Abstract
A change to a sentence can be small in the grammar and large in the work. A writer or editor might add one qualifying phrase, move one paragraph, delete one repeated claim, or answer one proof query. Yet the revision is never only the moment at which letters become different. Someone has to find the right place, decide what kind of change is needed, preserve or discard an earlier wording, compare alternatives, make the result legible, and sometimes explain the alteration to another person. Different writing environments distribute that work differently.
This paper calls that distribution revision friction: the situated pattern of effort, delay, risk, visibility, reversibility, and expectation that gathers around making or inspecting a textual change. The term does not mean that friction is inherently good or bad. A paper typescript with handwritten marks can make accumulated negotiation visible while making clean recopying costly. A word processor can make local substitution and rearrangement relatively easy while making earlier decisions disappear unless a comparison practice is retained. A review-visible environment can make a change inspectable while also creating a need for explanation, response, and care about who is watching.
Drawing on a university archive of annotated typescripts and rewrites, a history of word-processing software, writing-process research, editorial guidance, accessibility scholarship, and studies of reflective revision, the paper introduces the Revision Friction Ledger. The ledger gives writers, editors, teachers, and tool designers a way to describe what a revision environment makes easier, what it relocates, what trace it creates, and what new demand it quietly installs. The central lesson is modest but consequential: do not ask whether a tool makes revision easy until the revision work itself has been named. [1] [2] [3]
Working principle: A revision technology changes more than the effort to alter words. It changes the comparison, trace, audience, and expectation through which an alteration becomes responsible work.
1. The small revision that is not small
Consider a constructed teaching specimen. A paragraph originally says, “The new publishing system eliminated the distribution problem.” An editor inserts one word: “The new publishing system reduced the distribution problem.” The visible difference is tiny. The intellectual work is not. The revised sentence changes a causal claim, keeps the paragraph from promising too much, and may require the surrounding evidence to be reconsidered.
Now place the same change in three conditions. In the first, the sentence appears on a paper typescript. The editor writes a caret between two words, inserts reduced, perhaps circles eliminated, and returns the page for retyping. In the second, the sentence appears in a word-processor file. The editor selects one word, types another, and saves. In the third, the change appears in a review layer that shows a deletion and an insertion to the author, another editor, or a production colleague.
It would be easy to call the second or third condition faster and end the comparison there. But the question abandons the interesting work. In the paper condition, the physical mark preserves the local encounter between old and new wording, yet the page may grow difficult to read and the next clean copy may require recopying. In the file condition, the local substitution is immediate, yet the old wording may vanish unless the writer deliberately preserves a comparison path. In the visible-review condition, the older wording can remain inspectable, but the change can acquire an audience and an obligation to explain why it was made.
The three conditions do not measure a ladder from primitive to advanced. They show that a revision has several components: a unit to be changed, a purpose for changing it, an action that produces the change, a way to compare the changed text with something else, a trace that may remain, an audience that may see it, and an expectation that follows. The one-word alteration is therefore a useful problem because it prevents us from treating revision as a single, self-evident operation.
This paper uses revision friction to name that multi-part condition. Friction is not the number of seconds needed to type a word. It is the patterned work that a particular environment asks of a writer, editor, or reader in order to make, inspect, undo, justify, or carry forward a change. A revision can be mechanically easy and cognitively difficult. It can be hard to enter but easy to review. It can create a durable record while making a private working process newly visible. The point is to describe the pattern before assigning a value to it.
The constructed specimen is not a reconstruction of any historical author’s practice. It is a teaching device. Its purpose is to make a later question possible: When a tool lowers one kind of revision friction, which other kinds remain, move, or newly appear?
2. Revision is not one activity
A second mistake follows quickly from the first. If revision is treated as a single task, a sentence-level substitution, a structural reorganization, a consistency pass, and a final proof correction can all appear to be instances of one thing that simply takes more or less time. Professional editorial guidance instead distinguishes different kinds and stages of work. The Australian Government Style Manual describes substantive editing as work on content, structure, style, and meaning; copyediting as work on consistency, accuracy, and completeness after content is settled; and proofreading as final quality assurance against a master copy or production state.[4]
This institutional taxonomy is not a universal law of publishing. Projects differ, and a writer may perform several activities at once. Its value here is diagnostic. A revision environment does not have one cost. A tool that makes it easy to rearrange paragraphs does not necessarily make it easy to compare a proof against an approved master. A visible history that helps an editor inspect an insertion may create distraction when a writer is trying to rethink a whole argument. A system that captures every keystroke may generate a rich process trace without making a substantive decision any clearer.
The phrase revision friction therefore needs two limits. First, it is descriptive rather than evaluative. A delay can be a harmful barrier, a necessary moment of checking, or a neutral fact of a particular workflow. Second, it is situated rather than universal. The same keyboard shortcut, commenting feature, paper copy, or comparison view may be supportive for one writer and inaccessible, distracting, or socially risky for another. The ledger later in this paper is designed to keep those conditions visible.
3. A revision history is made of different kinds of work
A university archive can show why a clean technological story is too simple. The Rudolfo Anaya Digital Archive presents a documented development sequence for Zia Summer: early typescripts with handwritten editorial comments and corrections; rewrites that incorporated earlier comments and became relatively clean; later drafts with comments from the author’s wife and literary agent; material submitted to Warner Books and returned with editorial comments; and subsequent clean or re-edited drafts.[1] The collection also names notes, correspondence, typeset manuscripts, and page proofs among the surviving materials.[1]
The archive does not entitle a reader to infer the author’s private motives, the exact labor of every participant, or the complete chronology of decisions. It does, however, permit a narrower observation. Revision in this documented sequence was not only an alteration to a text. It involved marked pages, cleaned-up rewrites, correspondence, institutional return, and successive states in which some earlier negotiation was absorbed into a new copy while other negotiation remained visible in marks or notes.
That distinction matters. A heavily annotated page can make local discussion conspicuous, but it may not make a long passage easy to read as a finished whole. A clean rewrite can make a new reading possible, but it can also remove the immediate visual accumulation of prior comments. Neither form is more truthful by itself. They place different demands on anyone who wants to see what changed, what must be copied next, or who participated in the work.
The archive thus offers a material contrast rather than a nostalgia object. It warns against saying that earlier revision was merely slow and digital revision merely quick. Recopying can be a cost, but it can also be a way of reencountering the sentence in a clean context. Marking a page can preserve a local decision trail, but a dense marked page can make future reading laborious. The relevant question is not whether a paper page resists change. It is which revision functions the page makes visible, difficult, legible, or transferable in a particular sequence.
4. The word processor was never one thing
The phrase the word processor often covers too much historical ground. Otto Kruse and Christian Rapp describe a field in which more than 300 word-processing software versions existed in the mid-1980s before Microsoft Word emerged as a standard writing tool.[2] Their chapter also notes the later potential of keylogging and screen recording to change how writing processes could be studied.[2] The account matters because it replaces the simple image of a typewriter suddenly becoming a computer with a history of proliferating interfaces, files, conventions, and research possibilities.
A word processor can reduce the effort of local substitution, copying, and rearrangement. That observation is real but incomplete. It does not state whether a user can recover an earlier sentence, compare two copies, distinguish a writer’s revision from an editor’s, or see which version has been approved. It does not state whether the display supports rereading, whether the file is accessible to a collaborator, or whether the new ease of producing a clean copy creates a new expectation that a writer will revise again before sending work onward.
What changes, then, is not only the action of replacing characters. A word-processing environment can distribute revision into new combinations of selection, navigation, display, saving, duplication, comparison, and sharing. It can lower the material burden of a move while raising the importance of file naming, checkpoint habits, review procedures, or disclosure about which state is current. These are not defects that cancel the benefit of a faster local change. They are reasons to describe the benefit precisely.
This is also why the paper does not claim that word processing made writers universally freer, more productive, or more capable of revision. Adoption is uneven; interfaces differ; access conditions differ; and a feature that removes one form of labor may impose another. The historical contrast is valuable when it asks which work has been redistributed, not when it announces a winner.
5. A text can be revised as a product—or observed as a process
The contrast becomes sharper in contemporary process research. Rianne Conijn and colleagues distinguish product-oriented revision analysis, which examines the finished text produced, from process-oriented analysis, which examines the activity through which the writer produced it.[3] Their public dataset uses keystroke and eye-tracking data from 65 students writing argumentative texts and academic abstracts; it identifies 7,120 revisions and includes manually annotated and automatically extracted features.[3]
The study does not tell us how all writers revise. Its student population, language backgrounds, and tasks are specified. Nor does a measurable revision become an improved revision merely because it has been captured. The study nevertheless changes the teaching problem. A system that records temporally ordered writing activity can make some parts of revision inspectable that a clean final document cannot reveal: a return to a passage, a deletion and replacement, a pause, or a sequence of local changes.
That visibility is a capability and a burden. It can support process research, teaching, or a writer’s own reflection. It can also produce more material to interpret than a reader can reasonably use. A recorded sequence may show that a phrase changed several times without explaining why the writer changed it, whether the change was prompted by an editor, or whether the final wording is more responsible. Process evidence is not motive, and capture is not understanding.
The same distinction applies outside research. A revision history, comparison view, or marked proof can help a team see that something changed. It cannot settle whether the change is substantively right, rhetorically apt, or adequately supported. That requires a different kind of work: reading, judgment, and often conversation. The Revision Friction Ledger is designed to keep the difference between visible trace and adequate decision in view.
6. The Revision Friction Ledger
The Revision Friction Ledger is a comparative decision sheet. It is not a productivity scorecard, a claim that every writer should preserve every trace, or a substitute for editorial judgment. It asks a reader to name the conditions around a consequential revision before calling a workflow efficient, resistant, transparent, or accessible.
| Ledger field | The reader records | Why it cannot be skipped |
|---|---|---|
| 1. Revision unit | Word, sentence, paragraph, structure, layout, or proof element. | A local substitution and a reorganization of an argument are different tasks. |
| 2. Revision purpose | Meaning, organization, voice, consistency, accuracy, or quality control. | The purpose identifies what kind of judgment must be made. |
| 3. Required action | Mark, erase, retype, select, move, duplicate, accept, reject, compare, or query. | The action makes material and interface labor visible. |
| 4. Interaction/material cost | Physical handling, typing, navigation, selection, switching, waiting, or setup. | A change can be quick to enter but difficult to inspect or communicate. |
| 5. Comparison path | Memory, marked page, parallel copies, difference view, master copy, or review record. | Revision requires an answer to: compared with what? |
| 6. Reversibility | Difficult to undo, locally undoable, versioned, or recoverable only through a retained copy. | Reversibility depends on access to an earlier state, not merely a feature label. |
| 7. Trace produced | Margin mark, clean copy, duplicate file, history entry, comment, proof query, or no durable trace. | A trace can support learning or accountability while also creating exposure. |
| 8. Audience visibility | Private, writer-editor, production team, classroom, or public. | Visibility changes the social cost and explanatory burden of a revision. |
| 9. New expectation | Faster response, cleaner copy, greater justification, more alternatives, or more review work. | A tool can change what others expect a writer to do next. |
| 10. Compensating practice | Checkpoint copy, style sheet, reflection memo, quiet review, accessibility accommodation, or explicit pause. | A practice can mitigate a friction introduced by the environment. |
To use the ledger, begin with one revision and write a condition statement, not a verdict. For the constructed change from eliminated to reduced, the paper typescript may produce this statement: “The insertion remains locally visible against the old claim, but carrying the altered sentence into a clean reading copy requires a further recopying and comparison step.” A basic word-processor file may produce a different statement: “The substitution is easy to enter and the sentence is clean to read, but the prior wording is recoverable only if a copy or comparison practice has been retained.” A review-visible layer may yield another: “The old and new wording are inspectable to the designated audience, but the editor must manage the explanation, response, and eventual disposition of the marked change.”
The statements do not rank the environments. They identify the work that a simplistic account leaves invisible. They can also suggest a compensating practice. A writer working in a file may preserve a checkpoint before a large rewrite. An editor conducting a substantive pass may distinguish comments from final copy changes. A production team may compare a proof against an approved master rather than relying on memory. A teacher may give a writer time to identify a revision goal before expecting a response to a large set of comments.
The ledger also prevents a common error: treating a trace as an uncomplicated good. A visible history can make a revision inspectable, but someone must read it. A dense chain of alternatives can create a review burden. A permanent record can support accountability while revealing an unfinished process, outside help, or an accessible-workaround practice that the writer did not expect to expose. A ledger that includes both trace and audience makes those trade-offs discussable.
7. The counterweight: smoothness and resistance have no automatic virtue
There is a temptation, once revision friction is named, to choose a side. One story says the best environment removes every interruption between feedback and revised text. Another says that slower, more material, or more effortful work must produce deeper thought. The evidence supports neither shortcut.
Heather Lindenman and colleagues studied 152 revision-and-reflection assignments in a first-year writing program. Using document comparison, they classified 28 percent of the sample as substantive revision, 44 percent as moderate revision, and 28 percent as editorial revision.[5] Their more important result is qualitative: students’ reflective accounts and their actual revision practices connected in complex ways. Reflection did not invariably produce substantial or effective reimagining of a text.[5]
The lesson is not that reflection is useless. It is that an added pause has to be understood in its setting. A reflective memo, a comparison view, a printed proof, or a deliberately slower interface may help a writer notice a problem, articulate a goal, or separate a structural concern from a sentence-level correction. But none of those practices guarantees that a writer will decide well or revise deeply. Instruction, task design, experience, time, and feedback conditions matter.
Zhang and colleagues offer a narrow contemporary design case. Their 2025 CHI paper describes Friction, an interface that asks novice writers to organize feedback, identify issues, formulate actions, revise, and evaluate rather than simply accept an immediate replacement.[6] The reported 16-participant study concerns that particular system and baseline; it is not a general proof that AI-mediated delay is better, nor a recommendation to make writing harder. The relevant teaching point is smaller: some revision designs deliberately preserve a reflective interval because fast access to alternatives can bypass diagnosis and planning.[6]
The Revision Friction Ledger therefore treats a pause as a condition to be examined, not a virtue to be installed. What is the pause for? Who benefits from it? What task does it support? Is it accessible? Does it give the writer greater judgment, or only another procedural hurdle? These questions protect the paper from both technological boosterism and the romanticization of resistance.
8. Access labor belongs inside the ledger
A revision environment is never neutral simply because it is digital or because it offers assistance. Emily Q. Wang and Anne Marie Piper studied the academic writing practices of 11 dyslexic adults in the United States. Participants described limits in editing tools for academic language, workarounds that repurposed other tools, ad-hoc collaborations that compensated for gaps, and the burden of ableist expectations about written work.[7] The authors characterize the added time, effort, and social and material resources required to participate as invisible access labor.[7]
The study does not represent every disabled writer, every language condition, or every writing context. Its importance for this paper is disciplinary. It prevents an account of revision friction from treating the action available in an interface as the whole experience of revision. A spell-checking or editing feature may be present but inadequate for a particular form of academic language. A visible revision trail may support a reviewer while increasing pressure on a writer to justify an assistance practice. A request for another round of polish may appear routine to an editor while requiring an unrecognized network of time, tools, and peer support from a writer.
For that reason, the ledger ends with compensating practice rather than only a list of obstacles. The purpose is not to demand that writers overcome an inaccessible workflow. It is to ask what change in process, support, timing, interface, or expectation would make the work more viable. Examples may include allowing an alternate review route, setting a clear revision scope, preserving a private drafting period, providing an accessible comparison method, crediting collaborative assistance appropriately, or refusing the assumption that greater visible polish is a neutral measure of effort.
9. The expectation test
Before calling a revision environment an improvement, run five questions:
- Which work became easier, and which work merely moved? A substitution may be faster while comparison, storage, explanation, or recovery becomes a new responsibility.
- What comparison must occur for the revision to be trusted? The answer might be a marked page, a clean duplicate, a difference view, an approved master, or a reader’s careful re-reading.
- What trace was created, removed, or made newly visible? A trace can enable accountability and learning, but it can also generate exposure and review burden.
- Whose access needs and support labor are assumed by the workflow? An environment that appears smooth for one participant may require hidden work from another.
- What new expectation of speed, polish, response, or justification has the environment created? The social consequence of lowered effort can be a higher demand for further revision.
Return, finally, to the change from eliminated to reduced. The responsible conclusion is not that the insertion belongs on paper, in a file, or in a review layer. It is that the insertion has a working environment. Each environment creates a different relationship among action, trace, comparison, audience, and expectation. To notice that relationship is to see revision not as a frictionless correction of text but as a form of editorial work with material, social, and interpretive consequences.
References
[1] University of New Mexico Libraries, Center for Southwest Research. “Manuscripts for Zia Summer.” Rudolfo Anaya Digital Archive. https://anaya.unm.edu/manuscripts-18
[2] Kruse, Otto, and Christian Rapp. “Word Processing Software: The Rise of MS Word.” In Digital Writing Technologies in Higher Education. Springer, 2023. https://link.springer.com/chapter/10.1007/978-3-031-36033-6_2
[3] Conijn, Rianne, Emily Dux Speltz, Menno van Zaanen, Luuk Van Waes, and Evgeny Chukharev-Hudilainen. “A Process-Oriented Dataset of Revisions during Writing.” In Proceedings of the Twelfth Language Resources and Evaluation Conference, 363–368. European Language Resources Association, 2020. https://aclanthology.org/2020.lrec-1.45/
[4] Australian Government Style Manual. “Editing and Proofreading.” Updated 3 November 2021. https://www.stylemanual.gov.au/writing-and-designing-content/editing-and-proofreading
[5] Lindenman, Heather, Martin Camper, Lindsay Dunne Jacoby, and Jessica Enoch. “Revision and Reflection: A Study of (Dis)Connections between Writing Knowledge and Writing Practice.” College Composition and Communication 69, no. 4 (2018): 581–611. https://files.blogs.baruch.cuny.edu/wp-content/blogs.dir/9476/files/2023/09/Lindenman-et-al.-2018.pdf
[6] Zhang, Chao, Kexin Ju, Peter Bidoshi, Yu-Chun Grace Yen, and Jeffrey M. Rzeszotarski. “Friction: Deciphering Writing Feedback into Writing Revisions through LLM-Assisted Reflection.” In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, art. 935. ACM, 2025. https://doi.org/10.1145/3706598.3714316
[7] Wang, Emily Q., and Anne Marie Piper. “The Invisible Labor of Access in Academic Writing Practices: A Case Analysis with Dyslexic Adults.” Proceedings of the ACM on Human-Computer Interaction 6, CSCW1, art. 120 (2022). https://doi.org/10.1145/3512967