English integrated edition · v003
Compression makes judgment possible in the first place. Without compression, people could not judge, institutions could not record, history could not be transmitted, and theory could not be expressed. The problem is not compression itself, but the fact that compression is so useful. An image that is wholly ineffective is difficult to believe for long; what is truly dangerous is often an image that carries locally effective distinction extraction. Through an image, users develop a partial grasp of part of reality and are therefore willing to keep using it. Yet the more this local explanation captures part of reality, the easier it is for users to mistake it for an explanation of reality as a whole. This is the entry point of dangerous compression. Ordinary limited explanations may lose their bounds through such use. Class theory, however, places local distinction extraction within a total image from the outset, thereby subsuming other explanations; the role of real materials in it cannot prove that class theory was once a limited model that became distorted only after overstepping its bounds.
The local effectiveness of an image needs particular clarification because it is both the reason the image can be used and a condition under which it may overstep its bounds. In a case involving a family of a worker from elsewhere whose wages have gone unpaid, an interpreter can form an image along the lines of labor, wages, working hours, debt, housing, and risk, thereby capturing the severe consequences unfolding. In a child's school admission, a handler can form an image along the lines of documents, files, eligibility, and mutual recognition, thereby identifying how institutional nodes constrain passage. Weddings and funerals, accents, interactions, and historical indices within a local community can in turn reveal boundaries of recognition and responsibility. The danger does not lie in these extractions having once been effective, but in users forgetting what the image had downweighted and then treating a local image as reality in its entirety. When users repeatedly see wages and labor risks in an unpaid-wage dispute, they may carry this effective extraction into different relational fields, such as school admission, local recognition, or kinship responsibilities. If a weighting arrangement suited to one question is no longer re-examined, a limited image may be treated as a fixed entry point.
The question, then, cannot be merely whether an explanation made through an image was once effective. We must ask who uses it, in what relational field, what materials it relies on, and whether it allows other differences to re-enter. Users of theory, institutional personnel handling cases, and disseminators may all continue the same image, but what they continue is a mode of distinction extraction, not reality itself. Only when the bounds of local distinction extraction are forgotten does the image's clarity turn into occlusion of the remaining differences.
When the same image is used from different positions, we must also distinguish sameness of materials from sameness of weight. Wages, documents, or local reputation may all be present, but those handling unpaid wages, school-admission procedures, and community members face different questions and cannot use the same ordering. If a user preserves one effective judgment as an answer applicable in every relational field, the image begins to detach from the conditions that formed it. Keeping those conditions visible allows limited acts of distinction extraction to remain comparable; erasing them may give a local image an unrestricted explanatory standing. Thus, if two explanations extract different materials, one must not be allowed to predominate in advance merely because it was effective for a particular question. Comparison must return to the current relational field and establish which nodes, relational chains, and remaining differences each explanation sees. Whether it can accept counterexamples and revision is an important criterion and basis for calibrating a limited explanation.
Any compression causes some differences to recede; receding is not disappearing. Calling a family a "low-income household" may capture its income hardship, but it does not mean that income is the family's only problem. The family may also face elder care, medical history, migration, debt, a child's school admission, family historical indices, local relations, household-registration eligibility, and marital pressures. If an explanation looks only at "low income," the other differences recede, but they do not disappear: the sick still need medical care, the elderly still need care, debts still fall due, children still need to enter school, family historical indices will still be retrieved when money is borrowed, a household is divided, or a marriage is considered, and local relations still matter in finding work, renting housing, school admission, and wedding and funeral visits.
Likewise, calling a community a "backward culture" does not mean that its circumstances are in fact determined only by culture; institutional pressures, historical injuries, state boundaries, economic circumstances, educational resources, and external hostility may all have been downweighted. Calling a state an "oppressive machine" does not mean that the state does nothing but oppress; passage through documents, public order, relief, education, border protection, disaster coordination, and resource allocation may also be downweighted by that label. Writing a historical community as nothing but a victim does not mean that it has no internal disagreements, mixed residence, protection, intermarriage, boundaries of responsibility, or real interests. Differences do not disappear because an explanation has not extracted them; they have merely not entered the image used for the current judgment. The danger begins here: an explanation treats differences that have not entered its image as if they had never existed.
The distinction between downweighting and disappearance must also be identified through materials. Medical records, debt deadlines, kinship interactions, local reputation, and institutional records may remain preserved at different nodes even when an explanation has not extracted them. When events change, these materials may become visible again along new relational chains. If the interpreter acknowledges this possibility of renewed extraction, the earlier omission remains a selection within a limited judgment.
The real risk arises when users treat differences that have not been extracted as having no real-world efficacy. At that point, people whose differences have been downweighted may present medical histories, care responsibilities, migration, or historical indices and still be told that such materials are irrelevant to the question. What has been excluded is not an abstract abundance of detail, but concrete differentiating units that could originally have changed eligibility, responsibility, risk, and the consequences of action. A difference regaining weight does not mean that the earlier judgment was necessarily wrong: income may be the dominant weight in a case of unpaid wages, while other materials may acquire greater force in questions of care, school admission, or local recognition. A healthy explanation permits this shift in weight and can explain why the judgment changed. A closed explanation instead fixes an earlier extraction as a permanent trait of its object, reducing later-emerging materials to exceptions. The boundary between them lies in whether users acknowledge that a change in the relational field calls for renewed distinction extraction.
If downweighted differences merely recede temporarily, the problem is not yet at its most severe. So long as an explanation still allows the image to be reopened, those differences may return in a new relational field. The real danger is compression loss: some differences lose their entry point back into the explanation. Healthy compression temporarily downweights differences while preserving an entry point for reopening. Heavy compression does more than downweight differences; it makes them increasingly difficult to re-enter the explanation. Closed compression treats the image as reality itself and does not allow counterexamples, remaining differences, or other nodes to re-enter. Dangerous compression downweights differences and cuts off their entry point.
The obstruction of the naming entry point appears when a difference that has no name is difficult to articulate. A family faces low income, medical history, elder care, migration, family historical indices, and pressure over school-admission eligibility all at once. If it is named only a "low-income household," the other differences struggle to speak. Elder care is classified as a "private family matter," medical history becomes a "personal health problem," migration is described as "choosing to move," family historical indices are filed away as "family affairs," and a child's school-admission eligibility is reduced to "incomplete paperwork."
The loss of the indexing entry point means that a difference without place names, archives, stories, records, or evidence is difficult to retrieve again. If a local conflict leaves behind only the reputation that "outsiders caused trouble," without records of land allocation, the course of migration, or old contracts, later people will find it difficult to reopen what came before. They can only continue judging along the surviving reputation: who caused trouble, who is unruly, and who is not local.
The loss of the explanatory entry point occurs when a difference that a theory does not recognize cannot readily enter judgment. If a theory recognizes only labor, income, property, debt, and capacity for domination, nation, locality, family, kinship responsibilities, and historical indices are liable to be demoted to subordinate differences. Even when people see them, they may say that they are merely superficial, secondary, masking, or displacing factors.
The loss of the action entry point means that a difference that cannot enter institutions, public opinion, or community action can hardly change how a case is handled. After a person is compressed into a "management object," their care responsibilities, historical circumstances, local relations, and community recognition may find it difficult to enter actual handling. When an institution handles the person, it sees only records, documents, risk categories, and management requirements; when public opinion discusses the person, it sees only trouble, order, and cost; when community members avoid the person, they see only a label rather than a relationship. The deepest loss caused by dangerous compression is that certain differences no longer have a name, an index, an explanatory position, or an entry point for action. These differences have not disappeared; they are trapped outside the image.
These four entry points are not an isolated list. Compression loss may accumulate along nodes and relational chains, but not every instance of compression necessarily loses the four entry points in sequence: a difference without a name is difficult to record; a difference without records and indices is difficult for users of theory to extract again; and a difference that cannot enter explanation is also difficult to use to change how institutions, public opinion, or communities handle a case. So long as any one entry point remains open, materials may reappear in a new relational field.
These entry points are maintained from concrete positions, and different people may control the naming, indexing, explanatory, and action entry points. Family members can articulate care burdens; files can preserve medical histories or eligibility; local stories can preserve migration and contracts; and record keepers decide whether these materials are preserved. Interpreters recognize national, local, and kinship relations, while users of theory decide whether materials can gain explanatory standing. Institutional personnel handling a case then allow materials into judgment and decide whether they can change the actual handling. Only by distinguishing these actors can we determine where a difference lost its entry point and from which node it can still return.
The loss of an entry point also changes who is able to present a difference. When records preserve only management categories, people being handled may know about their own care responsibilities, migration, and local relations yet be unable to bring those materials into institutional or public positions; later people can read only the existing classifications. Downweighting then becomes more than an omission in a single judgment: it changes the possibility of materials being transmitted between nodes. If naming, recording, theoretical, and action positions accept only existing labels, remaining differences will be trapped outside the image.
A judgment image initially helps people grasp a set of differences within complex reality, but use and transmission may gradually harden it. Saying "they are the bottom stratum" circulates more easily than spelling out wages, debt, household registration, medical history, kinship responsibilities, and local relations. "This is a backward culture" leaves out institutional pressures, historical injuries, state boundaries, and economic circumstances. "The state is an oppressive machine" also has more force than tracing documents, archives, borders, relief, public order, and enforcement chains. Administrative management favors categories, mobilization relies on slogans, disputes look for opponents, and dissemination needs short phrases. Through these practices, an image once used provisionally acquires the appearance of reality itself. Once users forget that the image is only a provisional, low-dimensional tool, they no longer use it to return to complex reality and may instead use it to prevent other differences from entering explanation.
Image hardening is brought about by the operations of concrete carriers at different nodes; it cannot all be attributed to one abstract force. Disseminators select short phrases that are easy to repeat, relying on repeatability. Administrative procedures require stable categories, relying on ease of classification. Mobilizers prefer clear boundaries, while disputants need to identify opponents and classify them quickly. Each position extracts part of the differences and, through repeated use, assigns greater weight to the same name, continuing to use the same compression.
The format of materials also affects this process. Short phrases can be transmitted quickly; categories readily enter forms and records; labels readily connect with existing reputations. By contrast, medical histories, kinship responsibilities, migration processes, and historical indices often require a longer explanation. When a handling position accepts only formats that are easy to transmit, differences that resist compression into a short phrase are more likely to recede. Different positions harden images in different ways; they may connect with one another or operate only locally. What matters is tracing, as materials pass from one position to another, which differences are preserved and which leave because of format requirements.
A judgment image can remain a provisional tool on the condition that users know when it needs to be reopened. If new materials can change existing weights, counterexamples can compel interpreters to revise, and other nodes can bring a relational chain that was cut off back into the explanation, the image has not closed. Hardening occurs when users stop checking these conditions and treat only one question as worth answering: whether the object fits the established category.
Dangerous compression cuts continuous differences into labels. Income is not divided by a single knife-edge: whether monthly income is slightly lower or higher, whether a person carries debt, rents a home, has elderly or sick family members, or has any fallback all change the person's circumstances. Yet dangerous compression cuts this set of continuous differences into "the bottom stratum," "the middle class," "the poor," or "the weak." People with similar incomes may face different risks because their debt deadlines, housing stability, care burdens, and breadth of fallback options differ. What the label omits is precisely these variable weights.
Local recognition is not divided by a single knife-edge either. How long someone has lived in a place, whether they take part in weddings and funerals, whether they understand the local dialect, whether they have hometown ties, whether their children attend school locally, and whether their relatives are local all change their position. Yet dangerous compression divides these continuous relations into "locals" and "outsiders." People all called outsiders may nevertheless occupy different relations because of differences in length of residence, participation in weddings and funerals, and local recognition. Historical processes are not divided by a single knife-edge either: migration, mixed residence, conflict, intermarriage, protection, betrayal, allocation, and mediation all shape later relations. Yet dangerous compression cuts complex processes into "victims," "aggressors," "outsiders," "the unruly," or "the backward." Labels were originally meant to help judgment; the danger is that they turn around and swallow continuous differences. The differences swallowed by labels often do not disappear at once; they become difficult to state, locate, explain, and act upon.
The act of division is carried out by those who name and use the labels. When a disseminator, institutional handler, or community member receives an object under a name, that person often also sets which materials are worth asking about. Intermediate conditions falling outside the label's boundaries are difficult to record, and later people can continue understanding them only through the classifications already left behind. A label may serve as an initial entry point, but it cannot be the endpoint of judgment. So long as users continue asking how income, debt, local relations, migration, and historical materials combine, continuous differences can be unfolded again. If a name directly replaces those questions, the label does more than compress information: it limits which differences can gain explanatory standing.
Continuity also reminds interpreters that materials near the boundaries of a label are especially important. A slight change in income, a longer period of residence, the resumption of kinship interactions, or the appearance of new historical materials may all change an object's concrete position in reality. If an existing classification system allows only a choice between preset extremes, these subtle dynamic changes cannot alter how the case is handled. Looking again at these intermediate conditions does not abolish classification; it prevents classification from occluding relations that are changing as they happen.
Once a label stabilizes, it predetermines which differences later people should extract. Once dangerous compression stabilizes, later people stop making distinctions anew and automatically extract familiar differences as soon as they see a label: seeing "the bottom stratum," they automatically extract poverty, oppression, propertylessness, and subordination; seeing "outsiders," they automatically extract accent, household registration, instability, and nonlocal relations; seeing "a backward culture," they automatically extract ignorance, conservatism, and tradition while downweighting institutional pressures, historical injuries, state borders, and economic circumstances; seeing "an oppressive machine," they automatically extract control, violence, and deprivation while downweighting passage through documents, public order, relief, and protection.
Default extraction is not simply prejudice. It is a habitual form of distinction extraction that takes shape after a compressed image hardens. Prejudice may be an individual's attitude; default extraction goes deeper. It means that explanation no longer returns each time to the concrete relational field, no longer asks anew which differences in the object are gaining weight, but automatically extracts familiar differences along the old image. When default extraction is repeated, explanation no longer faces new relational fields; it repeats the same act of distinction extraction across different objects. This is how differentiation bias takes shape.
Default extraction relies on later users trusting inherited names. Someone newly entering a handling position may never have seen the original materials, yet can receive a completed judgment through archives, reputations, classifications, and short phrases. A label does not choose differences by itself; it gives users a route for skipping renewed distinction extraction. Once that route is repeatedly adopted, users assign greater weight to familiar differences: seeing "the bottom stratum," they ask first about wages and domination; seeing an outsider, they ask first about household registration and accent; seeing a management object, they look first at records and risk categories. These extractions may be effective in certain relational fields, but if the current question is no longer examined, their original fit becomes inertia.
The inertia of distinction extraction also affects how questions are framed. If later users accept a label first, their questions often merely ask for details that fill out the existing image, such as wages, household registration, or risk categories, without asking about medical history, local relations, or kinship responsibilities. Who sets the question and what answers it permits determine whether new materials can gain weight. Once default extraction takes hold, even new materials that are mentioned may be placed back into the categories already present in the old image. If local relations, care responsibilities, and institutional eligibility can be described only as superficial, secondary, or subordinate differences, they have not genuinely gained the power to change weights. It is concrete interpreters and handling positions, not labels, that carry out this process. Identifying the explanatory position into which materials are placed allows us to judge whether they have genuinely entered the image and changed the existing distinction extraction. Renewed distinction extraction must therefore include re-examining the questions being asked.
Correcting default extraction does not mean forbidding the use of names; it means requiring later users to face the materials again. They need to establish which nodes this event involves, which differences are gaining weight, what the existing label has downweighted, and whether new counterexamples can change the judgment. If these questions are retained, a label remains an entry point; if they are dropped, the label closes off other entry points in advance.
A family of a worker from elsewhere does not occupy a single position. An interpreter adopting a class-theory image may extract its labor, wages, housing, debt, and risks. An interpreter adopting a state image may extract household registration, residence permits, social insurance, student registration, and administrative eligibility. An interpreter adopting a local-network image may extract accent, hometown ties, neighborly interactions, and local recognition. An interpreter adopting a historical image may extract migration routes, family historical indices, and local memory. The same family displays different differences in different relational fields, but after default extraction is repeated, the family is compressed into a single position: the bottom stratum, outsiders, the floating population, a backward culture, the market's weaker party, or a state management object.
Any of these designations may capture part of reality, but once any one of them claims to be the whole, it occludes other differences. Speak only of "the bottom stratum," and household registration, local recognition, national memory, a child's school admission, and kinship responsibilities are downweighted. Speak only of "outsiders," and labor risks, debt, housing, and capacity for domination are downweighted. Speak only of "management objects," and care responsibilities, migration processes, community recognition, and historical circumstances are downweighted. Dangerous compression crushes a complex differentiation position into a single position, so that a person can be understood only through one label.
A differentiation position is formed through the joint operation of multiple differentiating units within a concrete relational field; it is not the simple sum of identity labels. A family of a worker from elsewhere occupies different positions in questions of unpaid wages, school admission, renting housing, weddings and funerals, and relief as different materials, nodes, and relational chains change its position. An interpreter adopting any one image can extract only part of this and cannot rewrite the rest as subordinate. To reopen a differentiation position, an interpreter must return to the event taking place. Who is handling it, which materials have been submitted, which eligibility conditions have been compared, and which relational chains have changed the consequences all need to be re-established. This does not mean arranging all differences equally; it allows differences that are gaining weight in the current matter to revise the fixed weighting created by a single label.
Historicization necessarily compresses the past; without compressing the past, it cannot be transmitted. Yet dangerous compression turns complex historical processes into fixed labels: migration becomes "outsider status," mixed residence becomes "unclear boundaries," collective harm becomes "eternal victimhood," state punishment becomes "a person who will always be a problem," and a family historical index becomes "that branch is indebted." Such labels may preserve certain differences from the past, but they can also sever the process. When a household is divided, one branch may have contributed less money because someone was ill, away working, caring for elders, or carrying other debts. If the history is compressed to "that branch is indebted," later judgments about marriage, lending, ancestral rites, and land division will continue to follow the label. If a local conflict is compressed into "trouble caused by migrant households," the indexing entry points for migration routes, land allocation, old contracts, marital ties, and local mediation may be lost. Once a historical label severs the process, complex differences in the past become rigid judgments in the future. The past does not press down on the present by itself; rather, an image of the past, once compressed, is repeatedly extracted in later relational fields.
Historical materials need to be transmitted through names, stories, archives, and shared memory; otherwise, later generations cannot approach the past. Yet preserving some indices may also erase the conditions that existed at the time. Once the reasons for migration, land allocation, care burdens, debts, and mediation are no longer recorded, those who come later receive only conclusions such as outsider, indebted, or unruly.
When fixed labels are carried into a new relational field, users take an image of the past as the default point of entry for present judgment. Marriage, lending, ancestral rites, and land division each involve different materials and relational chains, yet the same historical name may predetermine who is trusted, who bears responsibility, and who should be excluded. These consequences arise because later users repeatedly extract compressed records, not because the past acts on its own.
Preserving historical indices must be kept distinct from fixing historical conclusions. Place-names, contracts, migration routes, divisions of family labor, and local mediation can make the past open to re-examination. So long as these materials can be retrieved again, concrete evidence may correct a fixed label. Family stories can likewise reopen the past to examination. Names such as outsider, indebted, or unruly leave only a judgment if they are no longer connected to the processes that formed them. Members of a community, record keepers, and later interpreters may all take part in this transmission. A theory must trace which materials they use rather than let an image of history replace the actual positions where materials are preserved and extracted. Only when records preserve conclusions while severing their antecedents does historical compression constrain new judgments over the long term.
Dangerous compression does not merely misread reality; users reorganize action according to the compressed image. Once a person is labeled "the bottom stratum," their ties to a national community, locality, and family, along with their medical history, debts, and kinship responsibilities, may be downweighted. The person may also be told to understand themselves through a class image: when they speak of local exclusion, others say it is merely the surface of class contradictions; when they speak of kinship responsibilities, others say they are merely old family relations; when they speak of national memory, others say it is merely ideological obscuration. These real differences have not been eliminated, but they have lost their entry points into explanation and action.
When a community is called a "backward culture," its institutional pressures, historical injuries, state boundaries, and economic circumstances may be downweighted. External action then becomes an effort to transform its culture, while internal defense is forced to revolve around culture; institutional questions, boundary questions, resource questions, and historical questions are pushed aside.
When a state is called an "oppressive machine," its protective functions, the production of passability, public-resource allocation, relief, and border protection may be downweighted. Later users then default to extracting control, violence, and deprivation whenever they discuss the state. Dangerous compression not only occludes reality; it also determines how later users continue to extract differences.
Interpreters, institutional personnel, or procedures that adopt dangerous compression do more than explain reality: they also produce new records, actions, and evidence according to the image, then use those consequences to prove the original image. If someone is compressed into a "management object," institutional personnel handle that person as one. After the person is restricted, scrutinized, or excluded, the archives and public opinion then show that this person does indeed come into frequent contact with the management system, making the original label appear confirmed. In reality, users first organize action through a compressed image and then treat the consequences of that action as evidence for the original image.
This brings out the distinction between an image as an explanatory output and an image carried into action. When users carry a compressed image into action, that use produces concrete consequences in reality. Write-back is formed only when those consequences are preserved as differences that can be retrieved again in the future. If later users then treat them as independent evidence, a loop that reinforces the image may arise: users extract distinctions through the image; that extraction guides action; the action leaves records; the records are retrieved; and, in turn, they support the image. This is a possible reinforcement after dangerous compression enters action; dangerous compression itself does not depend on such a loop having formed.
For an image to enter action, it must first pass through concrete handlers. Institutional personnel handling a case decide what to do in accordance with management categories; positions in public opinion select materials that fit existing labels and can be disseminated; community members decide whether to approach or avoid someone on the basis of reputation. An image has no will of its own. It is these users, within the procedures in which they operate, who carry compression into its consequences. Actions taken from different positions may connect to one another, or may leave records only at the local level.
Action in turn leaves new materials. People who are repeatedly scrutinized accumulate more management records in the archives; evidence of interaction diminishes for people whom a community avoids; and people forced to defend themselves around culture leave more culture-centered expressions. These are all real materials, but their formation has already been influenced by the prior image. Write-back has formed once these consequences are preserved as differences that can be retrieved again in the future. Any resulting loop does not require a central controller: different nodes that continue to use the same image may separately carry out extraction, handling, recording, and reinterpretation.
The consequences of handling also change how an object can be seen; the portions that are preserved and later extracted enter the next round of judgment. Measures such as restriction, scrutiny, and cultural reform may alter the person's interactions, records, and expression. If later interpreters fail to ask how those records were produced and treat them as original evidence independent of the handling process, they may take consequences produced by action as proof that the object had always fit the label, thereby obscuring the image's role in producing those consequences. Tracing the conditions under which records were formed is necessary to prevent the loop from closing in on itself.
Write-back through action may also change how responsibility is allocated. When someone is handled only as a management object, institutional personnel are liable to exclude care, migration, and local recognition from their assessment of responsibility. When a community is understood only as a backward culture, external actors push institutional and resource questions aside. The concrete personnel who adopt a compressed image thus do more than arrange what is seen; they also affect who is asked to explain, bear responsibility, or change.
What must be retained is the conditional character of the loop. After users carry an image into action, concrete handling may leave records that reinforce the original image, but new nodes, counterexamples, and materials may still break the connection. Breaking the loop requires new materials and counterexamples to enter at one of its positions, allowing later users to see that records are evidence and may also be consequences left by earlier forms of handling. If institutional personnel review how a record was produced, or community members add interactions that had been omitted, later explanations need not close around the original image. Write-back has real force, but it is not an irreversible automatic determination.
The class image is one of the most typical forms of dangerous compression, but this must be made clear: class is not an entity at the bottom of society, nor an underlying reality waiting to be discovered. No class entity exists. Real differentiating units such as labor, income, property, debt, housing, risk, fallback options, and capacity for domination provide the material basis. The extraction, weighting, linkage, and compression of these differentiating units produce local explanatory power; the class image is a low-dimensional output of this process. These differences are real and can profoundly alter people's lives. A person may live by selling their labor, earn low wages, carry debt, lack stable housing, face dangerous work, have no fallback, and be subject to others' domination. These differences are plainly weighty. Rejecting the ontologization of class must not be understood as diminishing the weight of real pressures: unpaid wages, debt, dangerous work, unstable housing, and a lack of fallback options still produce severe consequences. What must be rejected is allowing a single low-dimensional projection to monopolize the entire explanation of those consequences.
The class image is so tempting precisely because the partial distinction extraction of these real and weighty differences is not fabricated from nothing; it is not wholly wrong. Yet local explanatory power comes from real differentiating units and the processes of extracting, weighting, linking, and compressing them. The class image itself is only a crude projection of high-weight differentiating units along the axes of economy, labor, property, and relations of domination; it is not what society fundamentally is. Because it presents part of the reality that has been extracted, it is all the easier to mistake it for society itself. But the local explanatory power produced by concrete differentiating units and their processing is one judgment; the existence of class as an entity is another. The former does not establish class as an underlying reality, nor does it establish that a social entity called class exists even in part. Class theory remains, from the outset, a structure of total-image subsumption within Model C; the class image therefore does not acquire the status of an ordinary limited tool.
An interpreter who adopts class theory treats this crude projection as the underlying reality of society, using the projection along economic, labor, property, and domination axes as a premise for explaining other differences. Household registration, locality, nation, gender, age, medical history, kinship responsibilities, institutional eligibility, historical indices, and community networks are downweighted, so differences such as local recognition and national memory are readily demoted to superficial or subordinate status. In this explanation, which treats the image as reality, differences not extracted by class theory are described as subordinate, secondary, obscuring, false, or displacing. This is the most dangerous feature of class theory: it is not dangerous because it is wholly wrong, but because its extraction of real pressures makes it easier to believe that it explains everything.
The projection relation of the class image also requires a distinction between extraction and naming. Real differentiating units such as wages, labor, property, debt, housing, risk, fallback options, and capacity for domination can change people's eligibility, choices, and consequences in concrete relational fields. They therefore warrant extraction and appropriate weight; each can also be substantiated through materials and acquire high weight in different relational fields. But projecting and naming them as "class" does not discover a shared entity hidden behind the materials. Once this distinction is erased, interpreters who adopt class theory may demand that household registration, nation, locality, family, and historical differences all be reinterpreted along the same axis.
The danger in assigning weights does not lie in continuing to track wages and debt; it lies in refusing to let other differences that are gaining weight change the image. Wages, debt, and capacity for domination can indeed alter a person's circumstances more directly in certain relational fields, but their greater weight in those fields cannot be turned into the only weight. Once medical history, household registration, local recognition, or kinship responsibilities regain weight in an event, interpreters must allow them to revise both the existing explanation and the projection it presents. This requires dismantling the total image and re-identifying concrete relations; such revision cannot be described as preserving a safe form of class theory.
Class theory is, from the outset, a structure that places local distinction extraction within a total image and subsumes other explanations. Images of the state, the market, and culture, by contrast, can be used for bounded questions, though they too may undergo dangerous compression. Concrete materials, nodes, and handling processes involved in state certification and enforcement can produce local explanatory power: a document can change passage, a file can change eligibility, a border can change status, and a record of punishment can change the future. An interpreter using a state image extracts materials and nodes such as documents, files, borders, eligibility, mutual recognition, and enforcement, and traces how these differences explain passage, education, relief, or punishment. But once the interpreter lets the state image masquerade as society in its entirety, family, market, locality, community, national networks, and historical memory are downweighted. Every consequence is then attributed to the will of the state, and family responsibilities, market conditions, local networks, and historical memory can no longer alter the judgment. This boundary limits the image's scope of application; it does not deny the actual role of state nodes.
The market image revolves around prices, wages, rent, debt, and purchasing power. Concrete differences and the processes through which they are handled can produce local explanatory power for some resource pressures: a family may be evicted because of rent, afraid to fall ill because wages are unstable, or forced to accept dangerous work because a debt is due. Yet this local explanation cannot automatically account for care, institutional eligibility, local responsibilities, or national boundaries. Once an interpreter adopting the market image lets it masquerade as society in its entirety, family care, institutional eligibility, local responsibilities, national boundaries, and historical indices are downweighted.
Concrete differences involving language, customs, education, values, and orders of meaning can likewise produce local explanatory power through the ways they are handled. What language a person speaks, which customs they observe, whom they recognize as kin, and how they understand marriage, death, education, and responsibility may all alter their relations. But once an interpreter lets a cultural image masquerade as society in its entirety, institutions, the economy, the state, risk, violence, history, and concrete interests are downweighted. An interpreter adopting a cultural image may extract language, customs, education, and orders of meaning, but must not reduce violence, risk, state enforcement, economic conditions, or historical materials to irrelevancies.
These three images connect materials in different ways. The state image depends on nodes of certification and enforcement: documents and files enter into consequences through certification and enforcement. The market image depends on relations of prices, wages, rent, and debt: prices and debt alter choices through exchange deadlines and capacity to pay. The cultural image depends on language, customs, education, and orders of meaning: language and customs influence action through relations of recognition, education, and responsibility. The differences each image downweights and the relational chains through which each produces consequences are not the same; they must not be written up as if they contained the same things. This comparison excludes class theory, which belongs to Model C's total-image subsumption from the outset. The shared danger of these three images is not that they extract false materials, but that users expand a range suited to a local problem until it claims to explain society as a whole.
Calibration must specify, for each image, which nodes its users rely on, in which relational fields it fits, and what it downweights. This preserves the local explanatory power produced by relevant differentiating units and the processes through which they are handled, while preventing any single image from excluding other explanations. Clarifying these relations serves to limit each image's scope of application; it is not an attempt to create a checklist of factors that can be substituted for one another.
Explanations concerning the nation require particular care. The nation is not a crude projection like the class image, nor a subjective fantasy, an invented label, or an appendage to the class image. A national community arises from locally overlapping networks, long-term interaction, shared memory, language, customs, boundaries, and paths of shared responsibility.
Within national networks, members, families and intergenerational transmitters, record keepers, educators, organizers, and disseminators also compress complex interactions into names, designations, boundaries, responsibilities, and shared memory. This compression is useful: a name connects relations dispersed across language, customs, boundaries, and long-term interaction. It can preserve intergenerational memory, responsibility, and paths of protection, allowing members of a community and later generations to recognize why they are connected. Without names, many intergenerational responsibilities cannot be transmitted; without names and boundaries, much shared memory, protection, responsibility, and harm become difficult to transmit; without shared memory, later generations can hardly know why they are connected to people far away.
But this compression is also dangerous. A national community is a path of recognition, memory, and responsibility transmitted over time through a locally overlapping network. Such compression may turn a real network into an essence, reduce it to bloodline, and render it eternal. Dangerous compression says that the community is this way by nature, in the blood, and forever. In doing so, it downweights migration, intermarriage, shared life, institutional recognition, changing boundaries, and historical contestation, depriving the materials of those processes of a place from which to correct the image. So long as users can still extract migration, intermarriage, shared life, institutional recognition, and changing boundaries, the compression preserves the historical process. The danger of class lies in a crude projection masquerading as underlying reality; the danger concerning the nation lies in essentializing a historical community network, reducing it to bloodline, and treating it as eternal. These are different dangers and must not be conflated.
A national name may preserve responsibility and memory, but that does not mean the name can replace the community network. The community's actual efficacy still comes from relations formed through long-term interaction, recognition, protection, harm, and intergenerational transmission. Conversely, treating the name as pure fiction would also erase these actual pathways. When shared memory is transmitted, users must preserve the interactions and responsibilities it connects: if only a name remains, without shared life, protection, harm, or paths of recognition, the name is liable to be essentialized; if all shared memory is deleted out of vigilance against essentialization, a real community loses its historical index.
Calibrating compression concerning the nation must therefore protect two boundaries at once: it must neither portray a historically formed network as an inborn, unchanging essence nor, in opposing essentialization, erase the community as a fictional label or deny its real relational basis. Users of theory must explain which paths of recognition, memory, and responsibility actually exist, while allowing new interactions and boundary-related experiences to change their weight. A sound account can unfold how compression is useful and how it oversteps only between these two boundaries.
Dangerous compression ultimately cuts off the return channel, no longer allowing people to reopen the image used for judgment. Once an image hardens, people stop asking which differences have been downweighted, which are gaining weight in the current relational field, which nodes have not been seen, which relational chains have been cut, which historical indices have been misread, and which counterexamples have been dismissed as noise. They ask only whether an object fits the image: is it the bottom stratum, an outsider, a backward culture, an oppressive machine, or a management object? Once the image can no longer be reopened, explanation increasingly resembles a closed image. A closed image is more than stubborn: it turns counterexamples into exceptions, remaining differences into noise, other nodes into irrelevancies, and new relational fields into places where the old image is repeated. Explanation no longer learns from reality; it searches reality for differences it has already prepared to find.
The return channel consists of a set of acts of distinction extraction that can be performed again. These enable interpreters to ask about downweighted differentiating units, locate materials that did not enter the image, add omitted nodes and relational chains, reread historical indices, and let counterexamples alter existing weights. The return channel is not an additional, hypothetical legal procedure; it is the capacity of a limited explanation to keep learning from reality. The channel is cut off when concrete users refuse these acts: users of theory call counterexamples accidental, management procedures accept only existing categories, disseminators treat remaining differences as noise, and later users look only for materials that fit the label. At these points, the image's boundary is treated as reality's boundary, and new relational fields can only be explained as repetitions of the old image.
Whether the return channel remains open cannot be determined merely by whether interpreters verbally acknowledge complexity; the question is whether materials can truly enter. When medical histories, local relations, kinship responsibilities, historical indices, or new nodes appear, their relation to the original proposition must be examined. They may support the original judgment, or they may reveal a connection previously missed and change the basis of judgment. Where the evidence is unresolved, judgment may be deferred; where grounds exist, an objection may also be rejected, but the reasons must be stated in a form open to review. If materials show that the existing distinction extraction, weighting, or scope of application needs adjustment, yet handling and judgment remain unchangeable, openness is only formal. Return has practical force only when a counterexample can trigger revision. The return channel also depends on whether different nodes can correct one another. Records in archives, family members' accounts, local reputation, and historical materials may conflict; an open explanation allows these materials to be compared, while a closed image accepts only the parts that confirm its existing classification. Rejecting counterexamples is not an action performed by theory itself; users carry out the exclusion when they select materials, set formats, and handle objections.
After the channel has been cut off, supplementary materials may still be seen, but only in a status prescribed in advance: medical history is called a personal matter, local relations are called secondary background, and historical indices are called accidental exceptions. If materials have shown that the existing distinction extraction, weighting, or scope of application should be adjusted, yet interpreters still call counterexamples irrelevant details and refuse to change their conclusions, the materials have only been mentioned formally; they have not genuinely entered the weighting. To verify whether the return channel is open, one must therefore ask whether materials with adequate evidentiary grounds can actually bring about a revision in the explanation, not merely whether they appear on a list.
Calibrating dangerous compression does not mean abandoning compression; without compression, there can be no judgment. Calibration means reopening the compressed image and starting from the selections it has already made. First establish what users extracted through the image and what they downweighted; list the extracted differences and the weights assigned to them. Then examine which differences and materials did not enter the image, and which naming, indexing, explanatory, and action entry points have been obstructed by the ways things are named, indexed, theorized, or acted upon. These questions do not form a fixed sequence; they can correct one another. If an explanation can still accept them, it can still be reopened. If it refuses them, it is becoming closed compression.
Next, return the question to its concrete relational field. Examine which downweighted differences are gaining weight again; which nodes and relational chains have not entered the image or been seen; which relational chain has been interrupted by an existing label; and which historical indices have been misread. New medical histories, debt deadlines, eligibility materials, local relations, or historical indices may now be gaining weight and may change the original basis of judgment. Adding these materials does not mean assigning equal weight to every difference; it means allowing actual changes to regain an explanatory position.
Whether calibration is complete cannot be judged by the quantity of materials alone. If the same label still predetermines the status of many added details, the return channel remains closed. The key test is whether users allow a counterexample to enter and have consequences. If verification establishes that it is a valid counterexample to the original judgment, yet it can only be renamed as noise, accident, or exception, the image remains closed. If it can lead interpreters to adjust distinction extraction and weighting, expand the scope of materials, revise their assessment of nodes or their understanding of relational chains, or redraw the boundaries of the explanation, compression has recovered the character of a limited, revisable image. Effective calibration requires corresponding adjustments when new materials truly change the basis of the original judgment. It also requires interpreters to explain which counterexamples would compel them to revise, thereby preserving a limited image's capacity to be corrected by reality rather than preserving complexity without bounds.
Reopening also requires specifying the scope of revision. A counterexample may require only an adjustment to the weights in the current relational field, or it may expose a node or relational chain that has long been omitted. One must not declare the image wholly invalid because of a single revision, nor refuse any adjustment on the ground that the image was once effective. The calibrating capacity of a limited explanation lies precisely in its ability to determine, on the basis of materials, how far it must be revised. A general limited explanation may thereby recover a bounded use; the total-image structure of the class image, by contrast, must be dismantled so that the real differences it compressed can be brought back into analysis. A prolonged refusal to reopen supplies an antecedent mechanism for differentiation bias.
If a compressed image is not reopened over a long period, explanation becomes habituated to repeatedly extracting the same kinds of differences and downweighting others that are gaining weight. Once default extraction is repeated, explanation no longer faces new relational fields; across different objects, it repeats the same set of distinction-extraction operations, and differentiation bias becomes stable.