loading . . . COMMENTARY by Adrian Lenardic and Johnny Seales, “Why Write?” **Joan Didion** _, in her essay_ ‘Why I Write’ _, wrote that she writes to think [1]. Didion’s essay was inspired by an earlier essay, of the same title, by_**George Orwell** _[2]. Orwell’s reasons for writing are not equivalent to Didion’s, but there is an overlap. Orwell wrote that writers write to see things as they are—to find out. A writer who writes to find out is a writer who writes to think. Our essay title follows the lineage of homage but leaves out one word. We worry that, in the not-too-distant future, the title of this essay will become a declarative statement that there is no need to write at all. In that future, the thoughts of Orwell and Didion will be quaint memories at best._
* * *
Courtesy WordPress.com
There was a time when writers were advised to _write what you know_. We also take that to mean _not writing about_ _what you don’t know_(if one is honest, that’s a much longer list). We don’t know what it’s like to be a professional writer. So, we won’t write about that. Our focus is on how writing is presented to students as a general skill (part of a liberal arts education) and on a profession that relies on writing but is not defined by it. Both relate to what we do know from being immersed in two worlds:
Courtesy Penguin Books
1) The world of higher education, and
2) The world of research science.
Within both worlds, we have seen changes in the perceived value of writing. The changes have moved from _Good Intentions_ to _Perverse Incentives_ to _No Incentives_
A range of university departments, beyond English and Journalism, have writing exercises as part of their classes. The foundational idea is that one need not aspire to be a professional writer to benefit from writing skills – a good intention. Perverse incentives arise when, as we have observed, the message presented to students is that the primary goal of developing writing skills is to help them secure a job and advance in their chosen profession.
That the goal of writing is to get a job reinforces the older idea that the goal of writing is to get an A on a class essay. These are externalized goals. If the goal of doing something is some external reward, not intrinsic to the thing itself, then why not save time doing it? Why not take shortcuts if they are available? Why not outsource the chore if outsourcing is possible?
In 2025, the outsourcing of writing is correlated with artificial intelligence (AI) and large language models (LLMs). It’s often assumed that the correlation shows causation: AI and LLMs are _the_ cause of _the_ decline of writing. That’s an easy tale to tell. It offers a clean story with a clear villain – an artificial one at that. The reality is that an artificial bogeyman can not absolve humans. Outsourcing, shortcuts, and other choices that stem from externalized reasons for writing are older than AI.
**John Traphagan** wrote an essay about how essay writing in educational settings has moved toward perverse incentives, paving the way for AI and LLMs, which are not the cause but a natural result [3]. Artificial intelligence and LLMs fit into the pre-paved path so perfectly as to be, in effect, the inevitable endgame.
Traphagan observes, from his perspective as a professor, that “schools have been programming students to write uninspiring, formulaic essays for years.” This began with a good intention: students should know why an essay received the grade it did in “the quaint olde days,” which would come from discussions with the instructor. The discussions would treat an essay as a holistic work of thought, creativity, and technique. Some time ago, academic managers decided that it was inefficient, not sufficiently objective, and did not serve the customer (i.e., the student). That led to the rise of rubrics.
When our university decided that writing skills could be beneficial for students across a range of departments, it accompanied this decision with training seminars for professors and teaching assistants. One seminar that stands out had a colleague presenting grading rubrics to us and espousing how they allowed for transparency and eased the burden of teaching.
Rubrics would also enable students to transition smoothly to writing at the university level. They could ease time burdens on students by allowing them to write more efficiently. Along with rubrics, essay topics would be replaced by essay prompts to help students know what was expected. Like many training seminars for university students designed to help them as customers, it was presented as an advertisement for educational progress. Like many ads, the promises fell short.
Rubrics are formulaic. They break writing down to a list of categories an essay is expected to contain and provide look-up tables that correlate essential elements within a category to points awarded or deducted per category. In effect, they provide a grading formula.
Graders grade to the formula, and writers write to the formula to get the desired grade (with the added aid of essay prompts). It’s less about creation and more about rules-based management [4]. Anyone with a rudimentary knowledge of computer programming will see that this lends itself naturally to an algorithm. Why would students continue to walk through the formulaic steps of writing an essay after a tool that is naturally suited to the chore has become available? That hits on Traphagan’s point: _The path to using AI for writing was paved well before AI._
Rubrics undermined the value of writing as a means of thinking. Writing to a rubric involves thinking, but it’s an artificial form of thinking. The messier human mode is bypassed in favor of following a formula to gain a reward – a move to perverse incentives.
Once something can follow the formula faster, why not use it – a move with no incentive. This follows a broader path that moved higher education from a place of learning to a place of training. Students became trained to write before AI was given training data to help it ‘learn’ how to write essays for students.
The shift from learning to training reflects universities adopting corporate management styles. Students became customers, and learning became knowledge transfer (a commodity exchange). The value of education is customer employability. Boxes to be checked minimize customer dissatisfaction as everything is transparent, objective, and quantifiable. Rubrics fit the need, as do precise “learning objectives” [5].
Providing the same learning experience for all customers is a sign of product consistency, and all paying customers deserve the same level of product quality. Very cool, clean, and efficient – like a shiny machine. Writing skills become another product. Something to be employed. Something external. Writing to think becomes as foreign as reading Orwell or Didion (not useful for employability). Any lingering worries about outsourcing and shortcuts are squelched as public relations departments come in to advertise how universities are committed to the use of ‘responsible AI’ [6].
But none of the above could creep into professions that rely on writing, could it? If any profession values _writing to think,_ it’s science. The education of scientists would respect that, would it not? A quaint thought.
In many science departments, what constitutes a PhD thesis changed some time ago (~20 years ago in our department). A document written by a single author (the “old school” thesis) became a collection of published, multi-author papers (often with authors not at the PhD-granting institution).
A good intention is that the publication can be a source of pride. That was not, however, the prime motivation. The primary motivation stemmed from a shift in the perceived purpose and value of published papers.
Published papers had become a form of capital, not only for authors but also for universities. Funding for students increasingly came from grants awarded to faculty advisors. Those grants came with product expectations – published papers. A more significant shift occurred when universities became corporatized and competitive with one another [7]. A prime way for universities to compete is to highlight the grants awarded and papers published by their faculty.
Pressure to publish increased as promotions came to rely on the number of papers and grants obtained per year, with competition among universities driving the expected numbers upward. As grants funded many PhD students, it made ‘sense’ to require them to produce published papers. Even if grants did not fund some, the university would gain competitive capital through an increase in the number of published papers. This morphed writing a thesis into producing academic capital for oneself, one’s research group, and one’s university.
From the earliest days of being mentored on how to be a scientist, students are now immersed in the paper-as-a-product model [8]. This view persists throughout a scientist’s career, particularly at universities that use paper counts for hiring and promotion decisions [9]. Perverse incentives follow. When scientific papers are turned into products, a separation occurs between conducting science and writing up the results.
Writing is not seen as an integral part of research; it’s a hurdle to clear in getting science out the door and into publication. What becomes undermined is the idea that writing is an active process of science; that writing can expose flaws or gaps in the research that need to be corrected, or that the research is tenuous or not publishable.
Thinking occurs when one does scientific research, but the idea that one writes up the science to think takes a back seat to the idea that one writes to produce a paper that will get published, i.e., to produce academic capital. Not just to produce capital, but to produce it efficiently – the ‘per year’ in papers per year metrics are powerful words.
A published paper is not the same as an A-grade essay, but they are both externalized goals. Writing papers to get them published may not come with a rubric handed out to scientists, but it has its own heuristics, rules, and formulaic writing that are learned over time. Something else that has been learned is that science papers becoming capital for universities is capital for publishing companies.
Scientific publishing, once a non-profit venture, has evolved into a for-profit industry that benefits from more scientists writing to get their papers published as efficiently as possible. That, together with papers becoming capital for universities, has helped fuel a continual increase in the number of papers published per year [10]. One could blame AI and LLMs for a glut of papers by suggesting that scientists are outsourcing writing to machines. Publishers have suggested just that [11]. The irony, as with universities having fueled formulaic writing before AI, is that publishers feed on and fuel the outsourcing of writing [12]. Figure 1 illustrates how they did so before blaming artificial bogeymen for the overproduction of scientific papers.
_Figure 1: Images from emails we received from Springer Nature, one of the largest for-profit science publishers. The left panel shows that Springer wants to help scientists be more efficient. If they are not, that will limit the number of papers sent out for publication, and Springer views that as detrimental to both science and profits. The right panel opens with: “It’s been a time since you last submitted work for publication …” In effect, Springer wants to help scientists be more productive via a gentle nudge that says: ‘Shouldn’t you be publishing more?’._
Writing is hard. It can hurt one’s head [13]. George Orwell was aware of that: “Writing a book is a horrible, exhausting struggle …” [2]. Academic managers saw it as a problem that needed to be solved. They lowered the hardships and the wasted time for their clients by pushing rubrics. Springer saw it as a problem that came with a new source of wasted time for their clients (i.e., not writing at a ‘strategic level’ that leads to publication). In an altruistic move to help time-stressed faculty, they offered help of the type in Figure 1 [14].
Why agonize about writing when Springer can save you the headache, just as rubrics save students from headaches. Both save time and increase the chances of rewards. Why not follow templates or outsource to get your rewards more efficiently? You’re on the path to automating your writing.
Wouldn’t it be great if something that is faster in the automated realm came along to save even more time? Enter AI and LLMs to free you from all the agony and, as a bonus, to provide a villain should a villain be needed to pin a deterioration of writing skills, or an overproduction of science papers, on.
In 2025, any changes in writing, from education to the professional level, are attributed to AI (an overgeneralization, but not by much). Is AI the cause of a decline in writing amongst students, be it in quality/creativity or in fewer students wanting to write at all, akin to a decline in desire and willingness to read [15]. Is AI the cause of an increase in the number of scientific and academic papers published per year? Did AI subvert the intrinsic value of writing? Did AI train students and scientists that writing should produce academic capital? If our LLM is worth its cost, then you should already know our answers.
If the goals of writing are external (a grade, a publication, capital gain), then it makes sense to want to know exactly how to write to achieve the goals. Rubrics, templates, and formulaic writing recipes become logical [16]. They allow you to adjust your writing and the time you put into it to hit the target. The reward is clear and singular. The goal is to reach it as efficiently as possible with minimal risk. If a new technology allows you to do just that, why not use it?
The technology is not the driver. _The driver is a glorification of efficiency, productivity, and external reward that has crept into education, a field whose core values are_**NOT** _about any of those things._ An intrinsic value of science overlaps with the value of education over training and with what Didion and Orwell saw as a value in writing: A desire to “see things as they are” [17].
If that is not seen as its own reward and future generations are conditioned to the idea that writing is for external gains, then we should not be surprised to hear more declarations of _‘Why Write!_ ’
AL & JS July, 2025
Cover graphic courtesy 123RF.com
**Notes and References**
[1] Didion, J., 1976, Why I Write, _The New York Times_ , https://jarrettfuller.com/design-writing/downloads/Didion.pdf
[2] Orwell, G., 1946, Why I Write, _Gangrel_ , https://www.orwellfoundation.com/the-orwell-foundation/orwell/essays-and-other-works/why-i-write/
[3] Traphagen, J., 2023, Should We Be Worried About AI in School?, _Future U – Neoliberalism in Higher Education_ , https://futureu.education/education/commentary-should-we-be-worried-about-ai-in-schooling/
[4] Following a rubric becomes akin to following a cooking recipe, step-by-step, to become a chef. Some readers may recall the guitar hero craze – a video game that had players hit guitar ‘chords’ displayed on a scrolling screen (the guitar was a plastic prop). If you could hit the chords as quick as the screen told you to, you became a ‘rock star’; follow the rules efficiently and you become be a musician.
[5] The words ‘learning objectives’ remind us of another Orwell essay: Politics and the English Language (Orwell, G., 1946, _Horizon_ , 13(76), http://www.public-library.uk/ebooks/72/30.pdf). In it, Orwell discusses pretentious words and words that lack clear meaning. The vagueness and grand sound of the words allows those who use them to do so in ways that lead readers into hearing them in a better light than that of the narrative that is being pushed forward. In this case, the narrative that the student as customer model is for the good of students and, more generally, for the good of an educated society.
[6] See endnote 5. If anyone wants to hire us for some university PR, here’s a taster: ‘By using responsible AI, we provide our students with a return-on-investment that will develop them into thought-leaders who will lead the knowledge economy to new levels of sustainability’.
[7] Mittelman, J.H., 2018, _Implausbible Dream:The World-Class University and Repurposing Higher Education_ , Princeton: Princeton University Press.
[8] For some, it starts in high school. Published research is now a form of capital that can increase the chances of being admitted to a university. Services are available to help high school students generate published papers (https://www.propublica.org/article/college-high-school-research-peer-review-publications ). The increase in for-profit publishers, eager for publications, makes the potential of being a published researcher out of high school ever more viable given the ‘appropriate mentoring’.
[9] Scientists have always produced papers, and papers could always generate rewards. However, papers were not always viewed as products, and the nature of rewards has changed. Prior to metric based assessments, papers were a means to communicate with colleagues and to establish priority of an idea (e.g., de Solla Price, Derek J., 1963, _Little Science, Big Science_ , New York: Columbia University Press). Rewards could be garnered in the form of enhanced reputation. That depended on the quality of the science and the logic and structure of its presentation. Both would be assessed by the science community under no fixed time frames. The introduction of paper counts to evaluate faculty annually, changed writing papers into producing quantifiable products over a fixed time window. Papers can still enhance reputation, but that is no longer the sole reward. The variable time frame for reputational reward, in a system that now assesses a scientists’ value on an immediate time scale, affects how that reward is balanced against those that come from yearly productivity. An interview with a Nobel laureate sums up the changes. In an interview with the Guardian, Peter Higgs noted that he would not be able to get an academic job, under the modern system, as he doubts he would be regarded as productive enough (https://www.theguardian.com/science/2013/dec/06/peter-higgs-boson-academic-system).
[10] Bornmann, L., Haunschild, R., & Mutz, R., 2021, Growth rates of modern science: a latent piecewise growth curve approach to model publication numbers from established and new literature databases, _Humanities and Social Science Communications_ , 8, 224. https://doi.org/10.1057/s41599-021-00903-w.
[11] Prillaman, M., 2024, Is ChatGPT making scientists hyper-productive? The highs and lows of using AI, _Nature_ , 627, pp16-17. https://www.nature.com/articles/d41586-024-00592-w
[12] Lenardic, A., 2024, Absurdities, Ironies, Hyper-Production and Artificial Bogeymen, _Zenodo_ , https://zenodo.org/records/12746889
[13] Although our focus is on writing, the fact that some things are difficult extends more generally to learning, as argued for by William Byers: “No one can learn for someone else. You cannot magically impart understanding to another. Learning, as I have repeatedly stressed, is intrinsically difficult” (Byers, W., 2014, _Deep Thinking_ , World Scientific Publishing). The use of the word ‘magically’ takes on a modern twist as many proponents of LLMs, who stand to gain from their adoption, present them as if they are magic (Nagy, P., and Neff, G., 2024, Conjuring algorithms: Understanding the tech industry as stage magicians, New Media and Society, 26(9), 4983-4954). The quote from Byers highlights a difference between learning and understanding versus being trained. The former is intrinsically more difficult. The student-as-customer model seeks to ease difficulties, so a customers’ path is smooth and efficient. This allows for training, that can serve a student in the job market, but it undermines education: “The existence of obstacles is an essential feature of education, which is often ignored” – William Byers.
[14] Springer is not the only publisher saving scientists time and it’s not just publishers. The for-profit company Academia.edu, which runs a privatized platform that scientists can share papers on, offers a service that automates grant proposal writing by taking an existing proposal and multiplying it into new proposals targeted at new funding opportunities – it’s called The Grant Multiplier (https://www.academia.edu/grant_multiplier/landing). Academia.edu knows that faculty being graded on how many grants they acquire need to expand their reach. Writing more grants is a headache. Let The Grant Multiplier do it for you. Everyone else is doing it. Your competition is doing it. Nuff said.
[15] Malesic, J., 2024, There’s a Very Good Reason College Students Don’t Read Anymore, _The New Your Times_ , https://www.nytimes.com/2024/10/25/opinion/college-university-students-reading.html
[16] That some things are logical within a system, even if illogical outside of it, makes placing blame on individuals, for choosing to outsource writing, akin to blaming artificial bogeymen. Both are misguided. Sociologists have long argued that systems are not just people and that assuming they are places false blame on individuals (Johnson, A.G., 2014, _The Forest and the Trees_ , Philadelphia: Temple University Press). A student choosing to use an LLM for an essay or a time-stressed junior faculty member choosing to do so to finish a paper, did not create the system they find themselves in. To blame them for making choices within a system that favors those choices walks the line of ‘blaming the victim’. Well intentioned individuals can do things within a system that they would not do outside of it. The questions to ask are how the system came into existence, and can it be restructured to not favor detrimental behavior. Or blame it all on robots, lazy students, and greedy faculty (that’s easier and more efficient).
[17] “To see things as they are” implies a regard for how things really are. The opposite of that is “without regard for how things really are” and that constitutes bullshit (Frankfurt, H., 2005, _On Bullshit_ , Princeton, NJ: Princeton University Press.). We could cite papers calling out a rise in bullshit and bullshit writing but better to let readers determine for themselves if that’s “how things really are”; if it is becoming harder to see things “as they are”; if there is less need to see things “as they are”; if AI is enhancing both; if it’s become more acceptable to see things as one sees them. We will, however, add a quote from another quaint writer: “The most essential gift for a good writer is a built-in, shockproof, shit detector” – Hemingway.
[18] One last quote to help this essay get us a big money book contract: “The great enemy of clear language is insincerity.” – Orwell.
Post Views: 199
Tweet
Pin It https://futureu.education/uncategorized/commentary-by-adrian-lenardic-and-johnny-seales-why-write/
