THE DILEMMA OF SERIOUSNESS Copyright Copyright (c) 2026 Momen Ghazouani. All rights reserved. This excerpt is shared for free listening and discussion and may not be republished as a substitute for the book. Author: Momen Ghazouani Publisher: Setaleur Book Publication year: 2026 ISBN: 978-9909-03-030-1 ISBN 978-9909-03-030-1 A Note to the Reader FREE EXCERPT (approximately 10,000 words from a full-length book of roughly 52,000 words) This is a free excerpt from The Dilemma of Seriousness by Momen Ghazouani. These pages offer a glimpse into the ideas, reflections, and ques- tions explored throughout the book. If this excerpt resonates with you and leaves you curious to discover more, I invite you to explore the complete edition. Continue reading. Discover the full book. Get your copy of The Dilemma of Seriousness and experience the complete work. Thank you for your interest, and for taking the time to explore these ideas. Momen Ghazouani Author of The Dilemma of Seriousness i About the Book Generative AI has changed the cost of appearing capable. Founders can now produce polished strategies, persuasive presentations, and professional-looking plans with unprecedented speed. Yet the ap- pearance of competence is not the same as the capacity to exercise judgment, navigate uncertainty, or accept responsibility. The Seriousness Dilemma examines the widening gap between visible output and underlying ability. Drawing on Michael Spence’s signal- ing theory, it explores how professional artifacts once served as costly signals of competence, and how generative AI challenges their relia- bility. At the center of the argument is the Wrapper Illusion: the risk that an impressive layer of generated work can conceal undeveloped judgment and an untested capacity to lead. Across twelve parts, the book examines this problem in education, consulting, law, scientific publishing, medicine, and digital labor mar- kets. It considers historical technological parallels, research on self- assessment and automation, competing arguments, and the effects of age, bias, and unequal access on evaluation. Moving beyond diagno- sis, it proposes practical methods for assessing judgment, including structured interview questions, a scorecard without a total, and a seven-line decision record. Written for investors, employers, educa- tors, founders, and those whose abilities are being evaluated, this book asks what seriousness means when execution becomes increasingly accessible, but discernment and accountability remain difficult to au- tomate. Its central question is not simply who can produce impressive work, but who can answer for the decisions behind it. ii PART ONE: DIAGNOSIS 1 The Dilemma of Seriousness Free Excerpt Two partners left the conference room at different speeds. It was a Tuesday, the kind of Tuesday that repeats itself in every city with money and ambition in it, and the pitch had gone the way pitches are supposed to go. The founder was twenty-four. Her product used machine learning to underwrite small-business loans in markets that conventional lenders had written off as un- profitable to evaluate, and she had walked the partners through an interface so clean it seemed to apologize for the industry it belonged to. The risk model was documented in a memo that anticipated the questions a skeptical credit officer would ask before he had time to ask them. The competitive slide did not flatter her; it named three well-funded rivals and explained, con- vincingly, why none of them had solved the specific problem she had solved. Nothing about the hour had felt like a performance. It had felt like evidence. The senior partner, walking to his car, was already drafting the internal memo in his head, the one that would recommend a term sheet by Friday. He had been doing this for twenty-two years, and the hour he had just spent matched almost exactly the hours that had preceded his best investments: the same fluency, the same command of detail, the same sense that the person in the room was several steps ahead of him. The junior partner, walking beside him, said nothing for a while, and then said that something about the meeting had bothered her, though she could not yet say what. She believed the product worked. She was less sure that the founder understood, in her bones, what would happen the first time the model was confidently wrong about a real business and a real family’s income. She could not point to a sentence in the pitch that had raised this doubt. She could only say that everything she had watched proved the product, and nothing had yet proved the person. 2 The Dilemma of Seriousness Free Excerpt That difference of opinion, replayed with small variations, is becoming one of the defining scenes of this decade in venture capital, in hiring, in universities, and inside companies evalu- ating their own people. It is not a story about a fraud being uncovered later, though such stories will keep happening for reasons that have nothing to do with this one. It is a story about two experienced observers, both acting in good faith, watching the identical hour of evidence and drawing different conclu- sions about what it licensed them to believe. The senior partner was reading an old signal in the old way. The junior partner, without quite having the vocabulary for it yet, had noticed that the signal no longer meant what it used to mean. It is worth being precise about why the senior partner’s instinct was, for most of the history of business, a reasonable one to trust. Economists have a name for the mechanism he was rely- ing on without knowing he was relying on it. In 1973, a young economist named Michael Spence, who would later share a Nobel Memorial Prize for the work, published a paper asking a deceptively simple question: if an employer cannot directly observe how capable a job applicant really is, what stops every applicant from simply claiming to be capable? Spence’s answer was that a costly signal can do the work that direct observation cannot. A university degree, in his model, does not need to make anyone smarter or more disciplined in order to be useful information; it only needs to be cheaper for the genuinely able to acquire than for the less able to fake. If a weak student finds four years of rigorous coursework nearly unbearable while a strong student finds it merely demanding, then the degree sepa- rates the two groups even though it was never designed as a test of anything in particular. The signal works because the cost of faking it is prohibitive. 3 The Dilemma of Seriousness Free Excerpt A signal is evidence that someone paid. When no one has to pay, the evidence keeps its appearance and loses its meaning. Nobody in that conference room had read Spence’s paper be- fore the pitch, and none of them needed to, because they had absorbed its logic through experience rather than theory. A flawless product demo, a rigorous memo, a competitive analysis that pre-empted its own objections: these used to be expensive to produce. Producing them required a founder who could either master a difficult technical domain herself or persuade someone who had already mastered it to work for equity and belief rather than a market salary. It required surviving the re- cruiting conversations in which capable people test a founder’s judgment before agreeing to follow her. It required enough dis- cipline to prioritize one feature over five plausible alternatives, because there was money enough for one and not five. But because the cost was real, the signal was informative, in exactly the way Spence’s model predicted a generation before anyone had used it to think about start-ups. What has happened since is not that founders learned to fake this signal. It is that the cost of producing it has collapsed for ev- eryone, honest and inexperienced alike. A generative system can now draft the technical documentation, anticipate the objec- tions, structure the market memo, and even simulate a working prototype, in a fraction of the time and without any of the old recruiting friction. The founder in that conference room was not lying about anything. She built the product, or directed enough of its building that the distinction barely matters at this stage of a company’s life. But the specific kind of evidence her presentation used to represent, the kind Spence’s economics explained so cleanly, has stopped separating the capable from 4 The Dilemma of Seriousness Free Excerpt the not-yet-capable, because it has become cheap enough that both groups can afford it. Spence’s separating equilibrium, in which only the genuinely able find the signal worth acquiring, has begun to collapse into what economists call a pooling equi- librium, in which everyone acquires it and the signal itself goes quiet. This book calls the resulting situation the Seriousness Dilemma: a growing gap between how impressive a person’s work appears and how much that impressiveness reveals about the person who produced it. The dilemma is not a claim that founders, job applicants, or students are pretending to be something they are not. Most of them are not pretending at all. It is a claim that the evidence evaluators have relied on for a very long time, evidence that once did real epistemic work almost by accident, has become unreliable precisely at the moment when it looks most persuasive. The most dangerous pitch is not the false one. It is the true one that has stopped proving what it used to prove. [The book then separates three different things that people mean when they say someone ”seems serious.”] It helps, before going further, to take apart a word that is doing too much work when people say a founder, a product, or a pitch “seems serious.” Three distinct things travel under that single word, and this book will keep them separate through- out, because conflating them is the exact mistake the dilemma exploits. The first is artifact seriousness: how functional, techni- cally sound, and commercially relevant the product, document, or deliverable actually is, judged on its own terms. This can be tested, at least in principle, by people qualified to test it. The second is personal maturity, sometimes called founder maturity 5 The Dilemma of Seriousness Free Excerpt when the person under evaluation runs a company: judgment, intellectual honesty, discipline under pressure, the willingness to make trade-offs when no option is comfortable, and the readi- ness to answer for outcomes rather than for intentions. This can only be observed over time, under conditions that actually test it, which is precisely what makes it expensive to evaluate and easy to assume. The third is perceived seriousness: how serious the person, the product, and the surrounding operation appear to an outside observer within the first hour of contact, built from polish, vocabulary, pacing, and the thousand small cues that read as competence. Perceived seriousness is what any evaluator actually has access to at the moment of first judgment. The other two must be inferred from it. For most of business history these three moved together closely enough that treating them as one thing was a defensible short- cut. Artifacts that were genuinely sophisticated tended to re- quire people with genuine judgment behind them, and people with genuine judgment tended to produce work that read, to an outsider, as serious. The correlation was never perfect. It did not need to be perfect to be useful, in the same way a smoke detector does not need to distinguish every kind of combus- tion to be worth installing. What generative tools have done is not to break the correlation everywhere at once, but to make it breakable in a specific and consequential direction: artifact seriousness and perceived seriousness can now rise sharply, and often honestly, without personal maturity rising at anything like the same pace. [The book then introduces two ideas that it relies on through- out: governance, and the difference between leverage that is understood and leverage that is not owned.] 6 The Dilemma of Seriousness Free Excerpt That word, govern, deserves to be introduced carefully, because the rest of this book leans on it. Governance, as this book uses the term, has nothing to do with corporate boards or legal struc- ture. It means the ongoing, personal capacity to decide what should exist and what should not, what to trust and what to discard, and who remains answerable when a decision turns out to have been wrong. Building something and governing it are different skills, acquired on different timelines, through differ- ent kinds of experience. A prototype answers the question of whether something can be made to work. Governance answers the harder question of whether it should have been made, in this form, at this time, and whether the person responsible can be trusted to notice when it stops working and to say so before someone else has to. Governance, in turn, requires a clean way of talking about delegation, because almost every founder, including the most disciplined ones in this book’s account, delegates enormous amounts of work to tools, contractors, and collaborators. This book will distinguish between two kinds of leverage through- out. The first is understood leverage: a founder has handed a task to a tool or a person, but she can explain what was asked for, evaluate what came back, recognize where it failed, and take responsibility for the result regardless of who or what produced it. The second is unowned leverage: the founder has received an impressive output and adopted it, but cannot explain its reason- ing, cannot say with confidence where it might be wrong, and when pressed, gestures toward the tool rather than toward her own judgment. Both kinds of leverage can produce an identical demo. Only one of them tells an investor anything reassuring about what happens after the money arrives. 7 The Dilemma of Seriousness Free Excerpt [It then explains why the person using the tool is often the last to notice the gap.] The cost shows up first inside the person, before it shows up in any spreadsheet, and it is worth taking seriously as a psycho- logical matter rather than only an evaluative one. When a tool suddenly expands what someone can produce, that person’s sense of what she understands tends to expand along with it, and the two expansions are not the same size. This is not evi- dence of vanity. It is closer to an ordinary feature of how people calibrate their own competence against the feedback available to them, and generative systems are an unusually generous source of that feedback. A tool that is fluent, endlessly patient, and always ready with a coherent answer will, almost by construc- tion, leave its user more confident than a tool that pushes back, stalls, or admits it does not know. Call this the generous mirror: a reflection that shows a founder a more finished version of her own thinking than she actually produced, without ever quite disclosing how much of the finish belonged to the mirror rather than to her. The psychological research on self-assessment gives this intu- ition a real foundation, though it is worth stating the founda- tion carefully rather than overstating it. In 1999 the psycholo- gists Justin Kruger and David Dunning published a study of how people in several domains, including logic and grammar, rated their own performance against their actual scores, and found that the weakest performers tended to be the most confi- dent that they had done well, largely because the same lack of skill that produced the poor performance also made it hard to recognize the poor performance for what it was. The finding has since become one of the most cited, and one of the most casu- ally misused, results in popular psychology, and methodologists 8 The Dilemma of Seriousness Free Excerpt have pushed back on parts of the original analysis, noting that some of the pattern can be explained by statistical regression toward the mean rather than by a pure failure of self-insight. The caution is worth carrying forward: this is a real and well- documented tendency, not a precise law, and it does not license treating every confident founder as unaware of her own limits. 9 PART TWO: HOW SIGNALS DIE 10 The Dilemma of Seriousness Free Excerpt In the last week of October 2025, the dean of undergraduate education at Harvard, Amanda Claybaugh, published a twenty- five-page report whose central finding could have been guessed by any student on campus. Sixty percent of the grades given to Harvard undergraduates in the 2024-25 academic year were A’s. A decade earlier the figure had been about forty percent, and two decades earlier it had been under a quarter. The median grade point average of the class of 2025 was 3.83, against 3.64 for the class of 2015. The report took care to rule out the flattering explanation. According to the college’s own course surveys, students were spending about as many hours outside class as they had for twenty years, roughly five and a half to six and a half hours a week per course, so grades had risen while measured effort had not. The dean’s conclusion, in language unusual for an administrator, was that grading as practiced had stopped doing the work that grades exist to do. No one at Harvard decided to destroy the signal, and the report is instructive about how it died. It identified a familiar clus- ter of pressures: instructors who feared looking harsh beside colleagues who were not, instructors who worried that strict grading would send students to other courses, administrators who encouraged faculty to weigh the hardships of individual students, and students who pressed for higher marks. Each of these pressures is defensible when taken alone. Together they produced an outcome that almost nobody wanted, in which the top grade no longer meant much. Princeton’s experience shows the same structure from the other side. In 2004 the university adopted a guideline limiting the share of A’s in undergraduate courses to thirty-five percent, and in 2014 it abandoned the guideline after a decade, explaining that students believed the policy put them at a disadvantage against graduates of other 11 The Dilemma of Seriousness Free Excerpt schools. A university that grades honestly while its rivals do not pays a price that only a coordinated move can remove. When Harvard did coordinate, by publishing the numbers and asking its faculty to act, the share of straight A’s fell from 60.2 percent in the 2024-25 year to 53.4 percent in the following fall term, according to figures the dean sent to faculty and the student newspaper reported. That is a genuine recovery and a small one, and it took twenty years of slide and an institutional decision to obtain it. Grade inflation is a slow and familiar case, and this part uses it as a specimen because it shows in miniature the mechanism that matters for the rest of this book. A signal is informative only while it costs the weak more than it costs the strong, as Part One explained, and that condition can fail in more than one way. It is worth separating the ways, because the remedies differ. The first way is the plainest, and the one this book is mainly about: the price of producing the signal falls for everyone. A handwritten letter, a fluent essay, a well-formed spreadsheet each once demanded hours of skilled effort, and a person who produced one had shown that she possessed the skill or the patience. When a technology makes the same output nearly free, the output stops separating anyone. Nothing about the signal has been corrupted. Its cost has simply collapsed, and with it the information it carried. The second way is corruption by pressure. In 1975 the British economist Charles Goodhart, writing about monetary policy, observed that a statistical regularity tends to break down once authorities lean on it for control, and the anthropologist Mari- lyn Strathern later gave the idea the compressed form in which 12 The Dilemma of Seriousness Free Excerpt it is usually remembered, that a measure which becomes a tar- get ceases to be a good measure. The American social scientist Donald Campbell had reached a similar conclusion about so- cial indicators: the more heavily an indicator is used to make decisions, the more pressure there is to corrupt it, and the more likely it is to distort the very processes it was meant to moni- tor. Grades under competitive pressure are a textbook example. Once a grade decides admission to a graduate program or a job, everyone with a stake in the outcome has a reason to push it upward. Publication counts in science, which the next part will visit, are another: a count that decides careers gets manufac- tured. The third way is the collapse of the market that depended on the signal. In 1970 George Akerlof described a used-car market in which buyers cannot tell good cars from bad ones. Buyers will pay only the average price, owners of good cars will not sell at it and leave, the average quality falls and the price falls with it, and in the limit the market may disappear. His paper also observed that societies build institutions to counter the spiral, and named guarantees, brand names, chains, and licensing as examples. Each is a way of restoring a cost that the buyer can see. The analogy to a founder’s pitch or a graduate’s application is direct enough to be uncomfortable. If every application looks equally polished, the evaluator cannot pay more for real capability, and the capable applicant, who could have been paid for it, discovers that no one is able to see what she has. The cleanest measurement of the first kind of death in the age of generative tools comes from a working paper by Anaïs Galdin and Jesse Silbert, posted in November 2025. They studied Free- lancer.com, a marketplace on which workers bid for one-off jobs 13 The Dilemma of Seriousness Free Excerpt and employers, who often receive dozens of applications within minutes, read each proposal for signs of who is worth hiring. Before generative tools, the researchers found, a proposal tai- lored to the job posting predicted whether the worker was hired. Employers were, in effect, paying for the effort that a tailored proposal displayed. The researchers built a measure of tailoring using a language model and followed it through time. After ChatGPT’s release, and especially after the platform introduced its own one-click proposal writer in April 2023, proposals grew longer, with the median rising from 79 words to 104 and the median proposal written with the platform’s tool running to 159, and the link between tailoring and being hired disappeared on the researchers’ measure. Employers had stopped paying for the thing that had stopped being costly. The paper then asks what follows for the market, and its model answers that the market becomes less meritocratic. Compared with the world before the tools, workers in the top fifth of the ability distribution are hired about nineteen percent less often, and workers in the bottom fifth about fourteen percent more often. Three cautions apply. The paper is a working paper and has not been through peer review, the setting is a market for short freelance tasks where the proposal is nearly everything an employer sees, and the hiring effects are outputs of an estimated model and not a direct count. A market in which employers can also observe a portfolio, a referral, or a probationary period will behave differently. The authors themselves point to one adaptation, the short trial contract, in which an employer learns about a worker by watching work instead of reading a proposal. It is a move from a cheap signal to an expensive observation, and it reappears throughout this book. Even with the cautions, the study captures the mechanism precisely. The signal did 14 The Dilemma of Seriousness Free Excerpt not become false. It became uninformative, and the capable applicants were the ones who lost by it. A signal does not have to be forged to die. It only has to become cheap. 15