Why We Write
An hours ledger, the meta analyses on writing to learn with their limits stated, and an honest account of what a year of publishing changed here and what it did not.
Date
Category
Writer
The third piece I published in this journal was read by eleven people in its first month, two of whom were me on different devices. It had taken about nine hours to write and another three to check. That is a cost of roughly an hour per reader, for a studio whose hours are the only thing it sells.
A year later I am still doing it, and this piece is an attempt to say why in terms that would survive an argument with an accountant. Not because writing is good for the soul, which may be true and is not a reason, but because there is a measurable case and a set of measurable costs, and both deserve to be stated plainly at the end of a year.
What it costs
Here is the ledger, as honestly as I can reconstruct it. These are my own figures for this journal, not a benchmark for anyone else.
Nine to eighteen working weeks. There is no framing of that number that makes it small. Any argument for a studio journal has to be worth roughly a fifth of a year, and most arguments for one are not.
Photograph by Tookapic, CC0, via Wikimedia Commons.
What the research actually supports, and what it does not
The claim that writing improves thinking is old enough that people repeat it without a source. There is a literature, and it is more modest than the slogan.
Bangert-Drowns, Hurley and Wilkinson published a meta-analysis of 48 school-based writing-to-learn programmes in the Review of Educational Research in 2004. Their conclusion, verbatim: writing "can have a small, positive impact on conventional measures of academic achievement". Two factors predicted better outcomes: "the use of metacognitive prompts and increased treatment length". Two predicted worse ones: implementation in grades 6 to 8, and, strikingly, longer writing assignments.
Graham, Kiuhara and MacKay revisited the question in the same journal in 2020 with a tighter inclusion standard, 56 experiments, restricted to science, social studies and mathematics. Their finding: "writing about content reliably enhanced learning (effect size = 0.30)", equally effective across the three subjects and across elementary, middle and high school students, and "not moderated by the features of writing activities, instruction, or assessment". An effect size of 0.30 is real and unspectacular.
Two honest caveats. First, both bodies of work study schoolchildren learning assigned content, not professionals reasoning about their own field, and I know of no evidence that the effect transfers cleanly. Second, the 2004 result that longer assignments predicted reduced effects is inconvenient for anyone justifying 2,500-word essays, and I am not going to pretend otherwise. The strongest reading the evidence supports is: writing about material produces a small reliable improvement in how well you hold it, and structured prompting helps more than volume.
The illusion you are paying to escape
The more useful finding is not about learning at all. Leonid Rozenblit and Frank Keil, in twelve studies published in Cognitive Science in 2002, demonstrated that "people feel they understand complex phenomena with far greater precision, coherence, and depth than they really do". They called it the illusion of explanatory depth. Its diagnostic is simple: ask someone to rate their understanding of something, then ask them to write a full mechanistic explanation, then ask them to rate it again. The second rating is lower.
The detail that matters for a design studio is where the illusion lives. Rozenblit and Keil found high miscalibration for devices and natural phenomena, and minimal overconfidence for procedural knowledge and narratives. People are roughly right about how well they can do a thing, and badly wrong about how well they can explain how it works.
Design practice is almost entirely in the second category. A designer can reliably produce a good grid and be entirely unable to say what a grid is for without reaching for received phrases. The studio's craft knowledge is procedural and mostly sound; the studio's stated theory of its own craft is explanatory and mostly untested. Writing is the cheapest available instrument for finding out which parts of what you believe are actually load-bearing, because the explanation either assembles on the page or it does not.
This has happened to me in a specific and repeatable way. I have started at least four pieces confident of a claim I had been making to clients for years, gone looking for the source, and found that the popular version was a distortion, or that the underlying study said something narrower, or that there was no study at all. Those are not writing failures. They are the writing working.
Publishing is a different act from writing
A private notebook does not produce that effect at the same strength, and I think the reason is accountability rather than effort.
Jennifer Lerner and Philip Tetlock reviewed the accountability literature in Psychological Bulletin in 1999, asking among other things "Under what conditions will accountability attenuate, have no effect on, or amplify cognitive biases?" and "Does accountability alter how people think or merely what people say they think?". One of the broader implications they draw is "the utility of treating thought as a process of internalized dialogue". That framing describes the experience of drafting for publication precisely. You are not recording conclusions. You are conducting an argument with an imagined reader who is better informed than you and slightly annoyed.
The imagined reader does work that no editor does, because the imagined reader is the one who will follow the link. Knowing that a figure is going to sit in a public sentence next to a URL is what turns "I remember reading that" into forty minutes of source-checking. I have deleted more claims than I have published, and almost all of the deletions happened at that moment.
Writing about something, even something you know well, usually shows you that you didn't know it as well as you thought.
That is Paul Graham, in "Putting Ideas Into Words", February 2022. Two other lines from the same essay match my experience exactly: "Half the ideas that end up in an essay will be ones you thought of while you were writing it", and "No one who hasn't written about a topic has fully formed ideas about it." The second is a strong claim and I am not sure it is true as stated. The first I would now defend on evidence.
The pieces nobody reads
Now the part that most writing-about-writing leaves out.
Ahrefs published a study on 1 December 2023 analysing roughly 14 billion pages from its Content Explorer database. The headline: "96.55% of all pages in our index get zero traffic from Google", with a further "1.94% get between one and ten monthly visits". The authors are careful about their own limits: the 14 billion pages are a subset of a 340.8 billion page index and skew towards higher-quality content, and traffic estimates come from roughly 651 million keywords, so very long-tail queries are missed. Take those caveats seriously and the picture is still stark. The overwhelming default outcome for a published page is silence.
That has been true of this journal. The distribution of readership across a year of pieces is not a gentle curve; it is two or three pieces carrying almost everything and a long tail of work that was read by a number of people small enough to count. Some of the tail pieces are, in my own judgement, the best work here. There is no relationship I can detect between how good a piece is and how many people find it, and I no longer expect one.
What a year changed, and what it did not
What I expected | What happened | My confidence |
|---|---|---|
Writing would become faster with practice | Drafting got marginally faster; verification did not, because the standard rose | High, measured against my own logged hours |
Publishing would generate direct enquiries | A small number, none attributable with any confidence; no counterfactual exists | Low, and I will not claim otherwise |
I would hold more of what I read | Yes, and the research predicts a small effect of roughly this size | Moderate; consistent with the literature, not proof of it |
Client conversations would improve | Clearly yes; explanations I had drafted arrived intact in meetings | Moderate, and entirely self-reported |
Beliefs about my own craft would survive checking | Several did not, including ones I had taught | High, because each correction is documented in a published piece |
The last row is the whole case. Over a year, the journal functioned as an audit of things I believed, conducted at a rate of roughly two beliefs a month, with the corrections written down where I cannot quietly forget them.
What it did not change is worth being equally clear about. It did not make the studio busier in any way I can demonstrate. It did not make me a faster designer. It did not produce an audience of a size that would interest anyone selling advertising. And it did not resolve the central uncertainty, which is whether nine to eighteen weeks of a small studio's year is well spent on this rather than on the work itself. I do not know. I have one year of evidence, a sample of one studio, and no control.
What I can say is narrower and I believe it holds. A studio that writes has to find out whether its claims are true. A studio that does not write can hold an inherited opinion for a decade without ever being contradicted by anything. Rozenblit and Keil measured the gap between those two positions; the cost of closing it, in this practice, is about twenty hours a piece. Whether that is a good trade depends on how much it matters to you that the things you tell clients are actually so.
One practical note for anyone starting. Put the writing on infrastructure you control. Measured decay on the open web is severe: Pew Research Center found that 38 percent of webpages that existed in 2013 were no longer accessible a decade later, and a 2016 study in PLOS ONE found that more than three quarters of surviving referenced pages had drifted from what they originally said. A journal on a platform you do not own is a journal with a half-life, and everything above is an argument about a long horizon.
Sources
Robert L. Bangert-Drowns, Marlene M. Hurley and Barbara Wilkinson, "The Effects of School-Based Writing-to-Learn Interventions on Academic Achievement: A Meta-Analysis", Review of Educational Research, volume 74, issue 1, spring 2004, pages 29 to 58, doi 10.3102/00346543074001029. Meta-analysis of 48 programmes; small positive impact; metacognitive prompts and treatment length as positive moderators; grades 6 to 8 and longer assignments as negative ones. https://journals.sagepub.com/doi/10.3102/00346543074001029
Steve Graham, Sharlene A. Kiuhara and Meade MacKay, "The Effects of Writing on Learning in Science, Social Studies, and Mathematics: A Meta-Analysis", Review of Educational Research, volume 90, issue 2, 2020, pages 179 to 226, doi 10.3102/0034654320914744. 56 experiments; effect size 0.30; consistent across subjects and grade bands; not moderated by features of the writing activity. https://journals.sagepub.com/doi/abs/10.3102/0034654320914744
Leonid Rozenblit and Frank Keil, "The misunderstood limits of folk science: an illusion of explanatory depth", Cognitive Science, volume 26, issue 5, 2002, pages 521 to 562, doi 10.1207/s15516709cog2605_1. Twelve studies; the illusion of explanatory depth; high miscalibration for devices and natural phenomena and minimal overconfidence for procedures and narratives. https://www.sciencedirect.com/science/article/abs/pii/S0364021302000782
Jennifer S. Lerner and Philip E. Tetlock, "Accounting for the effects of accountability", Psychological Bulletin, volume 125, issue 2, March 1999, pages 255 to 275, doi 10.1037/0033-2909.125.2.255. The four questions the review addresses, and the implication of treating thought as a process of internalised dialogue. https://pubmed.ncbi.nlm.nih.gov/10087938/
Paul Graham, "Putting Ideas Into Words", February 2022. The passages on writing revealing what you did not know, on half an essay's ideas arriving during writing, and on fully formed ideas. https://www.paulgraham.com/words.html
Ahrefs, "96.55% of All Content Gets Zero Traffic From Google", published 1 December 2023. Analysis of approximately 14 billion pages from a 340.8 billion page index; the zero-traffic and one-to-ten-visit figures; the authors' own caveats on sample skew and keyword coverage. https://ahrefs.com/blog/search-traffic-study/
Pew Research Center, "When Online Content Disappears", 17 May 2024. 38 percent of webpages from 2013 no longer accessible a decade later. https://www.pewresearch.org/data-labs/2024/05/17/when-online-content-disappears/
Shawn M. Jones, Herbert Van de Sompel, Harihar Shankar, Martin Klein, Richard Tobin and Claire Grover, "Scholarly Context Adrift: Three out of Four URI References Lead to Changed Content", PLOS ONE, 2016, doi 10.1371/journal.pone.0167475. Content drift affecting over 75 percent of checked references. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0167475
Related reading
Latest Articles.
© Moc Mallen Studio
Thoughts, ideas, and perspectives on design, simplicity, and creative process.
