Reclaiming the Narrative: From HR File to Human Story
The 1991 documentary Hearts of Darkness: A Filmmaker’s Apocalypse, assembled by Eleanor Coppola from footage she had shot during the production of Apocalypse Now, ends with her husband speaking directly to the camera. The film he is speaking about had taken three years to shoot, had cost roughly twice its original budget, had cost him personally enough that he had posted his family’s assets as collateral against the overruns, had nearly killed its lead actor during a heart attack the production concealed from its investors, and had nearly killed its director during a depressive episode he describes on camera in language that makes the documentary one of the more unguarded portraits of a working artist in mid-career to be found in the archive. The closing remark is not about any of that. The closing remark is a prediction.
To me, the great hope is that now these little 8mm video recorders and stuff have come out, and some — just people who normally wouldn’t make movies are going to be making them. And, you know, suddenly, one day some little fat girl in Ohio is going to be the new Mozart, you know, and make a beautiful film with her little father’s camera recorder. And for once, the so-called professionalism about movies will be destroyed, forever. And it will really become an art form. That’s my opinion.
The remark, in its delivery, is loose in the way off-the-cuff speech tends to be. The phrase some little fat girl in Ohio arrives in the middle of a sentence whose grammar has already begun to come apart, and the sentence stays apart. The construction does not so much break as drift. The so-called professionalism is the phrase that anchors it. The fat girl is the figure. The Mozart is the analogy. The camera recorder is the means. The argument is that the gatekeeping economy of his industry — the credentialing systems, the institutional permissions, the equipment costs, the union seats, the family-name advantages, the geographic centralization, the entire apparatus by which a person becomes a person whose work is allowed into the conversation — was about to be undone by a technological development the gatekeepers had not anticipated and could not control. The figure he reached for to carry the argument is the figure the rest of this essay is concerned with.
The Prediction and Its Stumbles
It is worth saying plainly that Coppola’s phrasing in the closing remark is, by contemporary standards, awkward. The figure he names is described in terms — little, fat, girl, in Ohio — that scan, on a flat reading, as condescending. The flat reading is not the only reading. The phrasing comes from a man at the end of a three-year ordeal speaking off the cuff, and it comes from a director whose entire mode of speech, in this and in other recorded interviews, is one of associative sentence-making in which the figures arrive faster than the syntax can place them. The stumbles are part of the delivery. The figure is what survives the stumbles. The figure is the part the argument turns on.
What the figure is doing in Coppola’s argument is naming, in five descriptors, the categories that the credentialed system of his industry excluded by structural default. Little names physical and social inconsequence, the absence of the executive presence the industry rewards. Fat names embodiment of a kind the industry, then and now, is not designed to put on screen or in the chair behind it. Girl names gender and age, the absence of the years-of-experience and seniority gates the industry uses to filter authorship. In Ohio names geography and class — the wrong school, the wrong network, the distance from the coastal capital centers in which the credentialing happens. Her father’s camera-corder names economic position, the consumer equipment with which a person with no access to professional equipment would have to work. Every one of the descriptors is, on this reading, a labor barrier. The descriptors are not insults. They are the schedule of gates the industry maintained, and the figure is the person whose path through the gates the existing system had been built to prevent.
Three published readings of Coppola’s figure preceded the one I am offering, and the lineage is worth noting because the lineage shapes what claiming the figure now is doing. The earliest set of readings, beginning with a Slate piece in 2006 and including subsequent essays by other male writers between then and roughly 2013, treated the prediction as primarily about technological democratization. The figure was, in this reading, a placeholder for whoever the next great filmmaker would turn out to be, regardless of demographic; the substantive prediction was about the tools. A second set of readings, beginning with Melissa Baumgart’s 2018 essay You Have to Fail a Little in Bright Wall/Dark Room — an essay she has discussed under the byline fatgirlinohio — and continued by Kate Hagen’s 2022 piece Fat Girls on Film in the Letterboxd Journal, claimed the figure identitarianly. The argument in those readings was that Coppola had named, however inadvertently, a specific demographic position whose underrepresentation in the cinematic conversation was the substance of his point, and that the right response was to recognize the people in that position and to read the work they were producing through that lens. Both predecessor readings are real interpretive contributions, and any further reading of the figure has to acknowledge them as such.
The reading this essay is offering is a third one. It treats the figure as primarily a labor argument rather than a technological one or an identitarian one. The descriptors Coppola assembles are descriptors of distance from the credentialed center, and the distance is the variable whose elimination he is predicting. The figure is who the credentialed economy refuses to see, and the prediction is that the credentialed economy will lose the ability to refuse her once the tools required to do the work are no longer in its exclusive possession. The labor reading does not displace the identitarian reading. It complements it. The identitarian reading describes who the figure is. The labor reading describes what the system is doing to her. The technological reading describes the lever that changes the dynamic. All three are present in Coppola’s phrasing. The contribution of the labor reading is to make the political economy of the system the figure stands against legible in a way the identitarian and the technological readings, by themselves, do not require.
The Prediction Considered Against Subsequent Evidence
Coppola’s prediction can be tested against what has happened since 1991, and the test is informative. The most direct realization of the prediction is the career of Sean Baker, whose 2015 film Tangerine was shot on iPhone 5S handsets using the FiLMiC Pro application on a budget of approximately one hundred thousand dollars. The film opened at Sundance, received broad critical attention, and established Baker as a filmmaker whose work is unambiguously authored from outside the credentialed system. His 2024 film Anora, made for approximately six million dollars with a forty-person crew and edited from his living room couch, won the Palme d’Or at the Cannes Film Festival in May 2024 and, the following March, the Academy Award for Best Picture along with four additional Oscars including Best Director. In his Cannes acceptance speech Baker described his work as my battle cry and used the platform to argue for the survival of independent cinema as a category. The films he beat for the Best Picture statue had budgets, in some cases, an order of magnitude larger than the one he had been working with. The prediction Coppola made in 1991 has, as a matter of empirical record, been realized by Baker’s career, in a form so close to the literal terms of the prediction that the realization is difficult to read as coincidence.
There is a smaller moment, in the timing of Baker’s Cannes win, that bears on the argument. At the same festival, Coppola was presenting his self-financed final film Megalopolis, which had cost him approximately one hundred twenty million dollars of his own money, accumulated over decades through the sale of his wine business and other personal assets, and which would in the months that followed gross about fourteen million dollars at the box office and require him to post the Sentinel Building — his San Francisco property — as collateral against the loss. Also at the festival, the Honorary Palme d’Or was presented to George Lucas, Coppola’s closest contemporary, by Coppola himself; Lucas then presented the competition Palme d’Or to Baker. The torch passed, in physical sequence, from the man who made the 1991 prediction through his nearest peer to the filmmaker who had realized the prediction. Coppola’s own film, made in the spirit the prediction had argued for, lost. Baker’s film, made in the form the prediction had described, won. The configuration is so close to a parable that one is tempted to treat it as a fabrication of the festival’s public-relations apparatus. The record indicates it actually happened.
The counter-evidence is also worth naming, because the prediction’s realization has not been uniform. The most striking counter-example involves another pair of filmmakers who began the way Coppola’s prediction described — outsiders, working with consumer-grade resources, from Ohio. Anthony and Joseph Russo are from Cleveland, are Italian-American, and made their first feature for approximately thirty thousand dollars on credit-card debt before Steven Soderbergh discovered them at the Slamdance Film Festival in the late 1990s. They are, in other words, the fat girls in Ohio of Coppola’s prediction, with the demographic markers slightly inverted. What happened to them, after Slamdance, is the part of the story Coppola’s prediction did not contain. They were absorbed by the credentialed system. They became, by some measures, the most industrialized directors in cinematic history, with their four Marvel features grossing in aggregate approximately six and seven-tenths billion dollars at the global box office. They reached the pinnacle by becoming, with full consent and apparent satisfaction, the professionalism Coppola had wanted to see destroyed. The outsider track, in their case, did not destroy the credentialing economy. It fed them into it. Baker refused the assimilation; the Russos embraced it. Both began in Ohio. The structural difference between their paths is the question Coppola’s prediction did not address. The prediction was about who could enter. The question of what the entered would do once inside — whether they would carry the values that had been excluded with them, or be absorbed by the values that had previously excluded them — turned out to be the more important variable.
Coppola’s Long Argument
It is worth pausing on Coppola’s own consistency, because the consistency matters for whether the 1991 remark should be read as a singular outburst or as a stage in a forty-five-year argument. The latter reading is, on the available record, the correct one. At Cannes in 1979, immediately after the original Palme d’Or win for Apocalypse Now, Coppola told the press that the major studios had become organizations whose primary function was the production of risk-free films, and that the consequence of risk-aversion at scale would be artistic stagnation. In a 1996 interview he described the rise of Wall Street ownership of the studio system as the Big Mac, the McDonaldization of an art form. In 2022, in the period before Megalopolis entered production, he described the Marvel Cinematic Universe in particular as one prototype made over and over, a description that the Marvel franchise’s subsequent commercial trajectory has done less to refute than the franchise’s defenders would have hoped. In the press surrounding Megalopolis in 2024, he said that the contemporary studios existed primarily to service their debt obligations and that the artistic ambitions of their leadership were, accordingly, downstream of their finance functions. The argument is the same argument across forty-five years. The forms in which the argument is delivered change. The argument does not.
What is striking, considered in this longer arc, is how closely the argument tracks the structural critiques that the management and economic-sociology literatures have been developing over approximately the same period. Randall Collins’s The Credential Society, published in 1979 — the same year as Coppola’s Cannes interview — argued that the proliferation of educational credentials in the postwar American economy was a class-closure mechanism, with the credentials operating primarily to exclude rather than to certify. Bryan Caplan’s The Case Against Education, published in 2018, argued that approximately eighty percent of the value of educational credentials in the labor market is signaling rather than human-capital accumulation. Daniel Markovits’s The Meritocracy Trap, published in 2019, argued that the merit framework had become an ideological cover for hereditary advantage. Michael Sandel’s The Tyranny of Merit, published in 2020, argued that credentialism had become the last acceptable form of prejudice in liberal political culture. Lauren Rivera’s Pedigree, Karen Ho’s Liquidated, Sam Friedman and Daniel Laurison’s The Class Ceiling, Anthony Abraham Jack’s The Privileged Poor, and Joseph Fuller and Manjari Raman’s reports on degree inflation through the Harvard Business School Project on Managing the Future of Work have all, in their respective ways, documented the operational mechanics by which credentialed economies maintain themselves and exclude alternatives. Coppola was making the same argument in 1991 that this body of scholarship was beginning to make and would continue to make for the subsequent thirty-five years. He was making it from inside an industry rather than from a university. The argument was the same.
The Credential Wall, Cracking Rhetorically
The argument’s present moment is a confused one, because the language of the argument has, in part, won, while the structures the argument describes have, in larger part, persisted. Approximately a third of U.S. companies dropped bachelor’s-degree requirements for some roles in the period leading up to 2024. Twenty-two states removed degree requirements for state government jobs in roughly the same period. Opportunity@Work, the workforce-policy organization, estimated that more than four hundred thousand state-government positions had been opened to non-degree holders through these policy changes. A Harvard Business School study published the same year, however, found that actual non-degree hiring had increased only by about eighteen percent against the much larger increase in the formal eligibility pool. The wall, in other words, was being talked about more than it was being torn down. The rhetorical change had been substantial. The structural change had been smaller, by a wide margin. The elite hiring practices Lauren Rivera’s Pedigree documented — the cultural-fit assessments, the family-network advantages, the prestige-signaling rituals — remained largely intact. The pay gap that Sam Friedman and Daniel Laurison documented in The Class Ceiling — the sixteen percent compensation differential between professionals from working-class backgrounds and their middle-class counterparts at the same companies in the same roles — had not noticeably narrowed. The wall continued to do its work. The discourse about the wall had simply learned to perform sympathy with the people on the other side of it.
What has begun to change, more substantively, is the technological ground the credentialing wall is built on. Solo-founder startups, as a share of new U.S. company formations, rose from approximately twenty-four percent in 2019 to thirty-six percent in early 2025, according to data from the equity-management platform Carta. Independent technical operators — the so-called solopreneur class — have begun producing software products at scales that, until a short time ago, would have required venture funding and engineering teams. Pieter Levels, the Dutch independent developer, has reported annual revenues in the range of three million dollars from a portfolio of products he operates without any employees. The work of Erik Brynjolfsson and his collaborators on artificial-intelligence productivity effects, including a study of more than five thousand customer-service agents, has found that large language model assistance raised productivity by an average of fourteen percent, by approximately thirty-four percent for novice workers, and by approximately zero for experienced ones. The productivity gain accrues most strongly to the people who were, in the previous credentialing regime, least able to compete with the credentialed practitioners on technical grounds. The lever the technology represents is, in this respect, the same lever Coppola was describing in 1991, updated for a different equipment set. The camera-corder has become the local model and the code-completion environment. The figure who picks up the tool and produces useful output without institutional permission is structurally the same figure. The economic implications are larger.
The professional economy, in the same period, has been shedding labor at scale. The technology sector alone recorded approximately one hundred fifty-three thousand layoffs across more than five hundred fifty companies in 2024, an additional one hundred twenty thousand in 2025, and ninety-five thousand more in the first quarter of 2026 alone, an average of about eight hundred seventy-four jobs eliminated per day during that quarter. Challenger, Gray and Christmas reported more than one million announced job cuts across all sectors through October 2025, the highest level outside the pandemic period. The robotic process automation category in which I had built my career — the work that, at PPG, I had been hired to do — saw its largest publicly-traded vendor, UiPath, lose approximately thirty-four percent of its market capitalization in a single trading session in May 2024 after the company disclosed deteriorating demand, and the stock continued to trade approximately seventy-five percent below its 2021 peak through the first quarter of 2026. The category I had been operating in was being eaten from two directions simultaneously: by agentic AI systems consuming the workflow-automation tasks RPA had been built to handle, and by offshoring of the analyst function to lower-cost geographies of the kind Essay 21 described. The destruction half of Coppola’s thesis — the half about the existing professional economy — is visible, in real time, in the labor data. The construction half — the half about what replaces it — is visible, more slowly, in the solo-founder and independent-operator data the prior paragraph described. The economy the credentialed institutions have been built on is being unbuilt and rebuilt at the same time, and the question of which direction the unbuilding tilts is the question the next decade will answer.
The Information Network Reading
I want to step out of the cinematic frame here, because the closing argument of this series requires a theoretical ground that the cinematic frame alone does not provide. The ground I want to lean on is the one Yuval Noah Harari developed in Nexus, his 2024 book on the information networks that constitute organizations and societies. Harari’s central distinction, the one the rest of the book turns on, is between centralized and distributed information networks. A dictatorship, he writes, is a centralized information network, lacking strong self-correcting mechanisms. A democracy is a distributed information network, possessing strong self-correcting mechanisms. The distinction is not, in his framing, a political one in the conventional sense. It is a structural one about how information moves through the network and how the network responds to information that contradicts its existing model of itself.
Applied to corporate institutions, Harari’s distinction does the work the metrics critique of Essay 17, the professionalism critique of Essay 18, and the transparency argument of Essay 20 were collectively trying to do. An organization that funnels all decisions to its leadership, that treats local knowledge as noise to be filtered out of the reporting that reaches the top, that suppresses error signals through non-disclosure agreements and non-disparagement clauses, that responds to internal critique by treating the critic as a personnel problem rather than as a source of useful information, is an organization operating as an information dictatorship. The internal information network is centralized. The self-correcting mechanisms are absent or have been disabled. The organization is, in Harari’s sense of the term, structurally unable to learn from its own errors, because the errors are not allowed into the conversation in which the learning would have to take place. The metric drops, when they eventually arrive, are the lagging indicators the silenced internal critics had been describing for years. The institution responds to the metric drops with the same rhetoric it used during the period when the metrics had been at their peaks. The rhetoric works on internal audiences for whom the gap between the rhetoric and the data is not directly observable. The rhetoric does not work on the external audiences who have been observing the gap from the beginning.
There is a related passage in Nexus that bears on the cost of the structural choice. In a completely free information fight, Harari writes, truth tends to lose. To tilt the balance in favour of truth, networks must develop and maintain strong self-correcting mechanisms that reward truth telling. These self-correcting mechanisms are costly, but if you want to get the truth, you must invest in them. The investment Harari is describing is the investment that the institutions this series has been describing have, by structural choice, declined to make. The investment is expensive. The alternative — the unbuilt self-correcting mechanism, the silenced internal critic, the metric system that cannot distinguish a humane workplace from an extractive one — is also expensive. The difference between the two expenses is that the first appears as a line item in a budget and the second appears in the form of catastrophes the budget did not anticipate. The catastrophes show up in the form of Computerworld rankings, in the form of class-action settlements, in the form of regulatory enforcement actions, in the form of the news stories that begin a former employee has filed. The expense the institution declined to incur upstream becomes the expense the institution cannot avoid downstream. The arithmetic only works in the short window in which the downstream expense has not yet arrived.
This is the framework in which the twenty-two-essay series is, I think, best understood. The series is itself an exercise in distributed information flow around an institution that elected centralization. The blog is what a self-correcting mechanism looks like when the institution refused to maintain one internally. The argument is not that the blog will produce institutional self-correction; the blog will not produce that, because the institution is not constructed to receive that kind of input from the kind of source the blog represents. The argument is that the information has to enter the broader network whether or not the institution receives it, because the broader network — readers, peers, future employers, journalists, researchers, others who have lived through similar experiences — is itself a self-correcting mechanism at the scale of the labor market, the industry, and the culture. The institution’s ability to manage its narrative depends on the institution being the only narrator. When the narrative is plural, the management is no longer possible. The institution can continue to deny. The denial can be checked against the available alternative accounts. The check is what self-correction at network scale looks like. The contribution of any one account is small. The contribution of the entire body of such accounts, over time, is what makes certain kinds of institutional behavior harder to sustain.
The Narrative Identity Reading
There is a parallel literature that bears on the personal side of what the series has been doing, because the series is not only an exercise in network information; it is also an exercise in the reconstruction of a self that an institutional event had attempted to define. The literature is the one that runs from James Pennebaker’s expressive-writing research through Dan McAdams’s narrative-identity framework, Jonathan Adler’s work on agency emergence in personal narrative, the post-traumatic growth research of Tedeschi and Calhoun, and the broader narrative-therapy tradition of White and Epston. The findings across this body of work converge on a small set of claims that the series, considered as an intervention rather than as a record, has been quietly relying on.
The first claim, from Pennebaker’s 1994 study of unemployed mid-career professionals, is that those who wrote about their job loss found new employment at approximately four times the rate of those who did not, with the Department of Labor classifying the causal evidence as high-quality and more than a hundred subsequent replication studies confirming the basic effect across populations and conditions. The writing is not a self-indulgence. The writing is the operational technique by which the experience is processed into a structure the writer can act from. The blog, in this respect, has been a labor-market intervention as well as a public-record intervention. Both functions are real. Both have been operating throughout.
The second claim, from McAdams and McLean’s 2013 review of the narrative-identity research, is that narrators who find redemptive meanings in their accounts of suffering report higher well-being scores than narrators who do not, with the effect surviving controls for the severity of the suffering. Adler’s 2012 work, using prospective measurement, demonstrated that increased agency-language in personal narrative preceded measured improvements in mental-health outcomes, not the other way around. The narrative work is not a description of recovery. The narrative work is part of the recovery. People author new identities before they inhabit them. The writing is the authoring. The inhabitation follows.
The third claim, from Tedeschi and Calhoun’s post-traumatic growth research, is that more than half of trauma-exposed individuals report moderate or higher growth across five domains: improved relationships, recognition of new possibilities, increased appreciation of life, greater personal strength, and spiritual development. The difference between people who experienced growth and people who remained traumatized was, in this work, characterized by the distinction between stuck narratives and moving on narratives. The moving on was not the absence of acknowledgment of what had happened. It was the integration of what had happened into a larger account of who the person now was. The integration is the difference. The stuckness is the failure of integration. The narrative-therapy tradition’s contribution, in the work of Michael White and David Epston, is the framing of identity as multi-storied — a useful term because it names the truth that a person is not the worst account of themselves that any single institution has produced, and that the alternative accounts a person can construct are not in competition with the institutional account but are additional to it. The HR file is one account. The blog is another account. The two accounts can coexist. The reader can compare them. The comparison is what the multi-storied life makes possible.
The fourth claim, from the small body of research on shame and disclosure, is the one Brené Brown has summarized in a sentence that has, in the years since she said it, become unavoidable in the relevant literature. Shame derives its power from being unspeakable. The unspeakability is the mechanism by which shame operates. The act of speech, conducted in a register the speaker can stand behind, is what disassembles the unspeakability. The blog has been, among other things, that act of speech. The disassembly is in progress. The mechanism is the writing. The writing is the recovery.
Two Personal Notes Before Closing
I want to include two observations from inside my current employment, because they bear on whether the patterns the series has been describing are particular to a single company or are, as the series has argued, structural across the industry. Both occurred at the company I have been working at since the PPG elimination. Both are small. Both are exactly the kinds of patterns the prior essays have described.
The first concerns a role I identified at the same company that aligned closely with my background and that paid materially more than the role I hold now. I am not going to specify the gap in dollars or percentages, because the specifics are not the point and the institutional discretion around compensation deserves some respect even from a writer who is otherwise on the record. What I will say is that the gap is meaningful and that the role’s posted qualifications — several years of experience, an MBA preferred — matched my profile cleanly. I hold the MBA and nearly a decade of professional experience, in a current role whose own posting asked for two years. I applied. The application was turned aside on two grounds delivered together: pay-band progression — the move would have skipped a band, which the process is not built to support — and tenure, since I had been in my current position for under a year. The tenure policy in question contains an explicit exemption clause. The exemption exists. It was not used. In the same conversation it was acknowledged that I met the role’s qualifications, and it went unmentioned that the goals attached to my current role — which amounted to learning the position and progressing toward that pay band on a runway of roughly four years — were goals I had already exceeded. The same credentialism that operates between employers in the form of degree inflation and years-of-experience floors operates within them in the form of tenure-based seat-locking policies that ignore actual fit, even when the policy itself contains a door built for exactly this case.
I want to say a word about what tenure-based loyalty policies, considered honestly, are measuring. They measure how long the worker has been at the institution. They do not measure whether the worker is loyal, in any defensible sense of the word, to anything at all. My own working history contains, as one data point, a first job out of school at a sawmill that is no longer standing, because the people who were running it ran it into the ground through a combination of operational incompetence and ownership-class indifference to the underlying business. My most recent job, at PPG, was reposted overseas after my elimination, by a set of decisions made by people who could not afford the cost of their own choices and who, in the meantime, were compensated at multiples of any worker’s pay for the same calendar year. The C-suites that made those decisions walked away whole. I and the thousands of other workers downstream of those decisions took the financial hit. To ask me, in that context, whether I have been at the new company long enough to be considered loyal, and to use the answer as a basis for excluding me from a role I am otherwise qualified for, is to apply a definition of loyalty that has the relationship between cause and effect exactly backward. Loyalty, in the working version of the word that has any meaning at all, runs to the people one has worked with and for — the colleagues whose careers one has helped build, the managers who have invested in one’s development, the teams one has been part of through difficult stretches. It does not run to the legal entity whose name happens to be on the paycheck this month. The legal entity has demonstrated, repeatedly and at scale, that its loyalty does not run to the worker; it would be strange for the worker to be expected to maintain an asymmetric loyalty in return. Tenure-based loyalty policies measure something. They do not measure loyalty. They measure the institution’s preference for workers who have not yet learned that the question is the wrong question.
The second observation comes from the same conversation, which is what makes the pair worth recording. I had been accepted into a master’s program in artificial intelligence that aligned directly with the work. I applied for tuition reimbursement, a benefit that exists, on paper, for exactly this situation. The request was denied. The stated standard was that the benefit is granted when the degree aligns with future roles the employee would qualify for at the company. In the same breath, I had been told I already met the qualifications for the role I had just been blocked from — a role whose posting preferred the MBA I hold and did not require the master’s at all. Read the sequence that one conversation produced. I qualify for the next role. I cannot apply for the next role, because of a tenure policy whose exemption clause went unused. And the degree that would strengthen the case for the roles beyond it cannot be funded, because funding requires alignment with roles I would qualify for — a standard the same conversation had already conceded was met. Each position is defensible in isolation. Delivered together, in one sitting, they form a closed loop.
I want to name the trap this produces, because the trap is the thing the policies are functionally accomplishing whether or not anyone intends it. The work the next tier of roles involves is work I had been doing at PPG for years, on complicated systems, with measurably greater scope, and had received exceeds-expectations ratings on in formal performance reviews. The same person, with the same skill set and the same documented record, was told in a single conversation that he qualifies, that he may not apply, and that the education benefit cannot fund further credentials because funding requires the very qualification that had just been acknowledged. The tenure clock is the operative gate, and the exemption clause that exists for cases like this one sits unused. The Catch-22 is mechanical. The institutional language of level — the same word that operates as a credentialing gate at the industry scale — functions here, at the policy scale, as a self-sustaining gate. The wall is inside the building. The wall does not need to be repaired between visits. The wall is what the wall does.
There have been no further conversations on the subject since. I want to be careful about what I conclude from that, because the honest answer is that I do not know why the exemption went unused, and I am not going to claim otherwise. What I can say is that the why matters less than it seems to. A policy loop that no one inspects produces the same output that favoritism would — the credentialed, experienced candidate held in place, the gate sustained, the record unexamined — without requiring anyone to intend any of it. That is the point Essay 18 was building toward, and it is the reason this series keeps returning to structure rather than villains. The loop does not need a bad actor. It only needs to go uninspected. The credential gate and the polish gate are, in the end, the same gate. One of them is just written down.
These observations matter because they confirm that the pattern this series has been describing is not a PPG pattern. The PPG essays could, on a charitable reading, be dismissed as an extended commentary on a single workplace whose problems were specific to its leadership, its restructuring decisions, and its industry position. The observations from the new employer — the tenure policy as transfer gate, the education benefit denied under an alignment standard the same conversation conceded was met, the broader operational patterns Essay 18 and Essay 20 described — indicate the patterns are present at a different company, in a different industry, under different leadership, in a similar form. The series’s thesis is that the patterns are industry-wide and structural. The observations from the new employer are direct evidence for that thesis. The patterns are not unique to PPG. The patterns are how this category of work is now organized across the relevant industries. The PPG case is the case I have the most detailed knowledge of. The pattern is the pattern. And the pattern, named plainly, is bullshit.
What the Series Did
I want to summarize, before the closing, what the twenty-two essays have, taken as a body, attempted to do. The series began with the corporate values rhetoric of large multinational employers and the gap between the rhetoric and the lived experience of the employees subject to it. It moved through the operational mechanisms by which the gap is maintained — restructuring, performance management, metric systems, professionalism conventions, hiring rituals, evaluation rubrics. It documented one particular instance of the mechanism’s operation, a four-and-a-half-year employment at PPG that ended in an elimination the company has characterized one way and that I have characterized differently. It then moved into the recovery and rebuilding portion, which considered what is left when the institutional account has been refused and the personal account has had to be constructed in its place. The recovery essays considered identity, metric systems, professionalism, community service, transparency, and geography in turn. The capstone essay, which you are reading, is an attempt to place the entire body of work inside a larger argument about credentialed economies and the people they refuse to see.
The arc of the series is the arc Jerry Muller described in The Tyranny of Metrics, the 2018 book that has been a quiet background presence across these essays. Judgment, Muller wrote, is a sort of skill at grasping the unique particularities of a situation, and it entails a talent for synthesis rather than analysis. A feel for the whole and a sense for the unique are precisely what numerical metrics cannot supply. The series has been an exercise in the judgment Muller is describing. It has not been an attempt to produce additional metrics. It has been an attempt to hold particular experience and systemic analysis in the same frame — the operation Muller argues numerical-metric systems are structurally unable to perform. The HR file, which is the institution’s account, is the metric-only version of the story. The blog, which is my account, is the judgment version. The two are different in kind, not just in source. The blog can do what the file cannot. The file can produce what the blog cannot. Both are real. Both are partial. Both are present in the record. The reader can compare them and decide.
What was confirmed, in the period during which the series was being written, was external to me. The PPG team I had been part of continued to operate on the legacy reporting infrastructure I had been hired to modernize, with a former colleague confirming as much in unsolicited communication. The Computerworld ranking that had been eighteenth dropped to thirty-eighth in the following year. The internal employee survey results, according to the same source, continued to deteriorate. The metrics the institution maintains about itself, when they eventually moved, moved in the direction the original assessment had predicted. The institution is not in a position to acknowledge the convergence. The convergence is on the record regardless. The case the series has been making is not a personal grievance. It is an analysis. The analysis has been confirmed, to a degree that surprises me, by the institution’s own subsequent reporting.
Closing
The Wharton finance researcher Alex Edmans, in a study published in the Journal of Financial Economics, found that companies appearing on the Best Companies to Work For list between 1984 and 2009 generated approximately three and a half percent of stock-return alpha above the market and approximately two percent above their industry benchmarks. The causality, his work demonstrated, runs from employee satisfaction to subsequent financial performance, not the reverse. Satisfaction precedes earnings improvement. The effect has been replicated across fourteen countries. FTSE Russell confirmed that a Best Companies index returned approximately eleven and seven-tenths percent annually against the Russell 3000’s benchmark over a comparable period. Treating people well is not, on the available financial evidence, idealism. It is a better strategy than the prevailing alternative, by a margin large enough that the prevailing alternative’s continued dominance has to be explained on grounds other than performance. The grounds, when one looks at them honestly, are cultural and ideological. The institutions are doing what they have learned to do. The data says they should be doing something else. The gap between what they should be doing and what they are doing is the gap the series has been describing, restated in dollars.
The figure Coppola named in 1991 was the figure the credentialed economies of his moment had been built to overlook. The figure stood, in his phrasing, for an entire category of people whose work the credentialing apparatus was structured to refuse. The first generation of writers who claimed the figure read her primarily as a placeholder for technological possibility. The second generation claimed her identitarianly, recognizing themselves in the description and reading the description as a recognition of their position. The third reading, which is the one this essay has been making, claims her as a labor figure — the worker the credentialed economies refuse to see, the worker who picks up whatever tool is available and produces useful output without permission, the worker who is told she is not at the level and whose response is to do the work anyway. The descriptors Coppola used for her are the schedule of refusals the system was running. The prediction was that the refusals were about to lose their force.
Closing (Redux)
There is one more observation to make before the essay closes, and it is the observation the series would be incomplete without. The figure Coppola named in 1991 was a girl. The prediction was that she — specifically a she, with all the descriptors he used to mark her position outside the credentialing system — would be the figure the apparatus could no longer refuse. The first essay in this series began drafting in early 2024 with a working assumption about who the prophecy’s fulfillment was likely to be. The drafts the artificial intelligence tools I have been working with helped me produce reached, predictably and consistently, for examples of fulfillment that were male. Sean Baker. The Russo Brothers. The independent operator solopreneur Pieter Levels. The CEO who got promoted at Walmart. Every example of the prophecy’s realization that surfaced in the drafts, including the drafts of this essay, named a man. The figure Coppola named was a woman. The figure my drafts reached for, again and again, was not. The pattern is not subtle.
I want to name this directly because the pattern matters and because the integrity of the series requires the naming. The artificial intelligence tools used to assist with research and drafting are trained on a corpus in which the canonized examples of independent filmmaking, the celebrated solo founders, the success stories of credentialing disruption, are disproportionately male. The training data carries the bias of the cultures it was scraped from. The drafts the tools produce carry the same bias forward. The writer using the tools, if the writer is not paying attention, carries the bias one further step into the published record. I was, throughout most of the drafting process, not paying enough attention to this. The result is an essay in which the prophecy of the fat girl in Ohio is honored at the conceptual level and undercut at the example level, by a writer and an assistant who reached for the male examples without questioning why those were the examples that arrived first. The fat girl in Ohio is not in this essay’s list of fulfillments. Greta Gerwig is not in the list. Chloé Zhao, who won Best Director the year before Baker won Best Picture, is not in the list. Kelly Reichardt, Lulu Wang, Céline Sciamma, Ava DuVernay, Maren Ade, Joanna Hogg, Eliza Hittman, dozens of other working filmmakers whose work realizes Coppola’s prediction at least as directly as Baker’s does — none of them are in the list. The list I built was the list the training data made easy to build. The list the prediction was about would have been a different list. The list the prediction was about would have started with the fat girl in Ohio.
This is, in miniature, the entire argument the series has been making, turned on the series itself. The institutions one is operating inside, including the artificial intelligence assistants one is using to write, produce the outputs the institutions were structured to produce. The structural bias is upstream of the individual writer. The individual writer who is not noticing the upstream bias becomes a transmitter of it. The remedy is not to stop using the tools; the remedy is to notice what the tools are doing and to correct for it where the correction is in one’s hands. The correction, in this case, is the paragraph you are reading. The naming is late. The naming is not nothing. The point of catching it is that the next list will be better — the next essay, the next prompt, the next reader who picks up the thread — because the catching is what teaches the apparatus and the writer both. The work of correction is, it turns out, the work the series has been arguing for all along.
What Was Built
I want to land somewhere honest about what this body of writing has produced, because the temptation, at the end of any long project, is either to oversell the result or to undersell it out of false modesty, and the series deserves neither. What was built, across the twenty-two essays, is a record. The record is, as it stands, the most thorough description I have been able to produce of what a particular kind of corporate dysfunction looks like from inside it, what it costs the people on the receiving end, and what is left when the institution has finished with them. The record is not the institution’s account. The record is mine. It exists in public, at a domain I control, on a platform whose distribution is not gated by anyone’s permission but my own. The record is, today, free to anyone with the link — no paywall, no login, no permission slip. But free is a term, and the terms are mine to set. Whether the record stays fully open, or whether some future part of it sits behind a gate appropriate to what the writing has cost to produce, is a decision that belongs to no one else. I am, for the first time in a long time, the one setting the terms, even when the term I have set is free. This is the inversion of the prior arrangement. The prior arrangement gave the institution the narrative and the worker the silence. The current arrangement gives the worker the narrative and reserves to the worker the choice about how, and on what terms, the narrative is made available. The record is, in the technical sense Coppola was reaching for in 1991, the kind of artifact the apparatus was not designed to receive and cannot retract. The terms of its distribution are, accordingly, the terms of someone who has decided to own them.
The making of the record changed the person who made it. The Pennebaker research I cited above predicts this, and the prediction has, in my own case, been borne out. I am clearer on what happened than I was when I started writing. I am clearer on what was structural and what was personal. I am clearer on what I value, what I will not absorb again, and what I am building toward. The clarity is not a finished state; it is a working state, the state in which one can act from one’s own assessment rather than from someone else’s assessment of one. That state is what I have, and it is not nothing. It is, in fact, most of what a person needs.
The other thing that got built is a set of connections I would not have without the writing. Former colleagues have reached out. People going through their own versions of the same experience have written to compare notes. A volunteer life I had been carrying alongside the corporate one has, in the period the series covers, become more central to how I spend my time and how I think about what I am doing in this city. The community Essay 19 was about turned out to be more durable, and more genuinely sustaining, than the community I had thought I was building inside the corporate apparatus. The trade was not, on examination, a trade at all. It was a relocation.
And the work itself, the actual analytical and technical work I do for a living, is in better shape than it has ever been. The tools have changed. The training I have continued through has changed. The next thing I am qualified to do is wider than the thing I was qualified to do when this series began. The credentialed economy that decided I was not at a level lost, in the period covered by this writing, its ability to define what my level is. The level is what I can do. What I can do has grown.
So the series ends here, not on a verdict but on a working condition. The destruction Coppola predicted has begun. The construction is in progress. The figure he named is still in front of us, and the work of seeing her clearly — in the canonical lists, in the celebrated catalogs, in the assistants I and others use to write — is the work that continues. I am one of the people doing some of that work, in the modest portion of it that is in my hands. I have a record. I have community. I have skills. I have clarity. The little fat girl in Ohio is going to be the next Mozart. So is the kid in the trailer park, the kid in the bad school district, the kid whose father did not own a camera-corder but whose phone can now do what the camera-corder did, and more. The list is longer than the list I built. The list is being written, in real time, by the people who are on it. The work continues. The work goes well.