{"path":"research/corpus-graph/what-a-move-is.md","content":"# What a \"move\" is, why it felt fuzzy, and whether there are kinds\n\n**Date**: 2026-10-11 · **Type**: a reflection on a word the corpus-graph work leans on, with one count\nof our own and two outside sources · **Status**: Claude's reading. Nothing here is ruled.\n\n**His words**, 2026-10-11, on the definition *what a passage does, said without its subject*: *\"Ok so\nlike the conceptual essence of a passage, somehow? Interesting, kinda psychedelic/fuzzy to me, but\nsure\"*, and then: *\"Important to reflect on what a \"move\" really is, perhaps there are different kinds\nof move etc\"*.\n\nThe word is used in the card test and in the judges' question ([README.md](README.md) § 18,\n[likeness-seen-from-outside.md](likeness-seen-from-outside.md)). This note says what it means there,\nwhy his sense that it is fuzzy is right, and what follows.\n\n---\n\n## 1. Plainly, with two examples\n\n**The textbook pair** (Gick and Holyoak 1983, their Table 1). A general cannot send his whole army\ndown one road, so he sends small groups down many roads at once and they meet at the fortress. A\ndoctor cannot send a strong ray through healthy tissue, so she sends weak rays from many directions\nand they meet at the tumour. Take away armies and rays and one thing is left:\n\n> wanting to bring a large force onto one point · unable to send it along one path · so sending\n> small forces along many paths at once · and the point is overcome.\n\nThat remainder is what we have been calling the move. The authors call it a *schema*, and give it\nfour parts: **what is wanted, what stands in the way, what is done about it, and what comes of it.**\n\n**A pair from our own notes**, which a reader tied and a judge called the same move for the same\nreason. One note stamps a fetched source with the time it was fetched, and warns that a copy\nfetched later must not pass as the original. Another, about how a memory system keeps facts, gives\nevery fact two times: when it was so in the world, and when the system learned it. Take away\nsources and facts: *wanting to say later what was known when · a later copy looks the same as an\nearly one · so every record carries two clocks and the first capture is never overwritten.* A\ncard-writer gave it the names *bitemporal data*, *chain of custody* and *write-once storage*.\n\n## 2. The word does three jobs, and we meant the second\n\n| Sense | What it is | Where |\n|---|---|---|\n| a move **in a text** | what a passage does for the piece it stands in: stakes a claim, concedes, corrects, justifies a method | the study of genres: *\"a discoursal or rhetorical unit that performs a coherent communicative function\"* (Swales 2004) |\n| a move **in the subject** | a way of getting something: the four parts above | Gick and Holyoak's schema; the *purpose and mechanism* of the analogy-mining work the corpus already read |\n| a move **in the map** | something a person can do next: answer a question, split a claim, object | this project's own use: *no gap displayed without a move offered*, the record of *moves tried* |\n\nThe corpus had the third sense first, so borrowing the word for the second was a collision. The\nproject's rule for a word that names a range is to name the layers and keep the word. In chat and in\nthe next briefs the second sense gets a plain name of its own, **a pattern**, with its four parts\nspelled out.\n\n## 3. Why it felt fuzzy: one passage does not have one pattern\n\nA passage can be restated without its subject at many levels. The stamped source is *a record with\ntwo clocks*, and also *keeping a history*, and also *not letting a copy pass as an original*.\nNothing in the passage alone says which level is the right one.\n\nGick and Holyoak say why. A schema is *\"defined by the correspondences between two analogs\"*: it is\nwhat is left when the differences between **two** cases are deleted. In their experiments every\ndevice for getting a schema out of **one** story failed (asking for a summary, stating the principle\nin words, drawing it). Given two stories, people produced the schema as a by-product of saying how\nthe two were alike, and how good that schema was predicted whether they solved the new problem.\n\nOur own counts say the same from the other side. Two cards written apart, each from one passage of\na pair, share a name for the move in 23 pairs of 67, and for finding the partner the card is a\nsecond net and no better net than the passage's own words ([README.md](README.md) § 18.6). And the\njudges' form asks for each passage's move in a sentence *before* the two are compared, which is\nthe single-story method.\n\nSo the pattern belongs to the pair. A card from one passage is a guess at which of its many\npatterns somebody else will also have written down.\n\n## 4. Are there kinds?\n\n**By what is done: a rough count that does not hold up.** Sorting the 134 cards by the first verb\nof their move gives four families: *arranging* something so that a thing happens by itself (51),\n*testing or measuring* (37), *noticing* a regularity (25), *cutting* one thing into two (17), and\n4 that fit none. But for ten pairs that both judges called the same move, the two cards fell in\ndifferent families in eight. The same pattern is written as *counts* by one writer and *keeps* by\nanother. Verb families are how a writer phrases a pattern. They are no typing of patterns.\n\n**By which parts are shared: a typing that follows from the four parts.** Proposed, untested:\n\n| The two passages share | The likeness is | Example of what one can take from the other |\n|---|---|---|\n| all four parts | the same pattern for the same reason | the other's condition for when it works |\n| what is done, with a different aim | the same means to another end | a use nobody had thought of |\n| the aim and the obstacle, with different things done | two answers to one question | the rival answer, and a reason to compare them |\n| the obstacle only | the same difficulty, noticed twice | that it is general |\n| one passage's *what is done* is the other's *what is wanted* | a practice and its ground | why the practice is right, or a test for it |\n| what is done, with opposite verdicts on it | one mechanism judged two ways | a disagreement to settle |\n\nThe right-hand column is what the judges were never asked. And a pair of one of the last four kinds\nis refused by a judge asked only *is it the same move*. Sixteen of the thirty-four real pairs were\nrefused, and read again by the session about ten of them hold a likeness of some such kind (an upper\nfigure: [likeness-seen-from-outside.md](likeness-seen-from-outside.md) § 4).\n\nA passage that only *notices* something has no *what is done*. It has a regularity and a condition.\nThat is one reason a noticing and a practice are never the same move, and are often the most useful\npair: one is the problem and the other an answer to it.\n\n## 5. What this changes (proposed)\n\n1. **The name.** *Pattern*, with its four parts, in chat and in the next briefs. *Move* stays for\n   what a person can do in the map.\n2. **The judge's form.** Fill in the four parts for each passage, say part by part whether the two\n   agree, and let the kind of likeness follow from the table. The yes or no on *the same move*\n   goes.\n3. **Where the pattern is written.** From the pair, by saying what the two share and where they\n   differ. A card from one passage stays what it was measured to be, a second rough net.\n4. **A test before any of it is relied on**, not run: the thirty-four real pairs and their decoys,\n   with the new form. *Predicted*: no decoy has all four parts shared; the eighteen pairs judged the\n   same move come out in the first two rows; and at least ten of the sixteen refused pairs land in\n   one of the last four rows, with something written in the right-hand column. *Against the typing*:\n   most refused pairs sharing no part at all, or decoys landing in the rows as often as real pairs.\n\n## Read depth\n\nGick and Holyoak (1983), *Schema induction and analogical transfer*, Cognitive Psychology 15: the\nabstract, and the pages holding Table 1 and the account of mapping and schema, as a search returned\nthem; not the experiments in full. Swales's definition: as quoted from *Research Genres* (2004),\np. 228, in two secondary sources; the book was not opened. The analogy-mining work and Gentner's\nfindings: from the corpus's earlier report,\n[what-research-says-about-finding-far-analogies.md](what-research-says-about-finding-far-analogies.md).\n\n## See also\n\n[likeness-seen-from-outside.md](likeness-seen-from-outside.md) (the kinds of likeness read out of\nthe sixteen refusals, which § 4 here derives from the four parts) ·\n[reading-for-every-kind-of-insight.md](reading-for-every-kind-of-insight.md) § 6.2 part 8 (the\njudge's forms) · [../the-analogy-daemon.md](../the-analogy-daemon.md) § 8 ·\n[../what-a-rule-rests-on.md](../what-a-rule-rests-on.md) (where a rule comes from: compared cases).\n"}