QUIT: Done With the Role, Not With the Work

A Reckoning With a System That Demands Your Best and Returns the Dangerous Cases as a Reward for Giving It

There was a moment in a German courtroom that I have returned to more times than I can count, not because it was the worst thing I ever witnessed in that setting, but because it was the one that finally put a number on what the preceding years had cost. A colleague, a professor of human biology, had just testified that a suspect appearing fully masked in video footage could be identified with a probability of 99.72 percent. No visible face. No biometric anchor. Just the information, folded into his reasoning before he had even looked at the recording, that similar items of clothing had been purchased through the same online platform. That information was in his case file. He had read it. It had settled into his thinking the way context always settles into thinking, quietly, without announcement, and by the time he watched the footage, the conclusion was already there waiting for the analysis to confirm it.

I was called to give a methodological assessment of this identification. I did. It was factual, thorough, and cited the published literature on the limits of identifying masked subjects from video. It was also, predictably, unwelcome. When the questioning of my assessment reached a point where it had ceased to be a scientific discussion and become something else entirely, I said, in open court, something that ended my presence in that room. I said: then let’s drop this nonsense. The judge asked me to leave.

I left. And I kept thinking about what I had just watched, not the specific case, but what it represented, which was a system in which a technically indefensible identification stands because the person who made it belongs to the system, and the person who challenged it is asked to leave because challenging it is the disruption, not the identification itself.

That was not the last hearing I attended. But it was the one after which I began to understand, with a clarity that discomfort had previously blurred, that my decision was already made.

What Genuine Neutrality Costs

The ideal of the independent forensic expert is written into German procedural law with a precision that does not survive contact with how courtrooms actually operate. The law says the expert is an assistant to the court, bound by scientific truth, not by the interests of the party that instructed him. The practice is something different, not everywhere, not in every case, but consistently enough, and in a consistent enough direction, that a pattern becomes visible to anyone willing to look at it over years rather than case by case.

The pattern is this: experts who deliver what the prosecution expects continue to receive work. Experts who do not, stop receiving it. The mechanism is never stated explicitly. No presiding judge sends a letter explaining that uncomfortable findings will reduce future referrals. No state attorney’s office publishes a list of preferred experts and what they are preferred for. But the lists exist, and they operate, and everyone who has worked in German forensic expertise long enough knows it.

When you are genuinely neutral, which means you write what the evidence shows regardless of which party benefits and regardless of whether the finding is convenient for the court’s preferred narrative, you do not receive fewer assignments immediately. You receive them differently. The comfortable cases, the ones with clear findings and institutional backing, migrate toward colleagues whose conclusions can be predicted in advance. The difficult ones, the ones that nobody else wants, the ones that carry political sensitivity or reputational risk, come to you. Not because you are respected. Because you are expendable.

In my final years as a court-appointed expert in Germany, the cases that arrived were not random. They reflected a selection that had been operating for years, quietly, through the mechanism of availability: if someone more predictable could take the case, they did. What remained were the cases where the established experts had already passed, the cases where the findings might go anywhere, the cases where someone with a record of honest inconvenient analysis was more useful than someone who might be embarrassed by their own conclusions later.

What this means deserves direct language. A forensic expert whose work contributes to sending people to prison for life does not live without risk, and the risk compounds when the cases involve organized crime, politically motivated violence, or structures with an institutional interest in ensuring that certain findings never surface. I worked those cases because no one else who would handle them honestly would take them. The consequence was that protective address suppression orders were filed for my personal data, and I live two hundred kilometers from Starnberg today, far enough that reaching me requires a deliberate effort. That is not paranoia. It is the straightforward logic of doing consequential and neutral work in areas where consequence and neutrality make uncomfortable enemies.

I am not complaining about the difficulty of the work. I am describing what the distribution of assignments communicates to anyone who follows it attentively over time: the safe cases went to the predictable experts. The dangerous ones came to me.

Every Case. Without Exception.

There is a contradiction in my story that I could not find a logical explanation for over many years. From a professional standpoint I was unmatched, and that is not arrogance, it is a description of the result: I solved every case I was given, without exception, whether as a court-appointed expert, as a specialist in police investigation units, or as an instructor for intelligence services who came to learn what I had developed in the field over decades. Neutrality has been and remains one of the foundations of my life, not as a professional ethos that one puts on at the entrance of a courthouse and takes off again on the way out, but as a conviction I do not relativize for convenience or to meet the expectations of whoever is paying. And yet I did not fit this system. I kept rubbing people the wrong way, again and again, and for years I did not understand why someone who solves every assignment and favors nobody in doing so is perceived by the system as a disruption.

At some point I stopped looking for a logical explanation. My gut gave me the answer that rational analysis could not formulate in words: stop.

Two cases stand in for what I experienced over years and what I could fill books describing.

The first was in Cologne. A shooting in a shisha bar, surveillance camera footage present and decisive. The responsible forensic examiner at the state criminal investigation office had overlooked a single frame, exactly the frame in which the perpetrator holds the firearm in his hand. This can happen, depending on which analysis software is used and how the image sequence is played back and examined. It is not a moral accusation, it is a technical failure with significant consequences. I brought that frame before the court. What followed was four hours of examination that does not deserve the word examination, because what happened in that room was not a scientific exchange but a systematic degradation. Four hours of being treated as an expert whom one might break through volume and repetition. Afterward I was sent home without any escort. Twenty minutes on foot to the train station. One should try to imagine that feeling: a forensic expert who has just presented a decisive finding in a case with everything that implies, alone on a Cologne street at night.

The second case is Thomas Drach, and about that one I will not begin here, because it would fill an entire book and I intend to reserve it for that book. What I will say: it was a trial that came apart entirely. Dutch police officers were insulted in the courtroom and left the room shaking their heads, with a dignity that the proceeding itself lacked at that moment. I was insulted and humiliated for days, I, as the expert who was doing his job to the best of his knowledge and conscience. In the wrong place at the wrong time, a description that in retrospect is more precise than any other. And when the proceeding was finally over, the justice system cheated me, through a formal application, out of the fee I was entitled to claim for work I had done. What a farce at a German court.

These are not exceptions. They are examples from a longer series. And at some point the sum of them means something.

The Interrogation on the Other Side of the Table

One case from my final period as an active court-appointed expert has stayed with me with particular clarity. I had written an expert opinion for defense attorneys in a case involving audio analysis, and was subsequently summoned by the court, initially as a witness rather than as an expert. I insisted on the correct designation as expert, because that is the capacity in which I had worked and the financial difference between the two roles in German proceedings is substantial, which is not an irrelevant consideration but a matter of procedural logic. The judge did not know me. After some back and forth, I was eventually permitted to appear as an expert.

What awaited me in that proceeding was a colleague employed by the state criminal investigation office, who had been working on the same category of audio file and had been unable to produce any usable improvement in quality. The work I had done, using methods I had developed and tested over years, had produced a result she could not replicate.

The questioning lasted several hours. It was conducted in a manner that I found, to use the most precise available word, peculiar. The technical questions were not the kind that a peer asks another peer when she is genuinely trying to understand a methodology. They were the kind that a person asks when she already has a conclusion and is looking for a path to it.

I want to be direct about what this situation illustrates beyond the specific case. A forensic expert employed by a state criminal investigation office is not, in any meaningful sense, independent. She works for the institution. She is embedded in its structure, its priorities, its relationships with the prosecution. This is not a personal failing, it is a structural condition, and the law that defines the forensic expert as a neutral assistant to the court has not resolved it. The independence required by that legal definition is not available to someone who reports to the same institution that investigates and prosecutes the cases she analyzes. The system has known this for as long as the system has existed, and it has not fixed it, because fixing it would require the institutions that benefit from the current arrangement to relinquish something, which institutions are not designed to do voluntarily.

The Body Keeps the Score

There is another reason I am done, and it is one that does not require any analysis of institutional failure to explain. My spine has sustained vertebral compression fractures that do not resolve. Every long drive to a distant courthouse, every hour seated in a witness chair waiting for a proceeding to reach the point where my expertise becomes relevant, every night in a hotel bed before a hearing that begins early, adds to a physical account whose balance has been negative for longer than I should have allowed.

Pain is an honest instrument. It tells you what is costing you more than it is returning, and it does so without regard for the psychological habits, the professional identity, the sense of obligation, that keep you saying yes to the next case even when you have already said yes too many times. I told myself I would stop many times. Another family with an unanswered question. Another case that no one else would take honestly. Another assignment where the finding mattered too much to leave it to someone who would shape it to fit what was expected.

The body was less patient than the reasoning. At some point the pain from a single trip became the argument I had been avoiding, stated with a specificity that professional considerations cannot rebut.

What Each Project Answers

I am not done working. The distinction I am drawing is between work that belongs to a system and work that belongs to a problem that needs solving, and the problems I intend to spend the remaining portion of my professional life on are the ones I kept encountering in courtrooms and case files and expert reports without any adequate tool for addressing.

The two books that have been waiting years for the time to write them properly are each a direct response to a failure I have watched up close. The first will answer a question that a professor once claimed to have answered by saying that the human face contains 246 distinct individual features, without ever publishing the list. I have spent years developing that list. It will be published with full methodological documentation, because an identification method whose evidential foundation cannot be examined is not a method, it is an assertion, and the difference matters in the specific way that it matters when a person’s freedom depends on which one the court is actually being given.

The second book will examine the errors that corrupt video forensic identification, the overconfident conclusions drawn from low-resolution footage, the software applied without understanding of what it does to the underlying information, the experts who deliver probability estimates that bear no relationship to what their methodology can support. I know what these errors look like from having seen them, challenged them, and been asked to leave rooms for challenging them. I know what courts believe about video evidence and I know how far that belief departs from what the science actually supports. That departure is what the book will document.

The VetBot exists because veterinarians, like the human medicine practitioners I have watched in court, are overwhelmed with documentation and administrative demands that leave too little cognitive bandwidth for the clinical work they were trained to do. The CrimeBot exists because I have watched defense attorneys walk into hearings without having read their own files, and I have watched the consequences of that. I have also watched the attorneys who did read their files, who engaged experts early, who built their defense on an understanding of the evidence rather than a hope that the prosecution’s case would contain an obvious flaw. The good ones work harder and their clients fare better. CrimeBot will give the good ones more time to do the work that matters, by handling the parts that do not require their specific judgment.

Tyra.chat exists because the data that flows through legal and forensic work is among the most sensitive that exists, and the tools available for working with it have not been built with the assumption that privacy is a requirement rather than a feature. I built it with zero-access architecture, with encryption that means not even I can read what users store, with automatic masking of personal information before it reaches any external processing, and with full legal compliance with European data protection law, because I have read enough files to know what happens when sensitive information reaches infrastructure that was not designed to protect it.

Encryptor exists because I spent decades reading what happens to people who had no adequate tool for protecting information they needed to keep private. The current version encrypts entirely client-side, in the browser, before any data touches the network, with AES-256 GCM and PBKDF2 key derivation. It exists because privacy is not an abstraction and I am not willing to build something that treats it as one.

The facial reconstruction software exists because there are victims who do not have names and there are families who do not have answers, and the computational tools for building scientifically defensible facial reconstructions from skeletal remains have not been accessible to the investigators who need them most. The voiceprint analysis system exists because I have stood in enough courtrooms where audio evidence was either dismissed for lack of a scientific framework or accepted on the basis of an expert’s confidence rather than a method that could be defended, and I know which of those two failures is easier to fix.

What I Will Not Do Again

I will not write expert reports for the German justice system or its investigative authorities. I will not drive to courthouses where my scientific conclusions are evaluated not for their accuracy but for their alignment with the outcome the proceedings are structured to produce. I will not submit to interrogations by colleagues embedded in institutional frameworks that make their independence structurally impossible, regardless of what they or the institutions they work for claim about their neutrality.

I will continue working with European courts outside Germany, with international law enforcement, and with investigators and attorneys in other jurisdictions. The difference in how that work is received is not subtle. There is a quality of engagement with the actual science, a willingness to follow evidence rather than to manage it, and in some cases a genuine expression of gratitude for findings that are inconvenient, that I have not experienced consistently in the German system. These are not trivial differences. They describe whether the expertise being sought is real expertise or the institutional validation of a conclusion that was reached before the expert arrived.

I will teach. I will publish. I will continue building tools that address problems I have spent decades watching go unaddressed. The work continues. The part of it that feeds machinery that takes without giving back, that protects the experts who deliver predictable results and allocates dangerous cases to the ones who insist on honest ones, that asks genuine neutrality to justify itself to institutional compliance and calls this a fair process: that part is finished.

The Sentence I Keep Coming Back To

There is a sentence from that courtroom that I have thought about more than any other from my years as an expert witness, the one I said in the moment when the proceeding had exhausted my willingness to pretend that what was happening was science. It was not a diplomatic sentence. It was not strategic. It was accurate.

Then let’s drop this nonsense.

The judge asked me to leave. I left. And everything that has followed, every project now in development, every decision about where my remaining professional time will go and where it will not, connects back to that moment and to what it clarified. The system I was being asked to serve was not the system the law describes. The neutrality I was expected to demonstrate was not the neutrality I had spent my career practicing. The choice between continuing to function as a component of that system and doing the work I actually believe in was not as difficult as the years of saying yes had made it feel.

I am done with the role. The work is just beginning.