I ran a surprise experiment to validate the underlying point in my articles about the culture of academia and the ways in which white men need to maintain the status quo, more scarcity mindset than intellectual curiosity. A bro who went by Bentham’s Bulldog wanted to remember that the machine is always there, and that boys will be there to stand up for it. I happened to have written at least 288% more words than him in my long career and taught college writing courses. But he’s a man who loves the hero of surveillance, and he knows an article he won’t engage with intellectually is machine-written. I was in a car, watching my partner, the same person who edits my text, drive across a state, and I had a moment to enjoy the culture of masculinity play out, undeterred by anything, certainly experience. In my interactions, not one noticed how they proved my original paper correct. Not one reflected on anything. For them, words are tools to bludgeon and cajole, but most importantly to assert their narrow frame rather than interrogate it. I filmed myself writing this piece, not as proof, because men like them cannot fathom frames that don’t replicate their knowledge. But for the rest of us, those of us whose lives are intersectional, we know how complications are, in fact, a tool for increasing the richness of our soul.
Let’s be clear about what happened here. Men who espouse clarity in academic systems often exhibit behavioral and attitudinal contradictions.
Someone took 770 words of my essay to check the likelihood of AI. He received 100% and then pompously posted his screenshot as proof of his ability to discredit my work.
Irony abounds, in a sense, as the essay was about how a white philosopher engaged in largely unnecessary auditing of the work of a Black professor. What happened after this man chose to discredit my writing helps illustrate the classic academic (or quasi-academic in the case of Substack) culture of carceral critique as a form of denuding minority voices of validity.
All the supporters of my detractor were white men. The irony is not subtle. The irony is not the point. The point is the mechanism. In their actions, we see in real time a demonstration of the culture of intellectual posturing and the falseness of epistemic injustice.
The three men who considered their own judgment authoritative appeared to feel that the means were easily justified, having effectively appointed themselves as arbiters of authenticity. The response from Bentham’s Bulldog was largely predictable. He stated that it was “just so obviously AI if you read it.” His stated qualification is having written approximately three million words. By comparison, I published extensive public blogs, precisely the category of text most commonly harvested for AI training, for two decades, averaging roughly 10,000 words per week. My total output significantly exceeds his.
A second commenter, James Alder, relied on the fair use doctrine. While this is legally accurate, it is worth noting that using someone’s text in order to discredit them, in the context of an argument about academic honesty, represents a notable contradiction.
Tom Hitchner declined to engage with the question of consent altogether. In his framing, published text is simply available for others to use as they see fit. His argument, however, only reinforced the original point regarding men in academic settings. Within that framework, all information is considered fair game so long as no consequences are attached.
Notably, none of these men left a comment on the original post asking why the AI detection setting had been disabled. I have worked in technology throughout my career. Extended writing sessions frequently generate false positives. Furthermore, AI systems often retain signatures of the text they receive. While the specific words may not become part of the system, structural characteristics can persist. Given what has been revealed about these technologies in retrospect, such acceptance rests on minimal knowledge of a fundamentally opaque system.
This episode offers readers insight into the narrow margin afforded to writers of color in intellectual settings. These men did not engage with the essay’s central argument, which is that academia operates according to a system in which white men frequently follow a different set of rules. They did not raise questions about my writing process. They did not approach me directly. Instead, they opted for public accusation in a forum where they could reinforce one another’s positions. Given that Cofnas now enjoys the active support of the Trump administration, with the US ambassador to Belgium, Bill White, writing to Ghent University to demand his reinstatement and warning that American ties to the institution were under review, their conduct, which mirrors Cofnas’s own approach in numerous respects, helps illustrate how the discrediting of voices functions as a form of anti-intellectualism.
None of this occurred in a vacuum, and none of these responses were rooted in genuine curiosity.
My essay argued that Nathan Cofnas, a self-described race “realist,” built his career within a system that rewarded him not for the quality of his scholarship, but for his facility with academia’s institutional frameworks. I documented miscited sources, a definition rewritten to survive its own counterexamples, and errors of plain fact. Within a matter of days, my own sentences had become the subject of examination.
Hitchner confirmed the target explicitly: “Your piece is based on criticizing Cofnas’s publications, right? What if he had prefaced all of them by saying ‘also, no one should criticize these, analyze their originality, etc.’ — would you then be honor-bound to not do so?”
He believes this constitutes a parallel. It is, in fact, the precise opposite of one.
My criticisms addressed his claims directly. I read them, quoted them, verified his citations against his sources, and documented my findings.
What was done to me was not an examination of my claims or of the central arguments in my text. These commenters did not dispute the substance of anything I wrote about Cofnas. They did not engage with the ideas at all.
They moved to discredit as a form of deflection. They discredited my voice. I discredited Cofnas’s ideas. This substitution is worth naming plainly. Addressing an argument requires thought. Discrediting an author requires nothing.
Ultimately, the system that convenes a symposium and a newspaper editorial on behalf of a race realist is the same system that designates a woman a machine. This behavior does not reflect a coincidence of individual temperament. It reflects a documented pattern. These commenters demonstrated, in real time, the culture that academics of color navigate daily.
I asked about consent. They responded regarding permission. These are not the same question, and the substitution is precisely the issue.
Nobody claimed the check was illegal. I never suggested that it was. Fair use functions as a legal defense against a copyright claim, and no copyright claim was ever made. The conversation was redirected toward more convenient ground, with the apparent goal of overwhelming the discussion through sheer confidence.
The philosopher Nicholas Shackel identified this rhetorical structure. The bailey represents the position one actually wishes to occupy, in this case the assumption that a stranger’s essay may be submitted to a machine in order to invalidate her. The motte represents the defensible fortress one retreats into when challenged, in this case the assertion that publication means the rules are whatever the reader decides. These are presented as a single position. Crucially, they are not the same claim.
Hitchner in particular maintained that text released into the world may be used however he chooses. “That’s the nature of publishing, is that you’re putting the text out in the world.” Quoting. Parodying. Checking. In his own framing, the position is unassailable. The structural resemblance to the logic of coercion is worth noting.
One relevant fact went entirely unaddressed by all of my accusers. My publication is configured to refuse AI training. That setting is not decorative. It represents a stated preference: published, visible, and accessible in a single click. My argument in the original essay, and in my subsequent comments, was never about legality. My point regarding power within academia stands. This was a demonstration of men choosing to assert it.
Not one of them raised the question in my own comments section. They said allowed. They said fair use. They said their actions presented no ethical problem. Every word was neutral. Every word was procedural.
Emilio J. Castilla and Stephen Benard conducted three experiments involving 445 participants with managerial experience. In organizations that explicitly described themselves as meritocratic, managers consistently awarded men larger bonuses than they awarded to female colleagues with identical performance records. Their paper, The Paradox of Meritocracy in Organizations, demonstrates that the language of fairness does not function as a constraint. It functions as a permission structure. Once an individual accepts that a process is neutral, active mitigation of bias tends to cease.
This effect is measurable throughout the academy. Lillian MacNell, Adam Driscoll and Andrea Hunt swapped the perceived genders of two online instructors and found that students rated the male identity more highly regardless of who was actually teaching, including on measures of fairness. Cassandra Guarino and Victoria Borden found that women faculty carry significantly more service responsibilities, controlling for rank, race, and field, the majority of it work that tenure committees do not formally recognize. Allison Morgan and colleagues, publishing in Nature Human Behavior, surveyed 7,204 tenure-track faculty and found that 22 percent have a parent with a PhD, approximately 25 times the rate in the general population, a figure that has remained stable for 50 years.
This is not a question of individual villains. It reflects a system that reliably reproduces the same outcomes year after year, while every individual within it believes the standard has been applied fairly.
The Limits of Detection
Here I will extend more fairness than was extended to me. The detection tool in question does perform better on false positives than the tools that preceded it, and I will not pretend otherwise.
However, it is important to consider the broader context surrounding AI detection. Weixin Liang and colleagues at Stanford, publishing in Patterns, found that detectors flagged more than half of non-native TOEFL essays as machine-written while handling US eighth-grade essays nearly perfectly. Debora Weber-Wulff and colleagues tested 14 tools, including Turnitin, and concluded that none were accurate or reliable. OpenAI withdrew its classifier in July 2023, citing low accuracy, after the tool identified only 26 percent of AI-generated text and falsely flagged human writing 9 percent of the time. Common Sense Media surveyed 1,045 teenagers and found that 20 percent of Black teens had been falsely accused of using AI on schoolwork, compared with 7 percent of white teens.
The instrument may well have improved. The tribunal has not.
That tribunal remains a man who determines, on his own authority, that another person’s work requires proof of humanity, and who then goes looking for it. That determination occurs before any tool is opened. No accuracy figure addresses it. Accuracy answers the question of whether the reading was correct. The question actually at issue is who granted the authority.
A Documented Pattern
Miranda Fricker introduced the concept of testimonial injustice, defined as the harm of receiving less credibility than one’s words merit because of who is speaking. Kristie Dotson extended this with testimonial smothering, in which a speaker truncates her own testimony because she recognizes that her audience is not competent to receive it. Sara Ahmed spent a career documenting what follows. She resigned from Goldsmiths in 2016 over the institution’s handling of sexual harassment complaints, and the resulting book, Complaint!, establishes the pattern precisely. The complaint becomes the problem. Not the conduct complained about. The complaint itself.
This pattern can be observed operating in real time. I wrote about a man auditing a Black scholar’s right to his position. Within a day, men were auditing my right to my own sentences. The same instrument was applied on a smaller stage. He had journals and a lawsuit at his disposal. They had a screenshot.
The accompanying tone is also familiar. Daniel Muñoz, writing two days earlier: “Ugh, getting really sick of Jason Arday discourse. It’s so boring and played out. Can everyone please just stop talking about this?”
The man is dead. Flowers were left at Trafalgar Square. Boredom does not constitute an argument. It functions as a request to close a subject, issued by someone who will not be affected by the outcome, and it remains by far the most common form this response takes.
On Bentham
Bentham wrote The Book of Fallacies, in which he cataloged the rhetorical devices used to shut down an argument without ever answering it. He gave us the terms eulogistic and dyslogistic, along with the question-begging appellative, which describes the practice of embedding a contested value judgment within a word so that the judgment itself never requires defense. Fair use is eulogistic. Interacting is eulogistic. Obviously performs more work in that thread than any evidence presented.
These men determined that they were the appropriate watchers without examining their positionality, motivation, or standing to watch.
Bentham’s most widely cited line is likely this one: “The question is not, Can they reason? nor, Can they talk? but, Can they suffer?”
That represents the founding move of the tradition Bentham established. Legality is not the relevant consideration. His entire project rested on the claim that legality settles nothing, because the only thing that ultimately matters is the suffering produced. To assert that there is no ethical problem is to confirm that the role being occupied is that of the enforcer.
Let me return to the question I originally posed. Nobody answered it, though I do not believe it was because the answer was difficult. They regarded the answer as immaterial. One does not ask when one has already decided. The system is employed to confirm existing bias. The machine does not constitute evidence. It constitutes a signature on a verdict reached before the tab was ever opened. Consent wasn’t the obstacle here. The original commenter proceeded to post his findings, mirroring Cofnas’ own approach. That he didn’t recognize the parallel only underscores how thoroughly prejudice has been naturalized in intellectual systems. I argued Cofnas’ work was poorly executed. Rather than engaging with those ideas, white men moved to invalidate me. The same procedure, at a different scale, serving an identical function. The essay itself remains unrebutted by any of these men. Not one of them disputed a single sentence about Cofnas. Race realism was certainly not their concern. Their concern was maintaining their status quo.
This world is f’d. I mean, truly. The world is burning. And we have kids like this. They’re blinded by their truth. What truth is that? They understand their systems of thought. They need the world to conform to them.
What you might notice is that this section sounds different than the previous ones. Why? After 20-some years with the keyboard, I can modulate my voice. I get text and modulate with abandon. What they were actually reading was a tone that I’d modulated to be purposefully flat, both to be legible broadly and because the material was unfamiliar. Har…
