# Extend Judgment Without Surrendering It

The Inkagai Way: 100 Practices for a Writing Life

By Anna Lishman & Brad Lishman

Practice 99 · Part 10: Completion and the writing life

Canonical URL: https://inkagai.com/the-inkagai-way/read/extend-judgment-without-surrendering-it

> Use tools, editors, and collaborators to extend judgment, not replace it.

Whatever a helper hands over is a proposal, whether it is a research assistant’s summary, an editor’s cut, a software suggestion or a collaborator’s scene, and that holds for any helper, human or not. A helper can supply what the writer lacks, a reader without the memory of composing, a source the writer hasn’t read, speed, or a cut the writer couldn’t bring themselves to make. The Work’s purpose and the promise it made to its readers stay with the writer, and they are the standard any suggestion has to meet. That reasoning is ours. A writer who hands over the decision hands the standard to someone who doesn’t hold it.

In the autumn of 1921, on three months’ sick leave from his job at a bank, T. S. Eliot worked on *The Waste Land* at Margate and then at Lausanne. The draft he showed Ezra Pound came back marked up, with large parts cut. The poem appeared in 1922 under Eliot’s name; he wrote the dedication to Pound, *il miglior fabbro*, the better craftsman, by hand in Pound’s copy, and it was first printed in 1925. In 1946 he wrote that he would wish the blue penciling to be preserved as “irrefutable evidence of Pound’s critical genius.” Asked in 1959 whether the cuts had changed the poem’s intellectual structure, Eliot said no. He thought the longer version had been just as structureless, only in a more futile way. Pound saw what Eliot had missed, and Eliot decided what stood and credited the help.

Raymond Carver’s case shows how murky the line can be. After Gordon Lish’s heavy edits to *What We Talk About When We Talk About Love*, Carver wrote a letter asking Lish to stop production and, two days later, another that was thrilled yet uneasy (Practice 70 tells it). Critics still disagree about whether his judgment was extended or replaced, and the record can’t settle that for anyone outside.

Software and AI make the question a daily one. In *Mata v. Avianca*, decided in the Southern District of New York on June 22, 2023, lawyers filed a brief citing court decisions that a general-purpose chatbot had produced and that did not exist. When the court asked for copies, they supplied documents that purported to be the decisions. The court’s own findings add that one lawyer had asked the chatbot whether the cases were real, and that it had said yes. The judge found subjective bad faith and fined the lawyers and their firm $5,000, jointly. The output entered a document the signers were accountable for without anyone checking it against a source, and the lawyers stood by it when challenged. Eliot tested Pound’s cuts against the standard he carried, and that is the step the filing skipped. Claims, quotations, citations and biographical details produced by AI are unverified until checked against an original (Practice 77).

None of this makes help suspect. One writer may use none and write well, and another may use a great deal and write well; the question is where the deciding happens. Chapter 6 set out the inspection questions for any tool, what it receives, retains and changes. This practice is about the allocation of decisions, which kinds of choice a writer is willing to have assisted and on what condition the help will be accepted.

Declining help is a legitimate choice, and a writer who declines for a stated reason, such as a voice kept unmixed or a wish to answer for every sentence, has made a decision. Accepting help carries the mirror risk, since a suggestion taken because it came from someone senior, or from a tool that sounds sure, hasn’t been judged. Some editors may see a book more clearly than its author can, though, and when a writer can’t evaluate a suggestion, deference with a stated reason is also a decision.

**Practice.** Draw a decision map for your current Work. List ten decisions, from large (what the book is for, whether a character lives) to small (a title, a paragraph’s wording, a citation, a publication date). Sort each into one of three columns: open to assistance from any source; assisted, then verified or accepted by me, with how I will check; mine alone. For each item in the third column write one sentence of reason. Keep the map beside the manuscript, and notice the first time you break it. The map takes forty-five minutes. Eliot’s second column would have read: cuts from Pound, accepted by me, credited in the dedication. Write yours in that form.

---

Copyright © Anna Lishman and Brad Lishman. Published by Inkagai, an Algodyne Corporation product.

This digital book is free to read and download for personal use. You may share the canonical public link and quote brief passages with attribution. You may not resell the book, remove attribution, or distribute modified copies as official editions.

To quote: use up to 100 words at a time, and no more than 300 words in total from this book in any one article, post, or work. Credit both authors by name, as on the title page: Anna Lishman and Brad Lishman, *The Inkagai Way: 100 Practices for a Writing Life*, `https://inkagai.com/the-inkagai-way`. Do not credit one author alone, and do not edit a quotation without marking the change. Longer excerpts, translations, audio editions, and adaptations need written permission through `https://inkagai.com/contact`.

Writers quoted or discussed in this book keep every right in their own work. Their words appear here as brief quotation for commentary and teaching; they are not licensed for reuse by this notice. Quote them from their own books and cite those. No writer named here is affiliated with, or endorses, Inkagai or this book.

Canonical home: `https://inkagai.com/the-inkagai-way`