Has Justice been Served? [The Repair 2] : post 7.2/9
This is part 7.2 (post 7.2/9) of a series on Justice.
Here, we continue our exploration of Repair from where we left off in 7.1, where we talked about "What is Repaired?", and move onto the "Who?" questions.
- ה -
2. Who is Repaired?
Because humans are deeply social, an Act rarely injures only a single node. The blast radius dictates who requires Repair:
(1) The primary Recipient:
The individual directly affected by the Act.
This is the original Recipient in the basic mechanism:
Actor → Act → Recipient
Their Repair may involve:
- material restoration;
- recognition of dignity;
- recovery of honor or face;
- safety;
- acknowledgement;
- compensation;
- restored agency;
- psychological closure.
This person should be the first object of Repair because they experienced the original disturbance.
However, as we have repeatedly found out, in modern Institutional Justice systems, they are often left out of the formal process of Repair, forced to pursue costly (time and money) alternative routes (civil court), or informal remedies (clan or tribe).
(2) The relational Recipient (family, friends, clan, tribe):
The people connected closely enough to the primary Recipient that the Act also changes their condition.
This may include:
- family;
- friends;
- dependents;
- caregivers;
- colleagues;
- clan or tribe;
- religious community;
- others who share identity, obligation, or vulnerability with the Recipient.
They may suffer:
- grief;
- fear;
- financial loss;
- caregiving burdens;
- damaged standing;
- inherited retaliation;
- loss of Trust;
- disruption of family or community life.
They are not merely witnesses to another person’s injury. They become Recipients of the consequences produced by the Act.
A murdered person is the primary Recipient.
The children who lose a parent, the spouse who loses income and companionship, and the family drawn into years of litigation are relational Recipients.
(3) The Institutional Recipient:
This one is Foucault's twist. The Institution responsible for governing the relationship may consider itself disturbed.
The Act may challenge:
- the authority of the court;
- the legitimacy of the State;
- the reliability of a professional body;
- the capacity of the police to protect;
- the military chain of command;
- the credibility of an election;
- the Trust placed in a school, church, company, or government.
Institutional Repair may require:
- enforcement;
- correction of procedure;
- removal of corrupt officials;
- public accountability;
- restoration of operational capacity;
- renewed demonstration that its rules apply;
- reform of the Institution itself.
But the Institution is not necessarily an innocent Recipient.
It may have enabled the Act, amplified it, concealed it, or failed in its Duty to prevent it.
In such cases, the Institution is both:
a damaged object requiring Repair;
and
an Actor responsible for performing Repair.
(4) The abstract Recipient:
The wider society may be disturbed even where most individuals were not directly affected.
A public Act may alter shared expectations about:
- safety;
- truth;
- equality;
- whether the powerful are accountable;
- whether Institutions can be trusted;
- which people count as legitimate Recipients;
- what conduct society is willing to tolerate.
This is not because society possesses a group-mind.
The social disturbance exists separately across many individuals:
- fear within one person;
- distrust within another;
- anger within another;
- a changed expectation within another.
When enough of these individual disturbances converge, they form a social condition.
Social Repair may therefore involve:
- public Truth;
- norm reaffirmation;
- commemoration;
- education;
- institutional reform;
- public acknowledgement;
- prevention;
- restoration of confidence.
(5) The Actor
This one is surprising, but obvious once considered.
In most Justice systems, there exists some need to Repair the person who committed the Act, alongside the many people affected by it.
The nature of that Repair depends upon whether the original Act belonged to Positive or Negative Justice.
Justice is not limited to punishing bad Acts. It also includes properly recognizing good ones.
A person who performs a Positive Act may be ignored, exploited, falsely condemned, or punished by the very Institution that should have recognized them. Repair may therefore require:
- recognition;
- reward;
- exoneration;
- restoration of honor or face;
- correction of the historical record;
- or reversal of an earlier judgement.
The threshold is often exceptionally high. Merely fulfilling an ordinary Duty may not create a claim for special recognition. Sacrifice, danger, or action beyond what could reasonably have been expected often becomes the evidence that distinguishes the Hero, Martyr, or Saint from someone who simply performed their Role.
In Positive Justice, the Actor may therefore require Repair because the response to their good Act was itself unjust.
Negative Justice presents a different problem.
Here, the Actor is the person who committed the harmful Act. Why should the Justice system concern itself with repairing them?
The predominant method of punishment in modern society, incarceration, creates several practical and moral reasons.
a. Incarceration is usually temporary
Most incarcerated people will eventually be released.
This means confinement cannot be understood only as removal. At some point, the Actor will return to a family, neighborhood, workplace, and society.
The relevant question becomes:
What kind of person is being returned?
If the Institution confines the Actor without addressing the beliefs, habits, circumstances, or internal Norms that produced the Act, it may have only suspended the disturbance.
The Actor may leave prison carrying the same Reference Set with which they entered:
Violence remains justified.
The rule is illegitimate.
My only mistake was being caught.
In that case, imprisonment may have punished the past without repairing the conditions that threaten the future.
Rehabilitation attempts to return the Actor to a state in which they can rejoin society without producing another Recipient.
Recidivism is therefore not merely evidence that an individual committed another Act.
It may also indicate that the Justice system successfully confined the Actor but failed to Repair them.
b. Incarceration is expensive
In 2023, the average annual cost of incarcerating one person in the United States was approximately $44,090, or about $121 a day. While precise costs vary by jurisdiction, the broader reality remains: incarceration requires massive, continuing public expenditure.
Permanent confinement is therefore an exceptionally expensive way to avoid asking whether a person can change.
Where safe reintegration is possible, Actor Repair may reduce:
- future harm;
- repeated prosecution;
- repeated incarceration;
- financial cost;
- disruption to families;
- and the number of future Recipients.
Repairing the Actor is therefore not simply an act of generosity toward the person who committed the wrong.
It may be one of the most practical forms of protection available to society.
c. Actor Repair is already embedded in the Justice System
Modern Justice systems already recognize this need through parole. A parole board does not ask only:
"Has the Actor spent enough time in prison?"
It also asks whether some form of Self-Repair has occurred.
The board may consider whether the Actor:
- recognizes the Act;
- accepts responsibility;
- understands the harm caused;
- expresses genuine remorse;
- has addressed the conditions that contributed to the Act;
- has demonstrated different behavior;
- possesses a realistic plan for life after release;
- and can return without creating an unacceptable risk of repetition.
The board is attempting to determine whether the same person, operating through the same internal Reference Set, still exists in the same condition that produced the original Act.
This process is imperfect and highly subjective. An Actor may learn to perform the language of rehabilitation without having changed internally. Another may change substantially without possessing the language or emotional performance the board expects.
Nevertheless, the existence of parole reveals an important principle already embedded within Institutional Justice:
Time served and Repair achieved are not necessarily the same thing.
The sentence may establish when release becomes legally possible.
The Judgement of Self-Repair helps determine whether return has become socially possible.
d. The Actor remains human
Perhaps the strangest reason is also the highest ideal.
Most people recognize that even someone who committed a terrible Act remains a human being.
They refuse to reduce the totality of a person to the worst thing that person has ever done.
This does not require denying the Act.
It does not require minimizing the Recipient’s injury.
It does not require forgiveness, release, or a belief that every Actor can safely return.
It requires holding two Judgements at the same time:
You are responsible for what you did.
and:
You may still possess the capacity to become someone who would not do it again.
That is an extraordinarily demanding position.
Anger naturally collapses the Actor into the Act:
"She betrayed us; therefore she is only a traitor."
"They harmed someone; therefore they belong permanently outside the moral community."
Repair requires separation without absolution.
The Actor owns the Act.
The Act remains wrong.
The Recipient remains owed.
But the Actor may still possess a future beyond it.
Incarceration itself may make this more difficult. It can expose the Actor to violence, sever family and community relationships, remove ordinary agency, reward behaviors necessary for survival inside prison but destructive outside it, and attach a permanent identity that makes legitimate return nearly impossible.
A Justice system that creates those conditions and then demands successful reintegration may be producing the very failure it claims to Judge.
The obligation is therefore not necessarily:
Society must restore the Actor to the person they were before the Act.
The previous condition may have been precisely what allowed the Act to occur.
Instead:
Justice must provide a legitimate path through which the Actor can accept responsibility, Repair what can still be repaired, alter the conditions that produced the Act, and return to society when return is possible.
The Actor becomes a target of Repair because an unrepaired Actor remains part of the disturbance.
Actor Repair serves several purposes:
- it recognizes Positive Acts that were wrongly ignored or condemned;
- it reduces the possibility of future Recipients;
- it prepares temporarily excluded people to return;
- it limits the social and financial cost of permanent confinement;
- and it preserves the principle that a human being need not remain permanently identical to their worst Act.
Society Repairs the Actor partly for the Actor’s sake.
But it also does so for everyone who must live with the person who eventually returns.
- ה -
3. Who is Responsible for the Repair?
This question is asking:
Where does the debt of Repair land?
That is not always the same as asking who deserves blame.
Moral responsibility concerns who was culpable.
Reparative responsibility concerns who now owes something toward restoring the damage.
One person or Institution may possess all three. But they can separate.
The Actor may be morally responsible but unable to pay.
An insurer may be responsible for compensation without being morally responsible.
An employer may bear legal responsibility because it controlled the conditions of the Act.
A later government may inherit responsibility for a wrong committed before its current officials were born.
An insurer may be responsible for compensation without being morally responsible.
An employer may bear legal responsibility because it controlled the conditions of the Act.
A later government may inherit responsibility for a wrong committed before its current officials were born.
The obligation to Repair may follow from several different relationships to the disturbance:
- causing it;
- participating in it;
- enabling it;
- failing to prevent it;
- benefiting from it;
- guaranteeing against it;
- or inheriting the system and assets through which it occurred.
Case: The Rana Plaza Collapse (Bangladesh, 2013)
On April 24, 2013, the Rana Plaza building in Dhaka, Bangladesh, collapsed, killing 1,134 garment workers and severely injuring 2,500 more. The day before the collapse, massive structural cracks had appeared in the walls, and the building had been briefly evacuated.
When we ask who was responsible, the blast radius of responsibility fragments immediately:
1. Causal and Moral Responsibility
The immediate causal and moral blame landed on the local actors. The building owner, Sohel Rana, had illegally constructed the upper floors to house heavy industrial generators whose vibrations ultimately shattered the concrete. The factory managers were also morally and causally responsible: on the morning of the collapse, they threatened to withhold a month's pay from any worker who refused to go back inside the visibly cracking building.
These individuals produced the disturbance. They were culpable. But they were either imprisoned, bankrupt, or dead. They possessed zero capacity to produce Material Repair for 3,600 devastated families.
2. Institutional Responsibility
The local municipal government and building inspectors had taken bribes to overlook the illegal construction and had failed to enforce the evacuation order. They enabled the Act and failed to prevent it. But the local government also lacked the vast financial resources required to compensate the victims and their dependents.
3. Reparative Responsibility
The debt of Repair ultimately bypassed the direct Actors and landed thousands of miles away, on the international fast-fashion brands (such as Primark, Benetton, and Mango) who contracted the factories inside Rana Plaza.
Did these global brands illegally pour the concrete? No.
Did they stand at the door and force the workers inside? No.
Under traditional criminal law, they were neither causally nor morally culpable for the structural collapse.
But they were structurally tied to the disturbance through benefit. They drove the hyper-aggressive pricing model that forced factory owners to cut safety corners to secure contracts. Because they benefited from the system, and because they were the only entities with the financial capacity to actually perform the Repair, international pressure forced them to assume Reparative responsibility.
Through a mediated international agreement, the Rana Plaza Donors Trust Fund was established. Global brands, many of whom fought fiercely to deny legal liability, ultimately paid roughly $30 million into the fund to compensate the victims and their families.
The collapse proved that in complex systems, the person who breaks the norm (Causal) and the person who acts with malice (Moral) are rarely the same entities who ultimately pay the debt of restoration (Reparative).
(1) The Actor:
The Actor is the primary source of the disturbance and therefore normally bears the first obligation to Repair it.
The basic expectation is:
The Actor may be required to:
- return property;
- pay restitution;
- apologize;
- correct a false statement;
- restore a position;
- perform service;
- accept punishment;
- participate in rehabilitation;
- prevent recurrence.
But the Actor may be unable to complete the Repair.
They may be:
- dead;
- unknown;
- insolvent;
- imprisoned;
- incapable of reversing the injury;
- unwilling to acknowledge the Act;
- or too weak to repair damage far larger than themselves.
The Actor remains responsible, but the Recipient’s Repair cannot always depend entirely upon the Actor’s capacity or cooperation.
And in many cultures, the Actor cannot be cleanly separated from the group to which they belong.
(2) The relational or collective Actor:
The individual may have performed the Act, but they did so as someone’s child, parent, spouse, clan member, employee, soldier, official, or representative. Their identity, resources, protection, and Authority may all come from the group surrounding them.
The injury may also spread beyond the primary Recipient into the Recipient’s own family or community. What began as an Act between two individuals can therefore become an obligation between two relational systems.
This does not necessarily mean that every member of the Actor’s group is morally guilty. It means that reparative responsibility may be distributed through relationships that existed before the Act and continue after it.
The group may become responsible because it:
- benefited from the Act;
- authorized or encouraged it;
- supplied the Actor’s resources or authority;
- protected the Actor from consequence;
- possesses the means to perform Repair;
- or accepts responsibility for the conduct of its members.
In customary systems, the Actor’s family or clan may provide compensation because the individual cannot satisfy the obligation alone. The group then assumes responsibility for controlling its own member and restoring peace with the Recipient’s group.
Modern Institutions use a similar structure under different language. An employer may be responsible for an employee’s conduct. A corporation may answer for the Acts of its officers. A military may answer for the conduct of its soldiers. A State may answer for officials acting under its Authority.
The underlying principle remains:
The individual performed the Act, but the individual did not necessarily act alone.
In certain cultures:
Moral guilt may remain individual, while reparative responsibility becomes relational.
Case 1. Somali xeer: the diya-paying group
Under Somali customary law, the relevant unit is often not the isolated offender but the diya-paying group: a number of closely related families that jointly provide security for their members and jointly pay compensation when one member injures or kills someone from another group. Traditionally, the payment was made in livestock from one group to the other.
The structure is:
Member of Group A harms member of Group B → Group A owes compensation
→ Group B receives compensation → the relationship between the groups is stabilized
The other members of Group A need not have participated in the Act. They pay because membership creates mutual obligations: they protect one another, answer for one another, and bear the cost when one member threatens peace with another group.
This also creates internal control. Because the entire group may be forced to pay for one member’s conduct, the group has a material reason to restrain, discipline, and monitor its own members. One report describes Somali elders eventually refusing to keep paying diya for militia members whose repeated violence had become intolerable.
Here, the relational group is not merely associated with the Actor.
It is the unit through which the Actor is made accountable.
Case 2. Germanic wergild: the kin group pays the kin group
Early Germanic legal systems used wergild, a formally determined compensation for killing or injuring a person. It was paid to the deceased person’s kindred, with the amount often varying according to social rank.
Although the exact allocation varied across periods and societies, the basic mechanism again operated between kin groups:
The Recipient’s side acquires a claim.
Payment could prevent the family of the slain person from pursuing vengeance. The purpose was not merely to enrich the Recipient’s family or allow a wealthy offender to purchase innocence. It transformed a potentially unlimited cycle of retaliation into a defined and terminable obligation.
That is a major function of collective reparation:
It gives the relational Recipient something other than collective revenge to demand from the relational Actor.
This form of collective reparation is not limited to far-off cultures or long-past times. It is alive and well today.
Case 3. Carter v. Reynolds (2003; 175 N.J. 402, 815 A.2d 460)
Alice Reynolds worked for a New Jersey accounting firm and was required to use her personal vehicle when visiting clients. While returning home from an off-site client visit, she struck a vehicle driven by David Carter. Reynolds committed the negligent Act, but Carter also pursued the accounting firm for the resulting damage. The New Jersey Supreme Court held that the firm was liable under the doctrine of respondeat superior because Reynolds had been acting within the scope of her employment: the firm required her vehicle, benefited from its use, and gained the ability to conduct off-site business without providing a company car.
The firm did not drive the vehicle or directly commit the Act. Its responsibility arose from its relationship with the Actor. Reynolds was performing the firm’s work, under conditions created by the firm, for the firm’s benefit. The Court summarized the underlying principle: one who expects to benefit from an Act performed on their behalf may also be required to answer for the injury it causes.
The Western legal system calls this vicarious liability. The negligent employee remains the immediate Actor, but the employer becomes responsible for the Repair because the employee acted as part of the employer’s enterprise.
The Somali clan, the Nuer kinship group, and the American corporation are not identical social structures. But they express the same underlying mechanism:
The individual performs the Act, but the larger unit that stands behind the Actor may inherit the debt of Repair.
The difference between cultures is not whether relational responsibility exists. It is which relationships are recognized as strong enough to carry that debt.
Case 4. Moffatt v. Air Canada (2024; BCCRT 149)
A customer relied upon incorrect information supplied by a chatbot on Air Canada’s website. The chatbot told him that he could apply for a bereavement fare after completing his travel. When Air Canada later denied the refund, it argued that it should not be held responsible for the chatbot’s statements, as though the chatbot were a separate entity responsible for its own conduct.
The Tribunal rejected this argument. The chatbot was part of Air Canada’s website. Air Canada had created or deployed it, placed it before its customers, and benefited from the service it provided. It therefore remained responsible for the information the chatbot communicated and for Repairing the loss caused by that information.
The chatbot produced the immediate statement, but Air Canada was the larger Actor through which the statement acquired Authority.
This is relational responsibility in a thoroughly modern Western form:
The individual employee, representative, automated system, or tool performs the immediate Act.
The case also improves the category itself. “Relational Actor” may be too narrow, because the immediate Actor need not even be human.
(3) The enabler
The Actor may perform the immediate Act, but another person or Institution may make that Act possible, easier, safer, more repeatable, or more difficult to detect.
This is the enabler.
The enabler may provide:
- access to the Recipient;
- money or equipment;
- professional Authority;
- a trusted position;
- a platform;
- secrecy;
- institutional protection;
- failure to investigate;
- failure to report;
- or protection from consequence after the Act.
The distinction from the relational Actor is important.
A corporation may be responsible for an employee because the employee acted on its behalf.
An enabler need not be represented by the Actor at all.
The enabler becomes responsible because it possessed some Power over the conditions of the Act and used that Power poorly, or refused to use it when intervention became necessary.
The mechanism may be active:
The enabler provides the weapon, account, location, information, legitimacy, or access.
Or it may occur through omission:
The enabler receives a warning, possesses the ability and Duty to intervene, but allows the Actor to continue.
A person who merely exists near an Act is not necessarily an enabler.
The stronger questions are:
- Did they know, or possess enough Evidence that they should have known?
- Did they have the Power to reduce or stop the danger?
- Did their action or inaction preserve the Actor’s access to the Recipient?
- Did they protect the Actor after receiving warning?
- Did the Act continue because the enabling condition remained?
The enabler may not share the Actor’s Intent.
They may not desire the injury.
They may even condemn it afterward.
But once their participation, protection, or failure becomes part of the causal pathway, some portion of the debt of Repair may land upon them.
Case 1: The FBI and Larry Nassar
Larry Nassar was the primary Actor.
He used his position as a physician to sexually abuse athletes under the appearance of medical treatment.
But by July 2015, allegations against him had been reported to the FBI’s Indianapolis Field Office.
The Department of Justice Inspector General later found that senior officials failed to respond with the urgency the allegations required. The office made fundamental investigative errors, failed to notify the appropriate FBI office or state and local authorities, and failed to take steps that could have reduced the continuing threat. Eight months later, another FBI office received the allegations but also failed to notify the appropriate authorities. During the period between the first FBI report in July 2015 and the eventual state investigation in September 2016, Nassar continued treating gymnasts and other young people.
The FBI did not commit Nassar’s assaults.
It did not possess his Intent.
It had not created the original disturbance.
But once the allegations reached it, the FBI possessed something the individual Recipients did not:
investigative Power;
access to other law-enforcement Institutions;
and the capacity to interrupt the Actor’s continued access to future Recipients.
Its failure did not transform the FBI into the primary Actor.
It created a second form of responsibility:
Nassar was responsible for committing the Acts.
This distinction matters.
The original Actor creates the danger.
The enabler allows the danger to remain active.
The mechanism becomes:
Actor → Act → Recipient
followed by:
Recipient reports Act → Institution receives Power to intervene
→ Institution fails to intervene → Actor retains access → additional Acts and Recipients
The Institution’s omission becomes a new Act within the larger Event.
In 2024, the United States Department of Justice settled 139 administrative claims alleging that the FBI had failed to conduct an adequate investigation, agreeing to distribute $138.7 million among the claimants. The Department acknowledged that the settlements could not undo the harm but might provide support for continued healing.
The settlement is important because the Reparative responsibility did not land only upon Nassar.
It also landed upon the Institution that had received warning, possessed the Power to respond, and failed to use it adequately.
That debt of Repair included more than compensation.
An enabling Institution may owe:
- public acknowledgement;
- disclosure of what it knew and when;
- preservation and release of Evidence;
- compensation for harm that occurred after intervention became possible;
- removal of officials who failed their Duty;
- procedural reform;
- independent oversight;
- and safeguards against repetition.
The Institution cannot Repair its failure merely by punishing the original Actor.
Nassar’s imprisonment answered what Nassar did.
It did not answer what happened after the FBI was told.
This produces an important principle:
When an Institution possesses both the Duty and the Power to interrupt a continuing disturbance, failure to act can make that Institution responsible for part of the Repair.
The enabler’s share of the debt does not erase the Actor’s responsibility.
It recognizes that the Actor may have been able to continue only because another person or Institution left the pathway open.
The Actor committed the Act.
The enabler preserved the conditions in which the Act could continue.
Case 2. Social Media: the Enabler Without a Moral Direction
The internet did not invent protest, terrorism, hatred, recruitment, or collective action.
It changed their scale.
Before social media, an Actor needed physical proximity, an organization, access to broadcasting, or control over an Institution to reach large numbers of people.
Social media gives ordinary Actors access to:
- an audience;
- coordination;
- repetition;
- targeting;
- social validation;
- rapid transmission;
- and contact with people they could never otherwise reach.
The platform may not share the Actor’s Intent.
It may not even know the Actor exists.
Nevertheless, it supplies Power.
a. The Arab Spring: enabling people to find one another
In 2010, photographs of the battered body of Khaled Said circulated online after he was killed by Egyptian police.
The Facebook page “We Are All Khaled Said” became a place where people could share information about police violence, corruption, and repression. It connected individuals who had separately experienced fear and anger but had no ordinary way to know how many others felt the same.
The page later helped circulate the call for demonstrations on January 25, 2011. Activists used Facebook’s groups, invitations, posts, and updates to plan protests and move people from an online gathering place into the streets.
Facebook did not create the police violence.
It did not create the public anger.
It did not march into Tahrir Square.
What it did was alter the relationship among the Actors.
Before:
Another person is afraid.
Another believes they are alone.
After:
Separate disturbances become a visible social condition.
A population previously divided by distance and fear becomes capable of coordinating an Act.
The platform enabled people to discover that their private Judgements were shared.
It converted:
R′, R′, R′
into something visible enough to act like:
∑R′
This is enabling in a positive sense.
The platform increased the Power of people who possessed little Institutional Authority of their own.
But the same mechanism has no moral direction.
It connects whoever uses it.
b. ISIS: enabling recruitment without proximity
ISIS used social media to distribute propaganda, locate potential recruits, normalize its Reference Set, and connect sympathizers in distant countries with people who could move them into territory controlled by the organization.
Mirsad Kandic, a senior ISIS member, used more than 120 Twitter accounts to spread propaganda and recruitment material. Through online communication, he assisted foreign recruits, including eighteen-year-old Jake Bilardi, with travel to Syria. Bilardi later carried out a suicide truck bombing in Iraq. Kandic also operated a private Telegram market where members offered weapons and military equipment for sale.
The platform compressed distance.
An ISIS recruiter in one country could reach an isolated teenager in another.
The recruiter no longer needed:
- a local cell;
- a physical meeting place;
- a printing press;
- control of a television station;
- or an existing personal relationship.
The platform supplied the meeting place.
Its network supplied possible recruits.
Its messaging system supplied private communication.
Its sharing functions supplied distribution.
Its visible likes, followers, and comments supplied the appearance of social validation.
The immediate mechanism was:
Recruiter → propaganda and personal contact → recruit
But the expanded mechanism was:
Recruiter → platform → amplified message → identified recruit
→ continued contact → travel → violent Act → Recipient
Twitter did not write Kandic’s propaganda.
Telegram did not form Kandic’s Intent.
But the platforms gave that Intent a reach it could not otherwise have possessed.
The original Actor supplied the purpose.
The enabler supplied scale.
c. Myanmar: when the platform does more than provide space
Myanmar demonstrates the point at which a platform’s reparative responsibility becomes much harder to deny.
Facebook became a central channel through which military-linked Actors and extremist nationalist groups circulated anti-Rohingya propaganda, dehumanizing claims, and false warnings of an impending Muslim takeover. Facebook’s own commissioned human-rights assessment concluded that the company had not done enough to prevent its platform from being used to foment division and incite violence offline.
The platform did not merely contain the content.
Its ranking and recommendation systems determined what users saw, how widely material traveled, and which messages received additional visibility. Amnesty International later argued that Facebook’s engagement-driven systems amplified anti-Rohingya hatred before the 2017 atrocities and that Meta owed reparations to affected Rohingya communities.
This changes the role of the enabler.
A wall upon which someone writes a hateful message is a location.
A system that identifies which people are likely to respond, recommends the message to them, measures their reaction, and circulates the most inflammatory material more widely is participating in the distribution of the Act.
The platform still may not share the primary Actor’s Intent.
But its machinery is no longer passive.
It makes choices:
- what becomes visible;
- what remains hidden;
- what receives priority;
- what is recommended;
- what is removed;
- and what is allowed to reproduce.
Those choices are usually made in pursuit of engagement rather than violence.
But lack of violent Intent does not erase causal involvement.
The Actor wanted to spread hatred.
Those different Intentions could still converge upon the same realized Outcome.
d. Algorithms: When the Enabler Begins Choosing
The Arab Spring and ISIS examples still began with identifiable human Actors.
One Actor wanted to organize resistance.
Another wanted to recruit people into violence.
The platform expanded their Power.
Algorithms introduce something different.
The platform no longer merely provides the space through which an Act travels. It begins deciding:
- which Act will be seen;
- which emotion will be repeated;
- which conflict will receive an audience;
- and which version of reality will become most visible.
No social-media feed can show everything posted by everyone a user follows. Something must select and order the material.
An engagement-based algorithm usually does this by predicting which posts will make the user:
- click;
- comment;
- share;
- argue;
- or remain on the platform.
The algorithm does not ask:
Is this True?
Will it improve understanding?
Will it restore homeostasis?
It asks:
Will this produce engagement?
That objective has no moral direction.
But human attention does.
Anger, fear, humiliation, moral outrage, and hostility toward an out-group often create powerful reasons to respond. The algorithm observes that response and learns to provide more of what produced it.
The Actor may begin with a political disagreement.
The algorithm discovers that the most hostile expression of that disagreement generates more engagement.
It then gives the hostile expression greater visibility.
Other Actors observe what receives attention and adapt their own behavior.
The mechanism becomes:
Actor creates divisive content → users react → algorithm interprets reaction as value
→ divisive content receives greater distribution → other Actors imitate what succeeds
→ the visible social environment becomes more divisive
The algorithm does not need to hate anyone.
It only needs to learn that hatred performs well.
A preregistered audit of Twitter’s engagement-ranked feed compared the posts selected by the algorithm with the same users’ reverse-chronological feeds. The engagement algorithm selected more partisan content, more hostility toward political out-groups, and substantially more anger. After viewing those algorithmically selected political posts, users reported feeling worse about the opposing political group and somewhat better about their own. The political content selected by the algorithm was also less preferred by users when they were directly asked whether they wanted to see posts like it.
This is revealing.
The algorithm was not simply giving people what they consciously said they wanted.
It was giving them what caused them to react.
Those are not the same thing.
A person may stop to stare at an accident without wanting more accidents.
They may respond angrily to an insult without wanting a society built from insults.
They may repeatedly argue with political opponents while sincerely wishing that political life were less hostile.
The platform records the behavior.
It cannot see the later reflection.
So the reaction becomes an instruction:
Show me more.
The study did not establish that a short period of algorithmic exposure permanently transformed each user’s political beliefs. It demonstrated something sufficient for the question of enabling: the algorithm selected a more partisan and hostile version of the users’ existing social world and immediately altered how they felt about the people on the other side.
Polarization does not require everyone to move toward a more extreme policy position.
It may also mean that people increasingly experience the other group as:
- stupid;
- dishonest;
- dangerous;
- immoral;
- or undeserving of ordinary Trust.
The disagreement changes from:
They are wrong.
to:
They are bad.
And eventually:
They are not legitimate members of the same moral or political community.
At that point, the algorithm has not merely enabled a message.
It has helped shape the Reference Set through which millions of people Judge one another.
Each side receives a distorted picture of the other.
- The most offensive statement is shown as representative.
- The most enraged response becomes Evidence of what the opposing group is supposedly like.
- Moderate people become less visible because moderation produces less reaction.
- Compromise begins to resemble betrayal because users repeatedly see the other side through its worst-performing Actors.
The platform then points to user behavior and says:
- This is what people chose.
- But the choice was circular.
- The algorithm selected what people would react to.
- The reaction taught the algorithm what to select.
- Political Actors then learned what the algorithm rewarded and produced more of it.
What begins as selection becomes a feedback loop:
- human bias trains the algorithm;
- the algorithm magnifies the bias;
- the magnified bias changes human behavior;
- the changed behavior trains the algorithm again.
This makes reparative responsibility difficult to locate.
Who caused the polarization?
- The politician who produced the inflammatory message?
- The user who shared it?
- The observer who responded with anger?
- The company that chose engagement as the governing objective?
- The engineers who designed the ranking system?
- The advertisers whose payments made attention valuable?
- There may be no single Actor carrying the entire debt.
But there is still an enabler.
The platform constructed the machinery that converted reaction into distribution. It chose the objective being optimized. It possessed the data showing what the machinery produced. And it retained the Power to alter the system after its effects became visible.
Its responsibility does not arise because the algorithm independently intended polarization.
It arises because:
the platform built a system that learned to profit from disturbance, then allowed that disturbance to become part of the social environment.
Repair cannot therefore consist only of removing individual posts or banning individual Actors.
The enabling mechanism may itself require Repair:
- changing what the algorithm rewards;
- reducing the amplification of out-group hostility;
- giving users meaningful control over ranking;
- making recommendation systems visible enough to be Judged;
- measuring effects beyond clicks and time spent;
- and treating social damage as a realized Outcome rather than an external cost.
The Actor creates one divisive message.
The algorithm decides whether it disappears among millions of others—or becomes the reality everyone else is forced to see.
The Actor supplies the disturbance.
The algorithm teaches the disturbance how to reproduce.
(4) The Guarantor
The guarantor occupies a strange position within Repair.
- They may not have caused the disturbance.
- They may not have participated in it.
- They may not have enabled it.
- They may not even have existed when the original relationship between Actor and Recipient was formed.
Their responsibility arises because they agreed in advance to absorb some portion of the loss if the Actor failed.
This may include:
- insurance companies;
- employers;
- banks;
- professional compensation funds;
- victim compensation programs;
- pension guarantors;
- or the State itself.
The guarantor says:
If the Actor cannot complete the Repair, I will stand behind them.
This solves one of the central weaknesses of Actor-based Repair.
The Recipient should not remain injured simply because the Actor is:
- insolvent;
- dead;
- unknown;
- incapable;
- or unable to reverse the damage.
The guarantor separates the Recipient’s Repair from the Actor’s capacity.
- A driver may be unable to pay for catastrophic injuries.
- A doctor may be unable to compensate a patient for permanent harm.
- A company may collapse after damaging thousands of people.
The guarantor creates a second source from which Repair may be drawn.
The mechanism becomes:
Actor → Act → Recipient
- Actor cannot complete Repair
- Guarantor absorbs the debt
- Recipient receives at least partial restoration
This is socially useful.
Without guarantors, every major Act would carry the possibility of total personal ruin for the Actor and permanent uncompensated loss for the Recipient.
Insurance and public guarantees allow people to:
- drive;
- practice medicine;
- operate businesses;
- lend money;
- employ others;
- build infrastructure;
- and undertake activities whose risks no individual could safely absorb alone.
The guarantor therefore makes socially useful risk possible.
But the guarantee changes the Actor before the Act occurs.
The Guarantee Changes the Reference Set
An uninsured Actor asks:
If this goes wrong, what will happen to me?
An insured or protected Actor may instead ask:
If this goes wrong, what will the insurer, employer, Institution, or State pay?
The material risk of the Act has been transferred.
The Actor may still bear:
- shame;
- legal punishment;
- higher premiums;
- loss of employment;
- professional discipline;
- or reputational damage.
But some portion of the consequence no longer lands upon the person making the decision.
This reduces the perceived cost of failure.
That reduction is not necessarily a defect. It is the reason the guarantee exists.
- A surgeon who personally faced unlimited financial ruin from every possible complication might refuse necessary but difficult operations.
- A driver who could never compensate a catastrophic accident might be unable to participate in ordinary economic life.
- A business might avoid every uncertain project, even where the social benefit greatly outweighed the danger.
The guarantor allows the Actor to move despite uncertainty.
But the same mechanism can weaken restraint.
- The Actor gains the benefit of Acting.
- Another party absorbs much of the loss when the Act fails.
The guarantee therefore creates a separation:
The guarantor absorbs the consequence.
The Recipient experiences the injury.
This is the classic problem of moral hazard, but the term alone does not fully capture what happens.
The guarantee can weaponize inertia.
The Weaponization of Inertia
Once failure has been priced, insured, and routinized, the Actor may have less reason to change the conditions producing it.
A company discovers that a product sometimes injures people.
It could:
- redesign the product;
- slow production;
- change its procedures;
- train workers;
- warn customers;
- or stop selling it.
But each change requires immediate effort and expense.
The injuries, by contrast, may be:
- rare;
- delayed;
- disputed;
- distributed among many Recipients;
- or paid through insurance.
The guarantee makes continuation easier than correction.
The system can remain as it is because the cost of each failure has already been assigned somewhere else.
The Actor does not necessarily decide:
I want people to be harmed.
The decision may be much less visible:
We will continue operating as before.
That is inertia.
The guarantor turns that inertia into a sustainable strategy by absorbing enough of its consequences to prevent the disturbance from forcing structural change.
The same mechanism can appear in public Institutions.
- A police department may repeatedly pay settlements for misconduct, while individual officers bear little of the financial cost.
- A corporation may treat injuries as insured losses.
- A bank may assume that the State cannot allow it to collapse.
- An employer may absorb the legal cost of an employee’s conduct while preserving the incentives that produced it.
The guarantor Repairs each Event separately while allowing the condition generating the Events to continue.
The case is closed.
The Institution remains unchanged.
Another Recipient appears later.
The guarantor may therefore produce local Repair without systemic Repair.
The immediate Recipient receives compensation.
But the repeated source of disturbance remains intact.
When the Guarantor Becomes an Enabler
- The guarantor and enabler are different roles.
- The enabler provides the Actor with the means, access, Authority, protection, or opportunity to perform the Act.
- The guarantor promises to absorb the cost afterward.
But the roles can converge.
When the promise of protection materially changes the Actor’s willingness to take the risk, the guarantee becomes part of the enabling condition.
The sequence changes from:
The Act occurred, and the guarantor responded.
to:
The Actor was willing to Act partly because the guarantor would respond.
The guarantor may then contribute to the disturbance without ever participating in the immediate Act.
This does not mean every insurer is responsible for every injury caused by an insured person.
The guarantee itself is socially necessary.
The relevant questions are:
- Did the guarantor price the risk accurately?
- Did repeated claims reveal a continuing danger?
- Did the guarantor require the Actor to reduce that danger?
- Did it continue protecting the Actor after the pattern became clear?
- Did the guarantee make harmful conduct economically easier to continue than to correct?
- Did the guarantor pay individual claims while ignoring the system producing them?
Once the guarantor knows that its protection is preserving a repeated disturbance, its reparative responsibility may expand.
It may no longer be enough to write another check.
Repair may require:
- changing coverage conditions;
- requiring preventive safeguards;
- raising the cost of repeated failure;
- withdrawing protection from reckless conduct;
- sharing information about recurring danger;
- or forcing structural correction before another claim is accepted.
The Guarantor’s Contradiction
The guarantor exists to prevent one failure from destroying both Actor and Recipient.
It pools risk, stabilizes relationships, and ensures that Repair remains possible even where the immediate Actor lacks the means.
But that same protection can reduce the pressure that forces Actors and Institutions to change.
The guarantor therefore stands between two forms of Justice:
With too complete a guarantee, the Actor may never need to Repair the conditions that produced the injury.
The guarantor makes risk survivable.
But risk that becomes too survivable may stop feeling like risk at all.
The deeper question is therefore not merely:
Who pays when the Actor fails?
It is:
Does the guarantee Repair the disturbance,
We explore this in 7.3/9.
- ה -
end of post 7.2/9
- ה -
Comments
Post a Comment