Chapter 05 What Is a Differentiating Unit? How Social Differences Become Differentiating Units That Can Be Extracted, Assigned Weight, and Connected to Consequences
I. Questions in the Situation at Hand and the Formal Definition of the Differentiating Unit
If grand concepts such as class, the state, the market, the nation, culture, and the family are not treated in advance as completed explanations of society, social analysis must keep asking: which differences actually enter into social consequences? Merely saying that society is complex, or placing many factors side by side, cannot trace how differences enter concrete relations and consequences. A child who cannot enter school, a worker who cannot obtain compensation, a cash gift that causes estrangement, a person excluded because of an accent—none of these can be left at the level of grand concepts. Which difference was extracted within which relation, and how did it change the result? That is the question that must be pursued.
When a child enters school, formal institutional factors such as household-registration conditions, school places, and application rules may be involved; where local policy expressly requires them, some documents concerning residence, social insurance, renting, or filing may also enter the review. Document deadlines, whether the parents can take time off, and whether the landlord cooperates in producing rental documents, by contrast, are living conditions a family faces while going through the procedure; a statement from the community is formal documentation only where the specific rules expressly make it so. Formal institutional requirements and informal living conditions jointly affect the result, but they must not be mixed together and written up as a single set of admission rules. The same applies to compensation for workplace injury. Domination, risk shifting, and difficulties of exit within labor relations matter, as do the determination of workplace-injury status, the assessment procedure, and rules of evidence. The concrete result must still be traced to whether a contract exists, whether medical records are complete, whether surveillance footage has been preserved, whether coworkers are willing to testify, who advances the medical expenses, whether the family can endure the wait, and whether the person concerned understands the procedure. Grand concepts provide direction, but they cannot replace tracing how consequences form.
A differentiating unit is a unit of difference in the world that can be distinguished, represented, assigned weight, and brought into comparison, and that may be extracted, connected, transmitted, and written back within a concrete relational field, thereby changing an object's position, eligibility, risk, evaluation, responsibility, action, or consequences.
The “unit of difference” meant here is not a fixed entity that exists independently of relational fields, nodes, and relational chains. Things, bodies, relations, memories, institutions, languages, spaces, resources, records, and actions in the world may all contain differences that can be distinguished. In a person, one can identify age, the body, language, income, debt, medical history, household registration, kinship relations, local belonging, platform records, and past experiences; family relations contain property, care relations, kinship order, past obligations left unfulfilled, and local reputation; local society contains accents, place names, roads, schools, customs, public memory, and structures of recognition. These differences do not naturally all become weighty at once, but they may be extracted in a particular event and enter relations and consequences.
What makes a differentiating unit is not a name or category in itself, but whether a difference can be identified, represented, compared, assigned weight, and change consequences within relations. Whether one has local household registration, whether social insurance is continuous, whether a contract is compliant, whether medical records exist, and whether there are records of cash gifts are all differences. Only when they are extracted in school admission, job seeking, compensation, borrowing, marriage or romantic relationships, care, exclusion, or the allocation of responsibility do they enter social consequences as differentiating units.
II. Potential States, Relationality, and the Empirical Levels of Representation
The differentiating unit first requires us to distinguish three things: whether a difference meets the conditions for becoming a differentiating unit, whether it has already been extracted in the current relational field, and whether it has already acquired weight and connected to consequences. Units of difference in the world that can be distinguished, represented, assigned weight, and brought into comparison, and that may change positions or consequences through extraction, connection, transmission, and write-back, belong to what this chapter calls differentiating units. The fact that one has not yet been extracted in a particular event does not mean that the difference does not exist, or that it has lost its potential as a differentiating unit. Only when specific people or procedures actually extract and use that difference, making it change qualifications, responsibilities, risks, evaluations, actions, or other consequences, can we further judge that it has taken effect in the current relational field. Analysts must therefore explain separately a differentiating unit's potential state, its state of extraction in the present, and its state of actually taking effect; they must not compress all three into the same moment.
Differentiating units have potentiality and relationality. A medical history may not become weighty in ordinary conversation, yet it may change a position when it enters a hospital, insurance, marriage or romantic relationships, or labor arrangements. Household registration may not become weighty in everyday chat, yet in procedures for school admission, obtaining local household registration, buying a home, or applying for benefits, it may change qualifications or become weighty. An accent is sometimes only a difference in sound, but it may become weighty in local recognition, exclusion, trust, or judgments of identity. Whether a difference produces consequences depends on which nodes extract it, which relational chains transmit it, and which specific people and procedures within institutions, platforms, schools, hospitals, or state service windows—as well as which family members, market participants, or members of local communities—recognize, amplify, downweight, or reject it. The same difference can acquire different weights in different relational fields; it may also temporarily cease to have an effect or reappear.
When the same person enters a hospital, a school, a labor relation, or a family event, the differences extracted from them, and the weights those differences receive, will vary. In a hospital, medical history, health insurance, ability to pay, and a family member's signature may become weighty; on entering the school system, household registration, student status, proof of residence, and the parents' social insurance may become weighty. Labor disputes bring out contracts, wage records, evidence of workplace injury, and coworker testimony; in family events, care responsibilities, past obligations left unfulfilled, kinship relations, and local reputation may become weighty.
A difference may have several forms of representation at the same time. Income can be represented in a wage statement, a money-transfer record, a family's oral account, or proof of debt; residence can be represented in a lease, a community statement, school documents, or neighbors' observations; bodily conditions can remain in pain, medical records, leave records, and care arrangements. These representations are not interchangeable, nor do they automatically have the same operative weight merely because they all point to the same fact of life. In identifying a difference, analysts must preserve distinctions between forms of representation, because different forms do not connect to the same nodes, scopes of authority, and consequences. Even where rules in a particular procedure may allow some materials to replace one another, that substitutability is conferred by the specific procedural rules; it is not a universal interchangeability inherent in the fact of life itself.
At the same time, the same experience may also be represented differently by the person concerned, family members, staff, and those who record it. Conflicting versions do not cancel the reality of the experience, but they require us to distinguish their sources, when they were formed, the occasions on which they are used, and the authority to adopt them. When specific people record and pass on the conflict and use it to change trust, procedures, or responsibility, it is these inconsistent representations and the way they are handled that enter into consequences, not an abstract “dispute” that acts on its own.
When differences are formatted in the course of handling them, some content is retained and some is downweighted. A form can record an address, income, and dates, yet may not record care burdens, waiting costs, and local support; an oral account can preserve the continuity of an experience, yet may lack the form a procedure requires. Analysis through differentiating units must recognize both the need for a form that can be processed and the omissions caused by that format; it cannot judge facts of life that did not enter a field on a form to be nonexistent. Correspondingly, the original experience, the result of its handling, and later explanation must also be kept separate. Pain, residence, and care first exist as facts of life; diagnosis, registration, and evaluation are ways specific people handle them under rules; later analysts then put forward differentiating units on the basis of these materials. Treating the result of handling as an unprocessed original fact overlooks how the preceding node helped shape the form of the difference.
The scope of a difference must also be made clear. Listing someone's “family background” as a differentiating unit is still too coarse. One must go on to ask who provides care, which property or debt has been extracted, which memory is recognized in a dispute, and which kinship position has changed the available exits. Listing “locality” as a differentiating unit does not complete the analysis either: local reputation, address, testimony from acquaintances, travel distance, and local rules may each enter different consequences. A name can be an entry point; it cannot replace breaking the category down. But this process is not endless: the level of detail of a differentiating unit should be governed by whether it can explain real consequences. If finer distinctions only add names without changing judgments about material, positions of responsibility, or the continuation of relations, there is no need to go further. But when stopping would conceal who handled something, which difference was downweighted, or which pathway was interrupted, finer distinctions cannot be omitted. The choice of scale is therefore constrained jointly by the question and the material, rather than arbitrarily decided by an interest in classification.
III. Mechanisms of Operation: Weight Assignment, Continuation, Positions of Responsibility, and Receding Effects
The existence of a difference does not mean that it is already operative. The world contains many differences involving the body, age, language, region, wealth, labor, institutional status, medical history, debt, kinship, rites and customs, education, and memory; they do not all naturally produce social consequences. A difference must first be distinguishable, and it must also acquire a representation. Within institutions, differences usually appear as credentials, files, numbers, documents, and records; in local society, they may appear as reputation, forms of address, wedding and funeral exchanges, seating order, and memories among relatives; in bodily relations, they may appear as medical records, scars, the capacity to act, or age-related conditions. Representation is not decoration. It bears on whether a difference can be identified, transmitted, and extracted.
Even when differences have acquired representations, concrete relations do not extract every difference in a person. A school may extract household registration, social insurance, proof of residence, age, and student-status documents; an employer may extract contracts, attendance, performance, bodily condition, and degree of compliance; local society may extract cash-gift exchanges, kinship relations, past relations, reputation, and social respectability; a court may extract evidence, written acknowledgments of debt, money transfers, judgments, and enforcement records.
Differentiating units that have been extracted acquire different weights. Weight is not a natural property of a differentiating unit, but the operative weight formed within concrete relational fields and relations: in school admission, continuity of social insurance may carry more weight than the parents' accents; in workplace injury, proof of a labor relation may carry more weight than everyday popularity; on occasions of cash-gift giving, whether one attended in the past may carry more weight than present income. Differentiating units must also enter comparison, pass through nodes, and connect to consequences along relational chains. Qualifications, credibility, responsibility, available exits, and judgments about receiving help or being excluded can all change a person's position. Household registration can connect to school-admission eligibility, continuity of social insurance to recognition in a city, contracts to labor relations, and medical records to compensation and leave; cash-gift records and accents may connect in different ways to local reputation, closeness, or exclusion. Extraction, weight assignment, and transmission are not a fixed, one-way process: they may intersect or branch, and a downweighted difference may be extracted anew in another relation.
Differentiating units also have temporal conditions. A difference raised today was not necessarily already taking effect in the same way yesterday; nor can a record formed later be retrospectively written into the material available for an earlier judgment. What must be traced is when a difference was preserved, when it was extracted, when it entered comparison, and when specific people connected it to qualifications, responsibilities, risks, or arrangements for action. Once temporal order is flattened, the relaying that actually changed consequences in the event disappears. Differences appearing at the same time have not necessarily been connected to one another. Wages and rent, medical history and responsibility for workplace injury, income and household registration can coexist in one event. Only when specific people or procedures use one of them to change qualifications, deadlines, risks, responsibilities, or available exits, leaving material and intermediary links, is there sufficient basis to judge that it has entered a causal linkage. Statistical correlation, similar names, and co-occurrence can suggest where to investigate, but they cannot replace carriers, acts of handling, and relational chains.
Within the same event, some differentiating units can substitute for one another, while others can only supplement one another. The fact of residence may be jointly supported by a lease, payments, a community statement, and school records, but whether any one can replace another depends on procedural rules; bodily pain and medical records can corroborate one another, yet they do not have equal force in every judgment of responsibility. Analysts must identify the boundaries between substitution, supplementation, and non-interchangeability. When materials conflict, subsequent handling changes as well. When facts of residence do not match procedural requirements, family memories contradict official files, or local recognition and state registration cannot be mutually recognized, the concrete participants must select material, explain their authority, and bear the consequences of their handling. Analysis should not rush to eliminate the conflict, nor write “conflict” itself as an actor. It must preserve the sources on each side, the positions at which materials are adopted, and the results of handling them.
Differentiating units also concern the allocation of responsibility. The person who raises a difference, the person who preserves the material, the person who assigns operative weight, the person who refuses recognition, and the person who bears the consequences may occupy different positions. Some only supply material; some have the authority to adopt it; some can initiate supplementation or correction. Writing a difference as “caused by the institution” or “caused by the family” conceals the differences between these positions. Making differentiating units concrete allows responsibility to be pursued through relational chains rather than handed over to abstract names. In judging responsibility, seeing a difference and having the capacity to correct it remain two different things. Personnel at the end of a process may know that the material is problematic yet have no authority to alter the format; those in higher positions may hold the power of correction without directly facing the person concerned. The weight of a differentiating unit can indicate where the problem lies, but responsibility must be pursued further through knowledge, authority, choices, and consequences.
Effects that run in the opposite direction within relations cannot be ignored either. Family support may reduce debt pressure while increasing long-term care responsibilities; local reputation can make testimony available, but it may also lead to monitoring and exclusion; institutional records can open a path to qualification, but they may also cause an erroneous determination to be extracted repeatedly later. Differentiating units have no fixed direction toward good or evil; their consequences must be explained through concrete connections. The receding of a differentiating unit also requires conditions. After a debt has been repaid, the current payment obligation may end, yet the associated credit records, family borrowing, and local evaluations may not disappear at the same time; when a qualification is corrected, an earlier refusal may still leave losses of time and opportunity. To explain that a difference no longer takes effect, one must separately check its direct consequences, how it remains preserved, and whether it is extracted again later.
IV. How Differentiating Units Differ from Ordinary Perspectives and Comparisons
An ordinary perspective can suggest where to look; differentiating units require us to go on to explain which concrete differences are operating and how they enter into consequences. Looking at a child's school admission from the perspective of the state brings household registration, school places, and administrative procedures into view. Yet whether the child can enter school must still be traced to how continuity of social insurance, rental filing, explanations at the service window, document deadlines, and community certificates operate within relations. Interpreters who adopt class theory can extract labor domination, risk shifting, and exit costs in workplace injury. Whether a contract exists, whether surveillance footage has been preserved, whether coworkers testify, whether medical records are complete, and whether the family can hold out until the assessment result arrives must still be traced one by one, because they change how compensation ultimately takes shape. A cultural perspective can point to rites and customs and local evaluations. Estrangement caused by a cash gift still requires tracing how past exchanges, kinship relations, and responsibilities for mutual aid enter into consequences. A perspective turns the field of vision in one direction; it cannot replace tracing the mechanisms of concrete consequences.
Differentiating units cannot be treated as permanent variables either. Today's household-registration documents may become weighty in a school-admission event, yet be merely background in tomorrow's dispute among relatives; the same medical record serves different functions in medical judgment, workplace-injury determination, and family care. That weight changes with relations does not mean that judgment is arbitrary. It requires a clear account of why a difference becomes weighty and of the nodes that take up its consequences.
In comparing different explanations, differentiating units provide common units for tracing, not common conclusions. Two explanations can begin with the same wage documents yet reach different judgments about their connections to housing, care, the nation, or law. What matters in comparison is not who lists more differences, but who can explain extraction, weight assignment, transfer, and consequences, and who can accept limits on the scope of their explanation imposed by contrary material. When several explanations extract the same difference, that does not justify merging them into the same theory. Economic analysis, legal judgment, and family research may all be concerned with income, yet ask different questions about material, responsibility, and consequences. Differentiating units make comparison among them possible, while still requiring that their respective scopes be preserved to prevent the formation of a new total image encompassing all research.
V. Judging Effectiveness and the Boundaries of Empirical Evidence
That everything can be differentiated does not mean that everything is equally important. If every difference is listed as important, analysis loses its capacity for judgment. The tables and chairs, dust, lighting, paper, sounds, and smells in a room can all be distinguished, yet they may not enter into the consequences of a child's school admission. Whether a difference matters depends on whether it has been extracted, assigned weight, brought into comparison, and connected to consequences. Rental contracts, continuity of social insurance, proof of residence, and school places may become weighty in school admission; surveillance footage, medical records, coworker testimony, and labor contracts may become weighty in workplace injury. The same difference may become weighty in different relational fields, or it may be unimportant. Whether a judgment can be made still depends on which differences acquire weight in the concrete relational field and what consequences they connect to.
Differentiating units can be downweighted, occluded, or suspended without thereby departing from reality. A fact of residence rejected by a school may still be preserved by the family; pain ignored by an employer may remain in the body and in medical records; local memories downweighted by a public narrative may still be transmitted among relatives and in commemorations. Analysts must distinguish between a difference not being extracted, being extracted without being assigned weight, and losing its effect during transfer after being assigned weight. These three situations produce different consequences and require different forms of correction. Differentiating units can also help analysts distinguish the nonexistence of a fact from its nonrecognition: residence, injury, care, or shared life may actually have occurred, yet fail to enter a procedure for lack of the required format, witnesses, or authority to retrieve the material. In that case, one cannot say that the difference does not exist; one can only say that it has not yet acquired the force required by the current handling. Conversely, recognition of a record does not guarantee that its content fully corresponds to the facts of life. Both directions require checking.
Connecting difference, representation, extraction, weight assignment, and consequences is meant to preserve the conditions under which each link holds. A fact of life may exist without yet being represented; material may be received without acquiring operative weight; a determination may be made without being transmitted to subsequent nodes. The establishment of any one link cannot automatically prove the others. Precisely because interruption is possible at every point, the concept of the differentiating unit can preserve a place for counterexamples, the unknown, and correction. Whether a difference can enter a causal explanation does not depend on whether the concept makes sense, but on how carriers preserve it, through what material it is represented, at which node it is extracted and assigned weight, and how it ultimately changes a concrete position. If a key connection is missing, at most one can propose a possibility that remains to be investigated. New material can fill a prior gap in knowledge, but it cannot be retrospectively written in as evidence already available at the time of an earlier judgment.
The adoption of material on one occasion can prove only that it acquired force in the setting of particular personnel, deadlines, and procedures; failure to find evidence that it took effect likewise permits only a limited conclusion of insufficient material, an interrupted pathway, or correlation alone being visible at present. To propose a more stable pattern of differentiation, one must also compare whether other events exhibit similar extraction, recognition, and consequences. If new material changes a judgment, one should preserve the evidential boundaries of the original conclusion, then narrow its scope, reduce its operative weight, or reconstruct the intermediate relations; one must not retrospectively recognize the unknown as causation already established. Absence, silence, or refusal sometimes produces consequences within concrete relations, but the fact that a result has appeared does not justify inferring a single attitude behind it. Failure to submit material, attend, or respond may arise from lack of knowledge, insufficient time before a deadline, lack of authority, bodily limitations, relational estrangement, or deliberate refusal. Only when specific people or procedures record and interpret that absence and connect it to consequences do the relevant differences enter the current judgment. The conditions producing the absence must still be kept separate from those who interpret it afterward.
The strength of material is not determined by the neatness of its form or its quantity. Contracts, work schedules, money transfers, coworkers' observations, and actual management can jointly support the existence of a labor relation, yet formal records may also be mere copies of an earlier judgment; an oral account that is difficult to format may instead preserve an account of time and of what people bore, neither of which the procedure recorded. When several materials are used together, one must check how they were formed, whether they are independent of one another, and who preserved and passed them on. Only genuinely complementary sources constitute corroboration; repeated appearances of information from the same source cannot masquerade as stronger evidence. A research question can determine which differences to trace first, but it does not authorize interpreters to exclude material that conflicts with their position. Omissions caused by limits of scope, time, and available information should be explained together with the reasons for selection, the current conclusion, and counterexamples that could change the judgment. If the conclusion is extended from an individual case to the family, locality, institutions, or a larger scope, the relaying and the basis for comparison must be explained separately. Making these boundaries public does not turn the main text into an appendix of materials; it lets readers know where an inference holds and where it can be opened up again.
A difference may occupy different positions when an analysis ends: some have been extracted by specific people and connected to consequences; some have been raised but not recognized; some already have the conditions for preservation but have not yet been triggered in the current event; others are only research hypotheses awaiting support from material. These positions cannot be compressed into a single list of factors, or unproved possibilities will acquire the same operative weight as effects already in force. Only by preserving these distinctions can subsequent research know whether it needs to add material, clarify authority, or adjust the original inference.
VI. Low-Dimensional Compression in the Class Image, Macro-Level Inference, and Changes of Scale
Class is not what society fundamentally is, but an explanatory image produced when a set of real differentiating units undergoes low-dimensional compression, projection, and naming. It is neither an isolated differentiating unit nor a single-point difference. Differences such as labor position, possession of wealth, employment relations, resource buffers, the bearing of risk, the ability to exit, debt, and housing pressure, once compressed within relations of economic domination, form the situations marked out by the class image. When a person's wages are overdue, labor relations, income buffers, rent pressure, the cost of asserting rights, and alternative employment opportunities become weighty together. Class theory subjects these real and weighty differences to low-dimensional compression, projection, and naming, thereby forming the class image. Yet the weight of the differences themselves cannot prove the existence of a social entity called class.
The class image formed in this way must still be brought back by interpreters into concrete relational fields and subjected to calibration against other differentiating units and counterexamples. Limited explanatory capacity comes from real differentiating units and the processes of extracting, assigning weight to, connecting, and compressing them; the class image is only a low-dimensional output and cannot on that account be passed off as what society fundamentally is. But this calibration does not mean that the class image can be preserved as a safe, ordinary, limited explanation. Class theory organizes differences into a total image from the outset; it is not an ordinary concept that was originally safe and only later lost its boundaries. The fact that it includes the extraction and compression of real differentiating units does not justify rewriting it as an ordinary entry point exempt from questioning of its structure.
When differentiating units enter macro-level judgment, a single difference must not be elevated into a permanent main axis, nor must all differences be placed side by side without distinction. If similar extraction, weight assignment, and ways of taking up and sustaining differences recur across multiple events, they can support a judgment of stable relations within a particular regional, temporal, and institutional scope. If small differences change consequences through similar people, procedures, and relations across multiple events, they can also support larger structural judgments. But obstruction experienced by an individual family cannot directly establish that an entire region follows the same pathway; similar results in several places still require comparison of rules, resources, and implementation.
Macro-level conclusions must retain local material, the dominant pull, exceptions, and new events that may change the ordering; they cannot automatically expand on the strength of the number of cases or the name of a concept. Macro-level names, in turn, must remain constrained by region, period, sources of material, and exceptions. Such inference must also answer whether similar results arise through the same relay process: if the format of qualifications causes obstruction in one place, while a shortage of resources causes it in another, apparently identical obstruction does not support a single pathway. Conversely, even when results differ, a structural effect with a bounded scope may still be identified if the same rules repeatedly change action space under different conditions for bearing consequences. The macro level is not a matter of smoothing away differences between cases, but of comparing how mechanisms persist or change direction under differing conditions.
Comparison and changes of scale must first establish whether the units are commensurate. An individual's monthly income cannot be placed directly alongside a family's total assets, nor can one medical record be treated as being on the same level as years of local reputation. When differences in residence move from a family's everyday life into eligibility within the school system or state registration, content from that life undergoes retention and omission. Analysts should state whether they are comparing time, qualifications, risks, the capacity to bear burdens, or relational positions, and explain what content can still be mutually recognized across scales and what holds only within the original relation. Moving from local differences to larger relations cannot omit relaying. Similar wage pressures in multiple regions initially create only a question for comparison. To propose stable relations across regions, one must also investigate similar rules, enterprise arrangements, document formats, resource allocation, and connections that can be repeatedly checked against material from events. Local material and macro-level theory are not two unrelated levels: macro-level judgments arise from local relations being continually taken up and carried forward, not from the automatic enlargement of the name “differentiating unit.” The importance of a differentiating unit is determined by its actual effects and the range of its continuation, not by the size of the scale; it therefore does not become the sole main axis merely by entering macro-level discourse.
VII. Not Becoming a New Main Axis, and Responsibility for the Boundaries of Material Selection
Listing differentiating units and then using the list to explain everything merely substitutes a new plane; it does not make society more three-dimensional. Without weight, the listed differences become only fragments placed side by side; apart from a relational field, one cannot judge which differences will be extracted. Differentiating units alone cannot explain how multiple differences jointly form an actual position. The entirety of possible differences not yet extracted is not equivalent to differentiating units; but a difference that meets the conditions for becoming a differentiating unit does not lose its potential as one merely because it has not yet been extracted in a particular event. Elevating differentiating units into a new main axis to replace class, the state, the market, the nation, culture, or the family merely repeats the error of flattening reality with an overarching concept.
Analysis through differentiating units establishes the conditions under which differences acquire social efficacy, not an infinite table of attributes. Details of life acquire corresponding operative weight in the current question only by entering consequences through carriers, material, procedures, and time. Conditions that have not yet become apparent may still really exist, but they cannot be written in advance as principal causes already in effect. This enables differentiating units to open up what grand concepts occlude without becoming a new total image that handles every event. Differences that have not entered the present event are not thereby abolished by the theory. They may remain in bodies, family memories, local relations, or institutional materials, awaiting extraction anew at another node. Distinguishing real existence from current efficacy is precisely the condition needed to avoid mistaking the window of analysis for the boundary of social existence.
A differentiating unit is not a social atom, nor an ordinary factor waiting to be used in every relational field; still less can it preserve, interpret, or write back effects on its own. The difference it denotes manifests social efficacy in a given relational field only through concrete carriers, relational pathways, and handling procedures. This boundary both preserves the reference to real differences and prevents differentiating units from being fixed as the smallest units of a replacement for the class plane.
The theoretical function of differentiating units is to return selection, representation, comparison, and causal linkage to interpreters. Without material, carriers, and pathways to consequences sufficient to support the relation at issue, conceptual possibility cannot acquire the standing of an empirical conclusion. If new material fills in the connections, differences previously excluded may re-enter. This boundary, which can open and close, enables theory to judge without filling the unknown with an illusion of completeness. At the same time, the extraction of differentiating units is not without limits. Where medical histories, family conflicts, or religious practices involve the private sphere, whether material can enter a judgment is constrained by the consent of those concerned, procedural authority, and existing responsibilities within relations. Completeness of research does not suffice to justify indiscriminate retrieval; the manner in which material is obtained changes trust and risk, and also becomes part of the assessment of consequences and the scope of judgment.
VIII. Conceptual Tools, the Reality of Relations and Groups, and Explanatory Boundaries
Concepts are not useless. Class, the state, the market, culture, the family, and the nation can all compress complex reality into an outline that can be articulated, enabling people to judge and argue; once treated as what society fundamentally is, they also occlude the differences beneath them. Interpreters who adopt class theory may occlude the differences among wages, debt, housing, risk, and the ability to exit. The concept of the state may occlude differences among household registration, service windows, legal texts, administrative interpretation, and fiscal allocation; the concept of culture may occlude differences among rites and customs, shame, gender expectations, local reputation, and long-standing habits. Concepts must return to concrete relational fields and be calibrated against differentiating units and the processes through which they enter consequences. Only then will concepts not be passed off as what society fundamentally is. General concepts can thereby leave room to be unfolded again, but the structure of class theory organizes differences into a total image from the outset; it is not an ordinary concept that was originally safe and only later lost its boundaries.
Some differences can be fully explained only within relations. The deadlines of creditors and debtors, what caregivers and those receiving care bear, and recognition between members and communities cannot be attached solely to one individual. Analysis must identify at once the positions of participation, the direction of effects, and the actual consequences. This explains how differences are constituted relationally; it does not establish a separate, pre-existing, fixed category of “relational differentiating units.” Differences at the group level, such as shared memories and shared harms, likewise persist only through real carriers. Family narratives, school education, rituals, recorders, witnesses, archives, and members' practices may jointly preserve and bring forward the relevant material again. Names of nations, families, or organizations can summarize these relations, but they cannot themselves remember, judge, or take responsibility. Reality at the group level comes from an ongoing process of taking up and carrying forward these relations and their contents, not from a name acquiring a unified personality.
That concepts are not what society fundamentally is does not mean that they are confined to researchers' language. Institutional classifications, forms of address within families, and names of communities, when used by specific people, can actually change qualifications, trust, and responsibility. What must be ruled out is writing a name itself as a pre-existing entity or actor. The same holds for differentiating units: they point to how real differences enter consequences, but they cannot operate on their own apart from people, material, relations, and procedures.