Chapter 11. On Keywords

We move into extraction, the first act of the work cycle. Before the technique of extraction can be addressed, what gets extracted must first be settled. The answer is the keyword. The guiding question of this chapter is: why is the world handled in keywords? We lay out why the unit of design is the word rather than the sentence, the item rather than the narrative, then take up, in order, the nature of the keyword as material, its varieties, and how to handle it.

1. The Keyword: The Minimal Unit of Design

A keyword is both the name that calls up each single item making up a world, and the minimal unit of design work. The white-copper coin, jianghu, the concealment mechanism, curfew. What a worldview document is, in substance, is a list of such keywords together with the explanation attached to each. Why is the unit the keyword rather than the sentence or the story? The reason lies in the demands of management. Recall the five conditions of the deliverable spec and the answer becomes visible. To be searchable, an item needs a heading; to be citable, an item needs to be cut cleanly; to be trackable, an item needs a unit of revision. The keyword is the only size that satisfies all three at once. The sentence is too fine a unit for search, and the story is too large a unit for revision — the same reason an encyclopedia is organized by headword rather than by sentence. The keyword unit has a second benefit: reuse. A setting dissolved into a story can only move together with that story, but a setting cut into a keyword can be dropped independently into a different story. For a worldview to become the shared foundation across works, the grain of that foundation has to be smaller than any single work. The analogy of assembly is exact: a finished product cannot survive being disassembled, but a part exists for the purpose of assembly. The design implication is a shift in the unit of thought. The designer is someone who imagines the world as story but records it as keywords, and this double sense is the foundation of the job description. Someone who only imagines stays a writer; someone who only records stays a clerk.

2. Reification and Objectification: The Technique for Handling the Formless as Object

There is one precondition to recording by the keyword unit: what makes up a world has to be something that can be cut out. A sword and a castle are cut out to begin with, but half of any world has no shape at all — mood, feeling, relationship, tension, the air of a given era. The technique for handling the formless is reification and objectification. Objectification is the operation of naming something formless and thereby turning it into a single object. The old hatred between two houses cannot be grasped, but the moment it is given the name the Northern Blood Feud, it becomes a registrable item. Once a thing has a name, an explanation can be attached to it, and once an explanation is attached, search, citation, and revision all become possible. Just as programming bundles data and behavior into an object, design bundles feeling and history into a keyword. Reification is objectification intensified, the operation of giving physical substance to something formless. Load the blood feud onto an object, the Dagger of the Vow, and the hatred now becomes something that can appear in a scene, be stolen, be broken. The formless acquires the fate of a thing. The design of a symbolic prop is exactly this technique, the generalization of the principle by which, in the world of home, history is carried by objects. The design implication is inventorying intangible assets. The more a thing sits at the emotional core of a world, the more likely it is to be formless, so unless it is deliberately objectified, it drops out of the document. Mood itself has to become a keyword before mood can be managed. Once the particular desolation of a given world can be written down as an item, what it is made of, that desolation can be reproduced in a derivative work too.

3. The Tangible and the Intangible: What Can Be Registered

With the technique of objectification in hand, a map of what can be registered becomes possible to draw. Place what makes up a world on a spectrum running from the tangible to the intangible, and settle a way of recording for each. At one end of the spectrum sit tangible things. Objects, places, people, things with shape and location, are recorded in the language of specification: size, material, location, owner. This is the category that fits a datasheet best. In the middle sit institutions and rules. Law, custom, organization, ritual, things with no shape but stated as clauses, are recorded in the language of the clause. The honor code and the infection-rules sheet belonged here. At the other end sits pure intangibility. Feeling, mood, the texture of a relationship, things difficult even to state, are recorded, after passing through objectification, in the language of definition and example: a sentence defining what the thing is, attached to a list of the scenes where it shows itself. The sheet of attitude that magical realism required was the specimen of this category. The point of this map is that moving toward the intangible end changes not whether a thing can be recorded, but how it is recorded. What goes unrecorded is not transmitted, and what is not transmitted disappears along with whoever was carrying it. The first things to evaporate from a world are always the most intangible ones. The design implication is a trinity of recording formats. Keep three formats ready in the worldview document system — specification, clause, and definition-and-example — and whatever component of the world arrives, there is a vessel ready to take it down. An organization with its vessels prepared accumulates at a different speed than one that has to puzzle out a format every time.

4. Cliché: Not Staleness but Asset

Now we move on to the varieties of material. What gets fished up most often at the site of extraction is, surprisingly, not the new but the old, the cliché. Cliché names an expression or setting worn down by long repetition, and craft theory usually teaches it as something to avoid. On the designer's balance sheet it sits in a different account. A cliché is not staleness but a verified, agreed-upon asset. That it has been repeated means demand has been proven; that it is worn means every reader already knows it. The value of already knowing was confirmed throughout Typology. The standard language of medieval fantasy erased the cost of explanation, and a single type name contracted away half the premise. The minimal unit of that economy is the cliché. The scheming chancellor, the hero from the countryside, the companion who betrays at the end, each is a ready-made part that can be deployed at zero cost of explanation. The problem is not the cliché itself but the unselfconscious use of it. Use it without knowing it is a part, and the work looks like an assembly of parts; use it knowing it is a part, and it becomes an object of arrangement and variation. The precise translation of the maxim avoid clichés is do not write without knowing your clichés. The technique of twisting just the vital point of the standard operates just the same at the level of the cliché. The design implication is inventory management for clichés. Holding a documented list of one's genre's clichés, with a verdict of use, vary, or avoid attached to each item, is the designer's stocktaking. Inventory with a verdict attached is an asset; inventory without one is just dust in the warehouse.

5. Trope: The Typified Device

Give a cliché a neutral name and it becomes a trope. Trope is the term for a narrative device typified through repeated use, a word that strips away the contempt cliché carries. The English-language TV Tropes wiki is the monument to this view. Hundreds of thousands of devices have been given names and catalogued, each carrying a definition, examples, and a genealogy of variations. It is proof that the entire material of creation can be organized into a parts database, and it also stands as a civilian precedent for the document system a designer should aim at. The practical value of the concept of trope lies in resolution. If the cliché is the lump, the trope makes it possible to break that lump down into its component parts. The cliché of the hero from the countryside decomposes into the conventional bundle of tropes: low birth, hidden lineage, the death of a mentor, the destruction of home. Once broken down, recombination becomes possible. Pull just the lineage trope out of the bundle and keep the rest, and out comes a hero who is familiar yet somehow different. Much of what a reader feels as freshness is not invention but exactly this kind of partial substitution. The principle of combination confirmed at the summary of Typology repeats itself here at the level of the device. The design implication is training the resolution of extraction. When breaking a work down, forming the habit of not stopping at the level of the cliché but splitting further to the level of the trope yields several times the reusable parts from the same source. Simply skimming the entry names on a tropes wiki already expands the vocabulary of decomposition, and the larger the vocabulary, the more parts become visible in the same work.

6. Metaphor: The Meaning a Keyword Carries

A part does not carry function alone. A keyword is also a vessel that carries meaning, and the principle of that loading is metaphor. Metaphor is the working of meaning by which one thing points to another. The keyword apple is a fruit, and at the same time it carries the forbidden fruit of knowledge, the loss of paradise, the discovery of universal gravitation, and the emblem of a particular company. The worth of a keyword as worldview material is proportional to this loading capacity. A keyword carrying no meaning at all remains a functional part, but a keyword loaded with metaphor carries theme just by appearing on stage. The earlier observation that a vampire's rules list functioned as terminals of metaphor was exactly this principle. Predation, desire, and exploitation are already loaded into the keyword bloodsucking, so the writer only has to run current through the terminal. Use it without knowing what has been loaded onto it, and an unintended current runs instead. A symbol grants no immunity to ignorance. Loading can also be carried out by design: choose a world's core keywords, decide the meaning to load onto them, and prepare in advance the series of scenes where that meaning will surface. At the peak of this method sits the precedent that loaded a single ring with the temptation of power and made it carry an entire world's theme. The design implication is managing a symbol register. Manage the keywords in a world that carry loaded meaning as a separate list, and specify, for each, the meaning it carries and the misuse to be treated as forbidden. Overuse and confusion of symbols happen in worlds that keep no such register. A world in which everything is a symbol becomes identical to a world in which nothing is.

7. Traditional Literature: The Archetype Repository of Keywords and Clichés

The richest vein of parts and meaning is traditional literature. It names the entire body of narrative asset that has survived through oral transmission: myth, legend, folktale, and nursery rhyme. Looking again, through the eyes of methodology, at this stratum that Typology treated as material, traditional literature turns out to be the oldest keyword database humanity has ever operated. Centuries of oral transmission were centuries of an editing process. Details that go unremembered drop away and only the strong skeleton survives; this selection process served as the refining process for tropes and metaphors. The dragon, the witch, the abandoned child, the three wishes, the broken taboo. The keywords of tradition are, one and all, high-purity parts that have passed through this process, and they do not even carry copyright. This is the first drawer a designer suffering from a shortage of fresh material should open. From the standpoint of extraction practice, traditional literature has a further value: it is a specimen of combination. Studies classifying folktale types have shown that stories from every corner of the world are combinations of a finite set of motifs, and this is old proof of the design proposition that an infinite number of worlds can be assembled from a finite number of parts. The design implication is dual extraction. When mining a keyword out of traditional literature, both the part itself and the way the part combines with others must be mined. Bringing over just the keyword dokkaebi, and bringing over the relational structure in which a dokkaebi strikes a wager with a human being, are extractions of different depth. The latter is what was earlier called transplanting at the level of condition, and its value on reuse multiplies several times over. The habit of mining down to relational structure can be learned from the morphological analysis done in folktale research.

8. Keyword Extraction: The Method of the Work

Now that the material is known, the action itself gets laid out. Keyword extraction is the work of combing a source and listing keyword candidates, and the procedure can be standardized into four stages. First, fixing the source. Decide what will be combed. From the primary text of the source work, to gathered materials, meeting minutes, competing works, and traditional literature, the list of sources is itself something to be managed. Second, a full read-through with marking. Read the source from the beginning, leaving a mark on any fragment that could become a keyword. The discipline of this stage is deferring judgment. Read while weighing usefulness, and the mesh of the net narrows, letting unexpected material slip through. Fish broadly first, and put off judgment for later. Third, listing and a rough first sort. Gather the marked fragments in one place, give each a heading, and sort them, in rough form, according to the map of the eight domains. Source notation gets attached at the same time. A fragment with no record of where it was fished from is one nobody trusts later. A keyword without a source gets re-investigated in full at the review stage, so the few seconds spent noting the source buy back hours of re-investigation. Fourth, candidate judgment. Go through the list and split it into registration candidates, holds, and discards. The standard of judgment is demand. Whatever touches the spec list or a priority excavation point becomes a candidate. The design implication is the recurring nature of extraction. Extraction is not a one-time event at kickoff but a standing process that runs every time the source is updated, and for a world still in serialization especially, every new installment is a new source. A routine that extracts each installment the moment it is published is the practical form that stopping the bleeding takes. Extraction left undone piles up like dishes left undone: the longer it sits, the less anyone wants to touch it.

9. Why Extraction Matters: The Gate of Design

Once the procedure is known, it is time to ask about this action's standing. Why does extraction sit first in the cycle? The answer is that extraction is the gate between existence and asset. An unextracted setting exists, but it is not an asset. A setting written somewhere in the source work, an idea sitting in a creator's head, a direction agreed on in a meeting, all of these exist. But none of them can be searched, cited, or checked against other items. A setting buried inside a sentence lives only as long as whoever remembers that sentence. The organization's world ends up mortgaged to an individual's memory. What the action of extraction does is transport these buried existences into a manageable state. From the moment a setting is put on the list, it gets checked against others, conflicts are found, and reuse gets considered. The pathologies seen earlier can be rediagnosed from this angle. The on-the-spot decision that bred paradox was a setting that went straight into the body of the work without ever passing through extraction; the divergence of derivative versions was differences that could not be checked against each other because they were never extracted. A world that vanishes along with whoever was responsible for it is a world where extraction never once took place. In short, most failures of worldview management have already begun before registration, in the absence of extraction. The design implication is a felt sense called the extraction rate: staying aware, even by rough estimate, of what proportion of existing settings has made it onto the list. In a world with a low extraction rate, all the refinement of definition and registration is a castle built on sand. However rigorous the gate, if no setting arrives in front of it, the world does not grow.

10. The Limits of the Keyword: What Extraction Cannot Hold

The last matter of this chapter is the limit of this unit. The keyword is the minimal unit of design, not the whole of the world. There are three things a list cannot hold. First, arrangement. The same set of keywords becomes a different world when arranged differently. The parts dragon, knight, and princess can assemble a rescue story, a tragedy told from the dragon's point of view, or a three-way political drama. The parts list does not include the assembly diagram. This is where the freedom to build different worlds from the same material comes from, and it is also where the trap of resembling other worlds because the material is the same comes from. Second, relation. The connection between one keyword and another, the relation a blood feud holds with a dagger, the relation jianghu holds with the imperial court, does not reduce to the explanation of any single item. The relation itself is a separate object of record, and this is the problem of interconnection taken up in the next chapter. Third, texture. A list makes a world manageable, but what a reader loves is not the list; it is the air of the world that the list, once assembled, gives off. Objectification turns the formless into an item, but the sum of items does not automatically restore that air. These limits are not defects; they are lines of division of labor. The keyword takes charge of material; arrangement and relation are the charge of definition and narrative; texture is the charge of creation. The moment someone believes the material unit is everything, design falls into a fetishism of the list; the moment someone skips past the material unit, management collapses. A tool that knows its own limit is a good tool. So the next question is self-evident: definition, the second act, the one that fixes meaning and name onto the candidates that made it onto the list.