accountability and the need for appropriate oversight bodies to have access to classified information. The real alternative is not 'everything public' or 'nothing reviewable', but controlled access and later public accountability to the extent disclosure does not create a new danger.

In practice this can mean graded access, secure records, logs of viewing and alteration, time limits on secrecy, periodic classification review and publication of aggregate data and findings when operational details cannot safely be released.

Use of force must leave a trace

When an institution uses force, especially firearms or other potentially lethal force, an oral explanation is not enough. UN Basic Principles require effective reporting and review; in cases of death, serious injury or other grave consequences they also require access to independent administrative or prosecutorial review and an independent process for affected persons.

That matters in small and federative systems too. Force that does not have to be recorded is force that is extremely difficult to constrain. A record of time, reason, method, consequences, participants and authorisation allows later review of necessity, proportionality, repeated patterns and the quality of training.

Serious consequences require stricter review. A death or grave injury should not be left only to the immediate colleagues of the guardian involved. Human-rights standards on potentially unlawful deprivation of life emphasise an independent, impartial, prompt, thorough, effective and sufficiently transparent investigation.

Hierarchy can be necessary in defence and security for speed and coordination. Precisely for that reason it must be clear that an order is not a moral or legal magic shield. UN Basic Principles establish superior responsibility where commanders knew or should have known of unlawful force and failed to act, and they reject manifestly unlawful orders as a complete defence.

For THY-REALITY this connects directly to Awakening into Natural Law: a person does not hand over all personal responsibility when accepting a uniform, office or order. Institutional discipline can be legitimate without demanding moral blindness.

A system should therefore also protect a person who lawfully refuses a manifestly unlawful order or reports unlawful force. If obedience is the only safe path, internal accountability has already failed.

An internal warning needs a safe exit

Many abuses are first seen by colleagues: an accountant notices an unusual contract, an operator an unjustified database access, an officer repeated excessive force, a soldier an unlawful order. If the system exposes the person who reports it to dismissal, transfer, harassment or isolation, future warning signals will remain hidden.

Council of Europe whistleblower standards emphasise internal and external reporting channels and protection against direct and indirect retaliation. This article therefore requires at least two safe routes: one inside the organisation and one outside the chain of command.

This does not mean every report is automatically true or that all classified information should be made public. It means a good-faith report based on reasonable grounds should not become professional suicide. The reporting person also remains subject to fair procedure and responsibility for deliberately false accusations.

This article faces another problem that transparency alone cannot solve: long tenure in a sensitive function can create personal networks, dependencies and conflicts of interest. UNODC in 2025 identified staff rotation in vulnerable positions as one practical measure for disrupting links between public officials and organised crime, while the OECD stresses disclosure and management of conflicts of interest and revolving-door risks.

THY-REALITY turns this into a risk-based, not universal rule. Some functions need longer terms for independence and expertise; others benefit from limited terms, task rotation or cooling-off requirements. The goal is not perpetual movement but preventing a sensitive position from becoming personal property.

Where a person holds power over large contracts, secret databases, surveillance, seizures, investigations or appointments, stronger rules on interests, gifts, financial links, family conflicts, later employment and recusal are justified.

The overseer must also be overseen

The independence of an overseer is valuable only if it does not mean unaccountability. A body that can investigate others, access classified material or trigger sanctions also exercises power. It therefore needs a clear mandate, transparent appointment, conflict-of-interest rules, a budget trail, limits on access, reasoned decisions and a route for review of its own conduct.

This is where the old question 'who guards the guardians?' changes form. We are not looking for the final person at the top of a pyramid. We build circular or networked accountability in which no institution checks everything and none is completely outside review.

The Green Book can serve here only as a supplementary intellectual source: its useful intuition is resistance to oversight of power becoming a permanently separate caste. This article does not adopt its political model; it filters that intuition through contemporary mechanisms of civilian oversight, independent complaints, rotation, public reporting and fair procedure.

Oversight must not disable security

Excessive or badly designed oversight can create the opposite problem: a guardian who needs five slow approvals for every urgent decision may fail to protect anyone in a crisis. This article therefore does not advocate the greatest possible quantity of oversight, but oversight proportionate to the risk and consequences of the power.

Some measures need prior approval; others may require a tightly bounded emergency mandate followed by strict ex-post review. Some information can be public immediately; some only after an operation; some only to an independent overseer. What matters is that the exception is neither empty nor indefinite.

Effectiveness and accountability are therefore not natural enemies. A system that learns from incidents, detects corruption patterns and removes incompetent or abusive personnel may be more effective over time than one that hides its own failures.

Practical audit: who guards the guardians?

Before a community or federative network grants someone a significant security, investigative or command function, it can ask these twelve questions:

1. What is the precise mandate? Which interventions does it permit and which does it exclude?
2. Who records the use of power? Is there a verifiable trail of decisions, access and force?
3. Who conducts internal review? Is responsibility in the chain of command concrete?
4. Who is independent of that chain? Where can an individual or employee complain?
5. Does the external overseer have real access? To the people, records, data and places it needs?
6. How is secrecy handled? Who can see classified information and when is secrecy reviewed?
7. What happens after force is used? When is an automatic independent investigation required?
8. Can a person safely refuse a manifestly unlawful order or report abuse?
9. Where are the conflicts of interest and capture risks? Is recusal, rotation or a term limit needed?
10. Who checks money, equipment and contracts? Are budget flows and procurement reviewable?
11. Who oversees the overseer? Does it also have a mandate, term, rules and route of review?
12. When is the system reassessed? Do safeguards change when risk or technology changes?

The audit cannot guarantee that abuse will never occur. Its purpose is more realistic: reduce the chance that abuse occurs without a trace, remains hidden, becomes routine or turns into an unpunished rule through institutional loyalty.

“Violence, Organised Crime and Community Security” showed that a community needs real security. “From Self-Defence to Federative Collective Defence” showed that it can expand that capacity through federative cooperation where necessary. This article adds the final link of this arc: the capacity for coercion must always travel with a capacity to oversee coercion.

That means internal controls and external independence, records and review, protection of secrecy without zones of total opacity, protection for reporting, personal responsibility for unlawful orders, risk-based rotation and oversight of the overseers themselves. No single mechanism is enough; the combination constrains power.

This closes the basic justice, security and defence arc from “After Harm: Restitution, Protection and the Limits of Punishment” to this article. “Monopoly, Plutocracy and the Concentration of Economic Power” will open a related but different problem: what happens when functional power becomes concentrated not mainly through weapons and official coercive functions, but through monopoly, property and economic dependency?