Preface: This article was co-created with ChatGPT.

To refuse “reform that bypasses power,” the target has to move away from process tweaks and moral training and back to three unavoidable questions: who holds discretionary authority over lawful violence, who can meaningfully impose consequences after the fact, and who controls the allocation of resources and tasks. If these three dimensions remain untouched, everything else is just a more presentable repackaging of the same power relations.

The first step is to transform the authorization to use force from something that is primarily justified after the fact into a legal threshold that meaningfully constrains behavior in advance.

At the constitutional level in the United States, the governing standard for police use of force is not simply “necessary and reasonable,” but the “objective reasonableness” test articulated by the Supreme Court in Graham v. Connor. This framework requires evaluating force from the perspective of a “reasonable officer on the scene,” rather than through hindsight or subjective intent. The problem is not the absence of a standard, but that this standard is structurally designed to be highly situational and deeply dependent on narrative reconstruction, which makes it far more effective as a tool for post hoc justification than for ex ante restraint.

The “reasonable officer” is not an empirically verifiable figure. It is an abstract reference point shaped by police culture, prior case law, peer norms, and institutional tolerance. When this reference point is repeatedly affirmed within systems that impose weak discipline and limited accountability, the boundaries of “reasonableness” predictably drift toward greater tolerance of violence. Formally, the standard is described as objective; functionally, it is anchored in a profession-specific viewpoint that is already systemically biased.

More critically, the framework centers post-incident storytelling. Once an outcome has occurred, the system permits investigators and courts to assemble an explanatory chain from fragmented contextual elements—lighting conditions, distance, perceived movements, prior records, hypothetical weapons, emotional states. No single factor determines the outcome, but they can be endlessly layered until nearly any use of force appears “understandable under the circumstances.” Situations that were low-risk, nonviolent, and amenable to delay can thus be retroactively legitimated as high-intensity control measures.

At the constitutional level, “objective reasonableness” does not require officers to demonstrate that they used the lowest feasible level of force, nor does it impose a strict proportionality requirement. It asks only whether the conduct was not plainly unreasonable. The result is that while use-of-force policies often describe escalation as linear, real-world practice frequently involves leapfrogging to higher levels of force. As long as the post hoc narrative coheres, the escalation itself is rarely interrogated.

Structurally, the standard favors the officer over the person subjected to force. The reference frame is the officer’s perception of potential risk, not the bodily vulnerability or life risk of the person being restrained. Consequently, individuals who are already handcuffed, immobilized, and repeatedly expressing distress are often legally discounted, while officers’ speculative assessments of “possible threat” are amplified. Safety, in this framework, is defined as the officer’s sense of security rather than the minimization of total harm.

If reform is serious about not bypassing power, this authorization logic must be narrowed at the level of law and binding policy. At minimum, the use of high-level or potentially lethal force must be restricted to situations involving necessity and an imminent risk of serious bodily harm, with an affirmative requirement to show that no lower-intensity alternatives were viable. Continued application of force against individuals who have been effectively restrained and deprived of mobility should be treated not as discretionary judgment calls, but as grave violations, potentially rising to criminal conduct.

The second core element is converting “de-escalation and retreat” from aspirational rhetoric into professionally protected, institutionally rewarded practices.

De-escalation is not a set of soft skills; it is a deliberate choice of trajectory. It involves pulling an interaction away from a confrontation-control pathway and back toward communication and management. In encounters defined by extreme power asymmetry, continued forward pressure—commands, proximity, physical dominance—predictably intensifies defensive reactions. De-escalation does the opposite: it slows tempo, reduces stimuli, avoids stacked commands, moderates tone and posture, allows the other party to speak, and accepts that comprehension and compliance may require time. Its core premise is not politeness, but the rejection of immediate obedience as the sole legitimate outcome.

Tactical retreat or disengagement is similarly mischaracterized. It is neither abandonment nor failure, but a recalibration of risk, time, and resources. In the absence of imminent danger, creating distance, altering modes of contact, or waiting for more appropriate support is itself a professional judgment. Many confrontations escalate not because the underlying issue is unmanageable, but because the system refuses to allow time and space, compressing all encounters into moments that must be resolved immediately.

Both approaches treat time as a tool rather than a threat. Escalation-oriented policing treats delay as loss of control; de-escalation and retreat recognize that time can reduce emotional intensity, clarify misunderstandings, and enable safer intervention pathways. But within police cultures that equate professionalism with dominance, retreat with weakness, and delayed compliance with danger, these practices are nearly impossible to sustain unless they are explicitly protected by policy and reinforced through evaluation systems.

The third requirement is making the consequences of violations predictable, swift, and external. As long as investigation and accountability remain embedded in “we investigated ourselves” structures, power will not be meaningfully constrained. Internal affairs units can remain responsible for minor disciplinary matters, but cases involving death, serious injury, suspected excessive force, falsified reports, or obstruction of medical aid must automatically trigger transfer to independent external bodies. These entities must possess subpoena power, full access to records, and the authority to publish findings. Without this, reform collapses into black-box assurances that something was “looked into,” while power remains intact and obscured.

The fourth unavoidable issue is police unions and collective bargaining agreements. Whether reforms succeed is often determined less by statutory language than by what labor contracts quietly permit through delay, buffering, and reversal mechanisms.

In many jurisdictions, contracts allow disciplinary records to be destroyed after a set period or sealed from public and external review. This actively erases behavioral history, forcing institutions to treat repeated misconduct as isolated incidents. When repeat behavior cannot be documented, provisions aimed at punishing repeat offenders become unenforceable by design.

Equally common are special interrogation protections that allow officers suspected of misconduct to delay giving statements and to review video, reports, and colleague testimony beforehand. This is not procedural neutrality; it shifts investigations from fact-finding to narrative alignment. When subjects of investigation can fully absorb the evidentiary record before speaking, even the strictest ex ante standards are neutralized by post hoc coherence.

Highly proceduralized termination processes and mandatory arbitration further weaken accountability. When arbitrators routinely reduce or overturn disciplinary decisions, including terminations, the message to the field is unambiguous: dismissal is not an endpoint, but a negotiable phase. Under these conditions, “zero tolerance” policies function as symbolic declarations rather than behavioral deterrents.

Finally, reform requires redefining the scope of what police are tasked to do. A substantial share of fatal encounters originate not in violent crime, but in mental health crises, addiction, homelessness, school discipline, and low-risk disputes. As long as armed police remain the default point of entry, situations will predictably follow a bodily control logic.

A reform agenda that does not bypass power must remove part of that gatekeeping authority. It must mandate diversion: non-armed professional teams as primary responders, with police relegated to perimeter support only when specific safety thresholds are met. Without this structural reallocation, reform simply asks the same armed system to perform the same functions, using softer language to reach identical outcomes.