Orienting to the cost-bearer, auditing toward real fixes.
Every sentence about harm has a cost-bearer. The fix-hiding register removes them. This framework puts them back.
You have read sentences like mistakes were made, tensions escalated, reforms are underway, civilians were affected. The person who wrote the sentence knew who paid. The sentence was built to hide it. That hiding is an operation. It has a name. It has diagnostics. It has a counter-move.
Kita is a pre-ethical grammar. The rules a sentence about harm follows to stay auditable: cost-bearer visible, agent named, time specified, real fix attached. Each missing element is a trace of a specific operation that was performed on you before you read the sentence.
The framework gives those operations names and teaches you to counter them. Name the cost-bearer. Name the fix. Hold the fix. Audit yourself the same way you audit everyone else.
Most political arguments fail not because the argument is wrong but because the cost-bearer never entered the room.
Two people argue for an hour about Haiti. Neither of them mentions a Haitian. Two people debate immigration policy. No undocumented person appears. Two people discuss sanctions. The 500,000 Iraqi children who died under the 1990s sanctions regime are not mentioned by name or number. The arguments can be technically correct on both sides and the cost-bearer never surfaces. This is the fog winning regardless of who won the points.
Kita's primary test. Can you get through a conversation about an arrangement without the people paying for it entering the room? If yes, the fog won. That is the whole test. Not "did I make the better argument." Not "did I correct the errors." Not "did they concede." The test is whether the cost-bearer was in the room.
The framework's operations are all in service of this test. The three questions (who did this, when, who pays). The body principle (put the bodies back in the sentences). The named fix (a specific action by a specific party by a specific time). The anchored fix (hold the fix through deflection). Everything else follows.
Naming what's wrong is step one. Naming what would fix it is the step that fog-operators can't survive.
A real fix (実済) has four properties. A specific party. A specific action. A specific time. A specific consequence for inaction. Once named, the beneficiary of the current arrangement has two options: publicly defend the arrangement, or be visibly seen refusing to. Both are progress. Fog only survives in the absence of this naming.
An anchored fix (錨済) is what keeps the fix in the room after it's named. Someone will deflect. The anchor move is one sentence acknowledging the deflection, one sentence returning to the fix. The anchor doesn't move. The current does. Most arguments are lost because the fix wanders off while its author is handling each objection one by one.
Empty fixes (虚済) are the opposite — the appearance of a fix with none of the four properties. "We need to have a conversation about healthcare" is an empty fix. "Congress restores the expanded Child Tax Credit by the end of 2026, with automatic escalators for cost-of-living, or the members who block it are named and opposed" is a real fix. The first lets you feel engaged while committing to nothing. The second puts weight on people.
Empty fixes are worse than no fix because they inoculate the audience against recognizing the absence.
Three questions. Who did this. When. Who pays. Run on any sentence about harm. Any failure is a trace of what was removed.
Three gates. Is anything moving between people. Does the description match. Is an expectation attached. Most fog fails at gate two or three.
Body principle. The fix-hiding register's primary operation is removing bodies from sentences. The framework's primary counter-operation is putting them back.
Cost-bearer anchor. Every conversation drifts away from the cost-bearer because keeping them visible makes people uncomfortable. The anchor is the practice of keeping them in the room anyway, through repetition if necessary, through re-specification when abstraction tries to take them out.
False debt (偽債). A debt installed rather than earned. Exit test: can the obligated party leave on terms proportional to what was actually exchanged. If not, the excess punishment is the shape of the false debt. From interpersonal to state scale.
Container audit (容器審計). Open words the way a mechanic opens an engine. Compare what the word claims to what it contains. Large containers (security, freedom, stability, regime, targets, resilience) carry the biggest mismatches.
Register swap (語域互換). Apply the diplomatic register to a crime, or the crime-report register to a diplomatic event. The pair makes the register permanently visible. Once visible, it cannot be made invisible again.
Realmanning (實論). Name the argument someone's conduct is running on, distinct from the argument they state. Often differs sharply from the stated version.
Symmetry. The framework applies equally to every party, including the framework user. Selective deployment is itself a fog operation, the most refined one in the catalog.
Twenty-plus other operations in the full document. These do most of the work.
A pre-ethical grammar. A set of rules for restoring the information substrate that ethical traditions need to operate on. The fix-hiding register strips that substrate at industrial scale, producing sentences where cost-bearer, agent, time, and fix have all been removed or softened. Kita's operations put them back. What ethics you then apply to the restored sentences is your call. The framework's opinion is only on whether the cost-bearer entered the room.
Symmetric by design. Every operation applies equally to every party, including the framework user. Audit one side and not the other, and you are doing advocacy. The symmetry requirement is load-bearing. Selective deployment is a refined fog operation — the framework's own tools can be weaponized against the framework's purpose if the user forgets this. Catch yourself applying the audit unevenly. That catch is part of the practice.
A skill distribution. If enough people develop the skill, cumulative effects compound over decades into something that looks from outside like a change. The change, if it happens, is mechanical and slow. Each individual adoption is valuable on its own regardless of whether a cumulative threshold is ever reached.
It borrows from discourse analysis, media criticism, philosophy of language, nonviolent communication, and institutional critique. It makes specific operational choices none of them make: name the operations that remove cost-bearers and fixes, teach the counter-moves, trust the user to apply them across every party.
The framework ships as a family of files at different scales, plus companion material.
kita.txt — main operational framework. Full operations, specimens, failure modes, pre-send checklist. Load as context in anything that gives a model text before generation.
kita-prefs.txt — preferences version for slots that allow ~14,000 characters. Compressed activation layer; assumes kita.txt can be loaded separately as knowledge file or project context.
kita-micro.txt — micro version for tight slots (8,000 characters or smaller). Standalone light framework.
kita-companion.md — human-facing companion. Essays, extended specimens, theoretical discussion. Read as a book.
fog-catalog.md — institutional sentences paired with diagnosis and restoration. Specimen reference.
Contributor and maintainer documentation is in the docs folder.
As a reader. Read kita-companion.md. Try the three questions on the next thing you read. Notice what is missing. Supply it, at least in your own mind. The difference between the original and your supplied version is the operation you just made visible. Do this for a week and the operations become impossible to unsee.
As a writer. Keep kita.txt accessible when you write about power, cost flow, or institutional arrangements. Run the pre-send check on drafts. You will find yourself writing fog you did not know you were writing. This is normal. Everyone does.
As a model user. Paste the appropriate file into preferences, custom instructions, or project context. Use kita-prefs.txt for most platforms. kita-micro.txt for tight preference slots. kita.txt where the platform supports loading knowledge files. Models given Kita produce output with bodies, agents, cost-bearers, and real fixes attached more reliably than baseline. Without the framework loaded, the model defaults back to the fix-hiding register that dominates its training data.
As a translator. Translate the stance, not the words. Every language has a register of trust: Chinese 白話, Arabic direct journalism, Spanish crónica, Japanese 常体, English Orwell / Baldwin / Carson / Didion. Find your language's version. Read its exemplars until their rhythm is in your ear. Write in that rhythm. The Chinese terms (壅, 実済, 錨済, 偽債, 實論, 兩口鍋, 贈明) can travel where they do compression work. Everything else becomes your language's own.
As a contributor. Fork it. Modify it for your situation. Publish your modifications as yours. Read kita-register-guide.md before significant rewrite work — register decisions, term-selection rationale, context exercise worth working through before starting. If you find the framework making claims its operations cannot justify, audit it and publish what you find. A strong critique is a gift. If you build something better, rename it and send your version into the world. The framework does not need to be preserved. The operations do. See CONTRIBUTING.md for contributions back to this repo.
The framework writes in a specific register. Short sentences. Bodies throughout. Agents named. The reader trusted to follow without being managed. No performative gravity. No protective formulas.
This register exists in every language with substantial public discourse. Kita points at it, writes one instance of it, invites readers in. Your language has its own exemplars. The register is what the operations produce when they run continuously on the writer's own output.
You can use the operations in any register. The register and the operations are related but separable. Reading the exemplars your language already has is part of how the framework lands.
Maintained as a public tool. Development has been conversational, iterative, open. Major versions tagged. Changes visible in commits.
The framework will continue to evolve. New operations will be added as they are named. Existing operations will be refined as practice reveals what they are doing. Some will be removed if they turn out to be doing less work than their presence suggests.
Contributions welcome. Translations into additional languages welcome. Critical essays welcome, especially ones that identify where the framework is drifting, over-claiming, or producing effects that contradict its stated commitments.
MIT licensed because MIT licensing is what lets tools propagate. The operations do not belong to anyone. This document is one pass at naming them. Better passes will come from readers who adopt the operations and develop them for their own conditions.
Put the cost-bearer in the room, name a specific fix with a specific agent and a specific deadline, hold the fix through deflection, and audit yourself the same way you audit everyone else.
MIT License
Every consequence has a source. I find the source.