I have spent years building in financial technology, where every serious decision carries consequences. A founder has to make judgments before every fact is known. You study the information available, choose a direction, and take responsibility for the result. But responsible leadership also requires something less celebrated than decisiveness: the humility to recognize when the original theory was incomplete or wrong.
A company that cannot challenge its own assumptions eventually fails. Leadership begins protecting the plan instead of solving the problem. Employees learn to bring the boss reassuring information rather than accurate information. Warning signs become annoyances. The cost of changing direction feels greater than the cost of continuing. By the time reality can no longer be ignored, the damage is larger than it ever needed to be.
Government faces the same human risks, but with far greater power and far higher stakes. It can compel records, restrict liberty, freeze lives in uncertainty, and place the credibility of the state behind an accusation. That authority is sometimes necessary. It also demands extraordinary discipline because the consequences of tunnel vision are not confined to a failed strategy or a bad quarter. They are carried by people, families, employees, customers, and communities.
Overreach rarely begins with an official announcing an intention to be unfair. More often, it begins when ordinary assumptions stop being questioned. A working theory becomes the organizing story. Momentum begins to look like proof. Facts that fit the theory receive attention, while facts that complicate it are minimized, compartmentalized, or explained away.
Confirmation bias backed by sovereign power is not an abstract psychological problem. It is an institutional risk with legal, financial, reputational, and human consequences. The legitimacy of government therefore depends not only on its authority to act, but on its willingness to test itself and change course when the evidence requires it.
The first safeguard
Power must be willing to test itself.
Investigators must form hypotheses. Serious inquiry is impossible without them. A theory helps organize documents, interviews, timelines, and competing explanations. The danger begins when that working theory stops being a tool and starts becoming a conclusion that the institution feels obligated to defend.
At that point, the central question quietly changes. Instead of asking, “What happened?” the process begins asking, “How do these facts support what we already believe happened?” Those questions may sound similar, but they produce very different searches for truth. One invites correction. The other rewards agreement.
Founders encounter this problem constantly. You can become attached to a product, a market thesis, or a hiring decision because you were the person who championed it. The only protection is to create a culture where intelligent disagreement is valued and bad news can reach the top without being punished. If a junior employee sees a flaw that the founder missed, the right response is gratitude, not defensiveness.
Government should demand at least that level of discipline from itself. A truth-seeking investigation should identify what evidence would weaken its theory, assign people to pursue that evidence, and document how competing explanations were evaluated. Contradictions should be treated as information, not inconvenience. Innocent explanations should be tested with the same seriousness as incriminating ones.
The public cannot see every investigative decision, nor should it. But the people entrusted with that process should be able to answer a basic question: what fact, if established, would cause us to change our minds? If the honest answer is “nothing,” the institution is no longer testing a theory. It is protecting one.
Confirmation bias
When a theory becomes an identity.
Confirmation bias is dangerous because it rarely feels like bias from the inside. It feels like experience. It feels like pattern recognition. It feels like a professional instinct that has been validated by years of work. That is exactly why accomplished people and respected institutions are not immune to it.
An investigation becomes especially vulnerable when its theory becomes intertwined with careers, supervision, agency coordination, or institutional credibility. People have briefed their leaders. Teams have devoted months or years to a direction. Resources have been approved. A narrative may have been repeated in meetings until it begins to feel like a fact simply because it is familiar.
Then reconsideration can feel like disloyalty. Revision can feel like defeat. An official who raises a serious doubt may worry about being viewed as weak, naive, or insufficiently committed. Ambiguous facts begin to look incriminating, while contrary evidence is minimized or forced into the original frame. The theory becomes resilient not because it is accurate, but because every possible fact has been made compatible with it.
The most dangerous stage arrives when contrary evidence is no longer merely possible, but visible, and the institution still cannot bring itself to say, “We may have been wrong.” At that point, the original mistake is no longer the only problem. The refusal to correct it becomes a new institutional choice, driven less by justice than by the fear of embarrassment, loss of status, or accountability.
Good leadership prevents that dynamic from hardening. In business, I want the person who can identify the flaw before a customer, regulator, or competitor identifies it for us. A leader who only wants confirmation eventually becomes the last person in the room to understand what is actually happening.
Public institutions need the same intellectual honesty. Separate inference from proof. Distinguish what is known from what is suspected. Record facts that cut both ways. Invite a reviewer who is not invested in the original theory. The purpose is not to make enforcement timid. It is to make enforcement accurate enough to deserve the authority behind it.
Institutional momentum
Sunk costs are not evidence.
Every entrepreneur eventually learns the lesson of sunk cost. You spend money, time, energy, and reputation on an idea that looked promising. New information arrives. The market does not respond. The economics do not work. The temptation is to keep investing because stopping would force you to admit that the earlier decision was wrong.
That instinct is understandable and destructive. The money already spent cannot make the next dollar a good investment. The effort already made cannot turn a weak product into a strong one. Mature leadership asks what decision makes sense now, based on the evidence available now.
Long investigations create the same psychological pressure. Time has been spent. Agencies have coordinated. Careers and reputations may feel connected to the result. Public resources have been consumed. A narrative may have been repeated so often that questioning it feels institutionally expensive.
None of that makes the narrative true. An investigation does not become more accurate merely because it has lasted longer. A theory does not become evidence because more people have worked on it. Consistency with an earlier position is not the same as consistency with justice.
The deeper the investment, the more valuable independent review becomes. A fresh set of eyes can ask questions that the original team no longer notices. A credible institution can narrow a theory, abandon an allegation, disclose a weakness, or acknowledge that an early inference was mistaken. Those decisions do not erase the work that came before. They show that the work remained accountable to its purpose.
Changing course after new evidence is not institutional failure. Continuing because too much has already been invested can be.
Public responsibility
The government's duty is larger than winning.
In Berger v. United States, the Supreme Court explained that a federal prosecutor represents a sovereign whose interest in a criminal prosecution is not simply to win, but to see that justice is done. That distinction is not rhetorical decoration. It describes a public responsibility fundamentally different from the incentives of an ordinary adversary.
A private advocate may be retained to advance a client's lawful position as strongly as possible. A prosecutor carries the name, resources, and credibility of the United States. The role includes enormous discretion over whether to proceed, which charges to bring, how evidence is evaluated, and how the government's position affects the lives of everyone involved.
The Justice Department's own published principles reflect that weight. They describe prosecution as a judgment with profound consequences for accused people, victims, and families, whether or not a conviction ultimately results. They call for reasoned discretion, individualized assessment, and protection of individual rights. They also recognize that probable cause is a threshold, not an automatic command to prosecute.
That matters because a legal minimum and a just decision are not always the same thing. The power to continue does not answer whether continuing is fair, proportionate, or supported by a candid assessment of the admissible evidence. Discretion is not a loophole in the system. It is where judgment, character, and responsibility become visible.
Winning cannot be the organizing purpose because government defines the field on which the contest occurs. Its higher obligation is accuracy under law. The public is served by the most truthful lawful account of the evidence, including evidence that complicates the government's position. A prosecutor who corrects a mistake serves the institution. A system that makes correction feel like defeat misunderstands its own duty.
Fair process
Due process is more than sequence.
Due process is sometimes described as a sequence of steps: notice, hearing, argument, decision. Those steps matter, but a process can look complete on paper while failing in substance. Meaningful process requires enough information, enough time, and enough institutional openness for a person to challenge the government's claims in a real way.
The Fifth Amendment protects against deprivation of life, liberty, or property without due process of law. That promise has practical meaning. Notice must accurately describe what is alleged. The defense must be able to test evidence and expose weaknesses. Required information must be disclosed in time to be useful. Decision-makers must remain capable of revising their conclusions rather than treating every challenge as an obstacle.
The Justice Department's disclosure policy recognizes the government's obligation to seek exculpatory and impeachment information from the prosecution team and encourages a broad view of materiality. That is important because evidence cannot be evaluated fairly if relevant information remains separated across agencies, teams, or files. A process devoted to justice cannot depend on whether a helpful fact happened to reach the right desk.
Government begins with compulsory authority, investigative tools, institutional resources, and the public credibility carried by an official accusation. The person on the other side often begins by trying to understand what has happened while protecting a family, a livelihood, and the ability to mount a defense.
Fairness therefore requires more than checking procedural boxes. It requires care proportionate to the imbalance of power. Due process asks more of government precisely because government begins with more.
The penalty before judgment
Overreach punishes before judgment.
The presumption of innocence is a foundational legal principle. Outside the courtroom, however, an accusation can begin imposing consequences immediately. Headlines reduce complex allegations to a few words. Search results preserve them. Customers, banks, partners, and even acquaintances may react before the evidence has been tested.
For a founder, reputation is not an abstract asset. It is woven into the confidence of employees, the willingness of partners to take a call, the patience of customers, and the ability of a company to keep moving. Years of disciplined work can be placed under suspicion in a day. Legal defense can consume time and resources that would otherwise support families, jobs, innovation, and service.
The personal cost is just as real. Families live with uncertainty they did not create. Children hear fragments they cannot understand. Health and sleep suffer. Every normal decision becomes shadowed by a process whose timing is controlled by someone else. None of those consequences proves guilt or innocence, but they explain why restraint and accuracy matter before judgment, not only after it.
An indictment is not a verdict. Yet its practical effects can resemble punishment long before a neutral factfinder reaches a conclusion. That reality creates an obligation to revisit assumptions promptly when contrary facts emerge. Delay is not neutral. Each additional month can deepen harms that no later ruling can fully reverse.
A justice system should be measured not only by whether it can impose a lawful sentence after conviction, but by whether it takes reasonable care to avoid inflicting unnecessary punishment through the process itself.
Emerging industries
Unfamiliarity is not criminal intent.
Pioneering a business in an emerging industry means operating while the map is still being drawn. Products evolve. Providers change. Rules develop unevenly. New risks appear, and controls mature as companies gain experience. That does not excuse misconduct, but it does make factual and technical precision essential.
Cryptocurrency magnifies this challenge. A single customer experience can involve software, identity verification, banks, payment providers, blockchain networks, wallets, retailers, and compliance systems. A person viewing one fragment may misunderstand who controlled a decision, when money moved, what a record represented, or how a product actually worked at a particular point in time.
Novelty is not concealment. Complexity is not intent. A startup's imperfect early process does not, without more, establish a criminal purpose. The relevant questions are concrete: What did the people involved actually know? What advice did they seek? Which controls existed at the time? How were concerns reported? What changed after a problem was identified? Which party performed each operational role?
Those questions require patient work. Technical language must be translated without stripping away meaning. Historical practices should be judged using the facts, rules, tools, and industry knowledge that existed then, not a cleaner system built years later. A retrospective narrative can make uncertainty look obvious and evolution look suspicious when neither conclusion is justified.
Entrepreneurs accept scrutiny. Building financial infrastructure requires it. What they should also be able to expect is that scrutiny will distinguish innovation from evasion, operational imperfection from criminal intent, and complexity from concealment. Government must investigate legitimate wrongdoing. It must also bring technical competence, proportionality, and humility to industries it is still learning to understand.
Institutional legitimacy
Public trust depends on visible fairness.
Trust in government does not require a perfect record. No serious person expects perfection from any institution staffed by human beings. Trust comes from something more realistic: the belief that mistakes can be identified, contrary evidence will be heard, and power will correct course before pride makes the error permanent.
I have learned that credibility in business is built most clearly when something goes wrong. Anyone can sound accountable when results are good. The real test arrives when the facts are uncomfortable, the correction is expensive, and admitting error risks embarrassment. Customers and employees watch what leadership does next.
The same is true of public institutions. A government that acknowledges a weakness can strengthen confidence because it shows that the rules are more important than the people administering them. An institution that refuses to reconsider can create the opposite impression. People begin to see self-protection where they expected neutral judgment.
The cost extends beyond one person or one case. Witnesses become less willing to cooperate. Businesses become more defensive with regulators. Communities grow cynical about equal treatment. Citizens may conclude that process is being used to validate decisions already made rather than discover what is true.
The government's strongest asset is not fear. It is voluntary respect for lawful institutions. That respect cannot be compelled by demanding deference. It is earned when government explains its actions honestly, applies standards consistently, confronts contradictory facts, and demonstrates that no official's ego is more important than justice.
The constitutional answer
Accountability is not retaliation.
Criticism of government power is not hostility to law. It is part of the constitutional design. Rights have meaning because citizens are allowed to invoke them, lawyers are allowed to press them, courts are allowed to enforce them, and the public is allowed to discuss how authority is exercised.
Accountability is also different from retaliation. Retaliation seeks to punish a person for an unwanted outcome. Accountability asks whether power was used lawfully, accurately, proportionately, and in good faith. It relies on records, standards, and review rather than threats or personal attacks.
The lawful responses to possible overreach are familiar: motions, hearings, appeals, supervisory review, public records, ethics processes, oversight, journalism, scholarship, and peaceful civic speech. These mechanisms can be slow and imperfect. They still matter because public power must never become immune from serious examination.
A government confident in the integrity of its work should welcome testing. Strong evidence becomes stronger when challenged fairly. Weak reasoning is exposed before it causes greater harm. Independent review gives conscientious officials a way to correct course without turning every disagreement into a contest of pride.
The hardest test of a justice system is not whether it can build a case. It is whether it can stop building when the facts no longer support the structure. A person is not a theory. An accusation is not a verdict. Authority is not diminished when it corrects itself. It is diminished when preserving the appearance of certainty becomes more important than pursuing the truth.
The remedy is neither rage nor revenge. It is disciplined, lawful, persistent scrutiny. It is evidence tested openly, rules applied consistently, errors corrected honestly, and power exercised with restraint. That is not an attack on justice. It is a demand that justice remain worthy of its name.
Author's note: This essay presents general principles concerning government power, institutional decision-making, due process, and public trust. It is not legal advice, does not state findings about any named official or pending matter, and should not be read as an attempt to influence any witness, court, or proceeding. Allegations in any case must be resolved through the judicial process.
Sources for editorial review
The public standards behind the argument.
These sources are included so readers can examine the governing principles directly. They establish the constitutional and institutional standards discussed in this essay. They do not resolve the facts of any individual matter.
U.S. Constitution Annotated: Fifth Amendment
The Fifth Amendment provides the constitutional foundation for grand jury process, protection against self-incrimination, and the guarantee that no person may be deprived of life, liberty, or property without due process of law.
Department of Justice: Principles of Federal Prosecution
The Justice Manual describes the reasoned exercise of prosecutorial discretion, the consequences of prosecution, the role of individualized assessment, and the evidentiary standards relevant to commencing or declining prosecution.
Department of Justice: Disclosure of Exculpatory and Impeachment Information
The Department's disclosure policy addresses prosecutors' responsibility to seek exculpatory and impeachment information from the prosecution team and encourages a broad view of materiality in close questions.
Department of Justice: Grand Jury principles
The Justice Manual's grand jury provisions describe the prosecutor's responsibilities within a process that carries significant investigative authority and operates under established legal constraints.
Berger v. United States, 295 U.S. 78 (1935)
The Supreme Court's opinion in Berger remains an enduring statement of the federal prosecutor's distinctive duty to seek justice rather than treat conviction as the only measure of success.