The first version of the community repair station's spreadsheet had only five columns: time received, location, problem category, person responsible, and completion status. There were no credit scores, no predictive models, and nobody intended to build a cage. It simply replaced the stack of notes behind the counter whose dates were gradually becoming illegible.
This is a fictional institutional archive. Residents can report problems with lighting, door locks, and handrails in common areas. The repair station initially arranged its work by slips of paper: whoever took a request wrote the location on paper and then found someone free to handle it. When volume was low, this method could function; once requests multiplied, omissions appeared. Some places were visited repeatedly; others went permanently unanswered.
The staff decided to use a shared spreadsheet. Each request was logged with its time received, location, problem category, person responsible, and completion status. The change was simple, yet it made certain previously unanswerable questions answerable: which requests had not yet been assigned, who was working on what, and where additional materials were needed. The spreadsheet did not solve every repair difficulty, but it gave the collaboration a more stable and visible form.
This chapter cuts vertically through the institution along the spreadsheet's successive versions: when five columns began to distribute opportunities, when completion status became performance, and when performance in turn changed what the staff worked on. We shall not presume that digitalization necessarily brings control, nor that paper-based collaboration is naturally more humane. The real object of inquiry is not some malicious designer, but how recording, allocation, evaluation, and responsibility are joined into a closed loop through one reasonable column addition after another.
RC describes social order as a coordinable, operable collection of certainties formed through the interaction of subjects. The spreadsheet is worth discussing because it shows how such a collection enters everyday work: certain matters come to be stably named, certain states come to be mutually recognized, and certain actions thereby acquire an order of priority. Order does not always appear as grand institutional declarations; it may also begin with "how should this column be filled in."
The Spreadsheet First Solved a Real Problem
Repair workers cannot remember every request at all times, and residents cannot continuously track the actions of every staff member. Without a shared record, the costs of coordination shift into repeated asking, repeated explaining, and mutual guessing. A record keeps communication that has already taken place in a place where it can be consulted again, and thereby saves these costs.
A unified format has a further effect: it lets the work of different people connect. After the person receiving requests leaves, the next one on duty can still understand where matters stand; if a repair worker temporarily cannot continue, others can see which steps have been completed. Standardization in this sense increases the organization's capacity to absorb changes in personnel.
If we mount a critique solely from the premise that "standards crush complex life into grid cells," we will miss these gains. Grid cells do omit detail, but omission can also keep certain necessary information from drowning in scattered narrative. The question is not whether the spreadsheet completely reproduces reality — it cannot — but which important distinctions it preserves for a particular task.
RC's principle of theoretical dimensional reduction does not deny the value of finite expression. It holds that theory is an interpretive system under a finite horizon, with boundaries of its own representation. Applying this principle to recording tools, we can require that a spreadsheet suit the coordination task at hand without requiring that it contain everything. A good tool needs clearly drawn boundaries, not an endless inflation of content.
When a resident reports that "the corridor light is out," the spreadsheet can first classify it as a lighting problem, so that someone with the relevant competence can be assigned. As for what this particular light means to different residents, that may have to be learned in the course of handling it. The initial classification is a path into the work; it need not become a final explanation of the whole situation.
As long as a tool's purpose remains clear, participants can use it while pointing out what it leaves out. Record and field continue to correct each other. People will not conclude that a problem does not exist merely because the spreadsheet lacks a column for it; nor will they treat someone who offers information beyond the spreadsheet as refusing to cooperate.
From Recording Facts to Distributing Opportunities
Now the repair station decides to order its handling by time of registration. This rule reduces arbitrariness and lets residents know roughly how their requests will enter the queue. The spreadsheet thereby ceases to be merely a store of information; it begins to take part in distributing scarce handling time.
The new use brings new boundary problems. If a late-arriving request concerns a passageway in constant use, while an earlier one is merely cosmetic repair, should order still follow time strictly? If adjustment is allowed, who has the authority to decide, and on what basis? If adjustment is not allowed, a fixed order may leave certain more pressing needs waiting indefinitely. These questions cannot be closed simply by invoking "fairness means treating everyone the same." Identical sequencing may realize one kind of fairness without fitting all consequences; ordering by urgency can respond to actual need, yet may also introduce individual judgments that are difficult to review. Institutional design faces a weighing of reasons against one another, not one principle annihilating all the rest.
The repair station therefore adds an "urgency" column. This change attempts to remedy the previous deficiency, but it also hands new evaluation power to whoever fills it in. Whether a given request is marked ordinary or urgent will affect how long others wait. Residents may begin to compete for higher ratings, and staff will need to explain why they made a certain classification.
RC's account of power hierarchy here acquires concrete meaning: the position able to define categories and decide ratings holds the capacity to shape the paths of others' actions. This is not necessarily illegitimate power. For an organization to accomplish its work, someone must make judgments. But the necessity of power does not automatically answer the question of how it should be constrained.
We can observe a continuous shift: at first, the spreadsheet let everyone know what had happened; later, it helped decide what to do first; later still, the rules for filling it in themselves became something participants had to compete over and explain. The tool did not turn into an institution at some single moment; it gradually acquired institutional meaning through the accumulation of consequences.
A Single Status Can Conceal Many Outcomes
The repair station next wants to know whether work has been completed. It keeps three statuses in the spreadsheet: pending, in progress, and completed. This is very helpful for basic coordination, yet it requires answering what "completed" actually means.
If a repair worker has been to the site, does that count as completion? If a part was replaced but the problem still occasionally recurs, does that count? If the resident says the facility is usable but the worker knows a follow-up inspection is still needed, how should that be recorded? A single status looks concise, yet behind it may lie quite different actual outcomes.
If the spreadsheet mainly serves to remind people of follow-up work, the organization may be willing to keep complex statuses. If "completed" is also used to measure efficiency, participants will pay closer attention to which events can enter that column. The same definition begins to serve simultaneously the linking of work and the comparison of performance, and these two uses are not always compatible.
Suppose the repair station stipulates that an item originally marked completed is reopened only if the resident submits a request again. This rule reduces the recording burden of staff-initiated checking, but it may transfer the work of discovering problems, re-explaining them, and persisting in follow-up onto residents. A rising completion rate in the reports does not automatically mean that recurring problems in fact diminished. The point here is not to suspect every completion status of deceit, but to see how it is produced. A record can be interpreted only in conjunction with its rules of use. The absence of a repeat request may mean the problem is solved; it may also mean the resident considers raising it again pointless, or has for the moment no opportunity to pursue the matter. Which interpretation fits better requires supplementary material.
RC uses the bidirectional representation of language to show that expression aids cognition while also producing a loss of experience. "Completed" is precisely such a compression. It allows the organization to grasp work progress quickly, while possibly making unresolved follow-up conditions disappear. The reasonable course is not to ban the word, but to keep its meaning, its evidential basis, and the conditions for reopening it stateable.
When Evaluation Begins to Change the Work Being Evaluated
The repair station subsequently uses each staff member's monthly completion count to arrange rewards. Staff now know that records not only show what has been done but also affect what they receive. They accordingly have reason to take this consequence into account in their daily choices.
In our stipulated scenario, some requests end easily while others require repeated inspections; some interventions rapidly increase the completion count, while preventive work may reduce future breakdowns without immediately producing a comparable number of records. If rewards recognize only the first kind of result, the second may struggle to receive adequate time. This inference follows from the rules explicitly posited in the scenario; it is not an empirical conclusion about all performance regimes.
Even if no staff member falsifies records, the mode of measurement can still change the composition of the work. They may favor items that are easy to count, defer complicated problems, or split tasks that could have been handled together into multiple entries. We cannot conclude from a change in completion counts alone that such behavior has occurred, but we should understand which possibilities requiring further observation the rules have created.
Managers, too, may be affected by their own indicators. If the level above asks only about total completions, the manager has all the more reason to fix attention on totals. Residents' repeated explanations, preventive inspections in the field, and the differing difficulty of different problems may not enter reports in equally legible form. Locally reasonable reporting behavior may, level by level, shrink the work the organization is able to see.
RC's "evaluation alienation" leads us to a key question: does evaluation still help uncover value, or is it gradually leaving participants able to act only around the value that evaluation recognizes? There is no line between the two that can be drawn from the name of a tool alone. It depends on whether, once deviations are discovered, the institution is willing to reinterpret and adjust its standards.
If a manager who sees completion counts rise while repeat complaints also rise is willing to return to the field and examine the relation between the two, the indicator remains a cognitive tool. If he concludes that the reports already prove improvement, so that all contrary experience can only be interpreted as residents demanding too much, then the indicator may pass from a representation requiring validation into a warrant that refuses validation. This is a closed loop: work according to the standard, produce data conforming to the standard, then use those data to prove that the standard needs no revision. The stability of such a loop need not derive from anyone's intentional deception; it can also arise from the organization no longer preserving any channel of feedback outside the standard.
"Everyone Agrees" — Agrees to What, Exactly?
Facing the new evaluation method, the manager convenes a meeting and asks whether anyone objects. No one speaks, and the minutes record "unanimous agreement." What can this record show? At minimum, that no objection was recorded at this meeting. But whether it can further prove that everyone endorses all the rules requires additional conditions.
Staff may sincerely believe the new method workable, or may not yet understand its consequences; they may feel the problems are not serious enough, or may not know whether raising objections would afterwards mark them as unwilling to work. The meaning of silence is not singular. We can neither automatically count it as evidence of oppression nor take it as sufficient proof that no disagreement exists.
To strengthen the interpretive force of assent, participants can first be shown how the standard will handle concrete examples, told which parts can be adjusted, and brought back into discussion after a trial period. In this way people's agreement gradually connects with consequences they can undergo, rather than resting on an abstract name.
When RC discusses conceptual consensus, it links the predictive power and practical value of interpretive frameworks with intersubjective comparison. In institutional contexts this reminds us that consensus can be tested through practice, and cannot be regarded as permanently completed by a single formal expression. Agreeing the first time to use the spreadsheet is not the same as thereafter agreeing to every one of its uses.
When a limited authorization is interpreted as an endorsement without boundaries, the subsequent experience of participants easily loses its place. Staff may accept the recording of work progress without accepting that the entirety of contribution be judged by entry counts; residents may accept the order of registration while hoping that urgent situations receive separate handling. Such disagreements need not mean they oppose the collaboration; they may be helping to delimit its scope.
Enabling an institution to hear such distinctions is more valuable than requiring everyone to constantly reaffirm their support for it. If consensus is to endure, it must be able to state what it currently covers, what it does not yet cover, and how new questions enter the next round of discussion.
Exceptions Are Not Merely Trouble, Nor Are They Automatically Right
A staff member at the repair station finds that one request, placed into any of the existing categories, cannot accurately convey what its handling requires. He can pick a category at random so the process continues; or he can propose adding an explanation, but that will take up coordination time. How an institution responds to such situations displays its attitude toward exceptions.
If every exception is treated as an individual problem obstructing efficiency, the spreadsheet will find it increasingly difficult to learn from the field. But if every dissenting opinion immediately changed the overall classification, others might be unable to form stable expectations. Corrigibility does not mean the absence of stability, nor that whoever raises an inconvenience is necessarily right.
We need to distinguish case-by-case handling from rule revision. A given request can first be completed with a supplementary note, while recording why the existing classification was insufficient; if similar cases recur, the classification can then be examined for adjustment. This ordering of steps allows the organization to keep operating while accumulating knowledge about its own boundaries.
Some exceptions may also reflect particular interests. If residents familiar with the manager always obtain higher urgency ratings, the vaunted flexibility erodes the common rules. In analyzing exceptions, one should ask whether the reasons can be stated to the other participants, whether they accept comparison with like cases, and whether the decider bears a duty of explanation.
RC's limited order does not demand that the same form be preserved forever; it concerns whether contradictory pressures can be absorbed and processed. Exceptions provide a concrete test: can the organization transform difference into discussable information, or does it merely classify difference as noise obstructing work? The latter response may reduce contention for a time, yet it also reduces the material required for improvement. One can therefore neither abolish exceptions in the name of institutional stability nor abolish all comparable standards in the name of respecting individuals. The more sustainable direction is to make explicit which differences require individual handling and which signal rule failure, and to keep both judgments open to review.
Who Bears the Work Outside the Spreadsheet
Any seemingly simple institution may leave a portion of its work outside the formal record. Residents must convert complicated problems into descriptions the spreadsheet can recognize; the front desk must chase down missing information; repair workers must explain why a problem cannot be closed the same day; the person on duty must placate those dissatisfied with the pace. These actions sustain the process, yet they may not produce a single new completion record.
If evaluation ignores these inputs over the long run, the organization may mistakenly suppose they cost nothing. Work that can be counted then receives more recognition, while the coordinating labor that keeps it flowing is assumed to be expandable at will. The master spreadsheet remains clear, while those who actually bear the load find it ever harder to free up time to respond to new problems. Here we can apply RC's distinction concerning how pressure flows. An advantaged position can absorb pressure through mechanisms of responsibility, or it can keep transferring the losses to positions with a smaller available margin. A manager's demand that the front desk "explain things clearly" is not always unreasonable; but if the front desk has no authority to change decisions and no adequate time to handle recurring disputes, the demand may amount only to turning an institutional problem into an individual burden of communication.
Residents, likewise, may bear work that goes unrecorded. If every report of a recurring fault requires filling in all the information from scratch, and the repair station keeps no internal linkage across episodes, then the search costs the organization saves may be shifted onto residents. Some people can keep completing this process; others may drop out earlier. The records that remain therefore need not cover all unmet needs.
Bringing these inputs into evaluation does not require immediately putting a price on every explanation. One can begin by acknowledging that they are part of the work, observing which links repeatedly consume time, and only then deciding how to improve handovers, authority, or information retention. Making invisible costs discussable comes closer to the problem than simply adding a "service attitude" score.
The public character of evaluation means here that those affected by the rules are able to state the costs the rules omit. If only the rule-makers can define which costs are worth counting, then however precise the calculation becomes, the evaluation may continue to tilt in the same direction.
Revising Standards Without Overturning the Whole Collaboration
In the next stage of the thought experiment, the repair station finds that a single completion count cannot adequately indicate the quality of work. It could go to either extreme: keep everything as it is and deny the deviations, or abolish all records and return to an arrangement relying solely on individual memory. In fact, a more limited space of revision also exists.
For instance, the organization can distinguish first-time handling from subsequent re-inspection, letting the completed status retain conditions for reopening; it can set complicated problems apart with separate explanation instead of comparing all entries directly; it can give preventive checks and coordinating work a visible place. The changes listed here are not a scheme already validated for every setting; they show that revision can target specific connections.
The more important change is to separate the use of a tool from the evaluation of the tool. People can go on using the spreadsheet to arrange work while periodically examining whether classification, ordering, and rewards have produced consequences running against the aims. Following the current procedure then no longer equals admitting that the procedure is forever correct.
RC's processual completeness locates the completeness of cognition in the capacity to iterate. For an institution to embody this capacity, it needs to form records of its own judgments: why this standard was adopted, what problem it was meant to solve at the time, which deviations appeared later, and which material prompted the modification. Rule changes should not consist merely in officeholders changing their wording; they should also let participants understand the reasons for the change.
Explaining rule changes also protects stable expectations. If standards can change at any moment without explanation, participants can hardly know how to arrange long-term action; if standards can never change, practical experience can hardly enter into shared judgment. Stating clearly the grounds and the scope of effect of a change can reduce oscillation between these two risks.
RC's sustainable decision-making emphasizes keeping optionality continuously present. At the scale of a repair station, this means that an improvement should, as far as possible, preserve the capacity for later revision, rather than wagering all resources on a single process that cannot be withdrawn. It also means that the costs required by trial runs should not be borne entirely by staff or residents who hold no decision-making power.
After a standard is revised, the comparability of old records still needs to be handled. If the new rule lists re-inspections separately, one cannot directly announce that efficiency has risen or fallen by comparing the two completion totals; the definition has changed, and so has what the numbers denote. Writing this change out clearly can prevent a revision aimed at improving evaluation from becoming instead a source of new misunderstanding. Old records may be kept, but they should be understood together with the criteria in use at the time.
What Determines the Boundaries of Obedience
Through this spreadsheet we can see that obedience has distinct layers. Staff fill in information according to the entry format so that others can link up the work; they handle requests according to the ordering, thereby accepting the current arrangement for allocating resources; being required to raise no questions even after finding the standard failing involves a deeper restriction. The three should not be swallowed whole by an undifferentiated concept of "obedience."
Collaboration requires participants to observe certain established conventions, but the reasonableness of a convention cannot be proven merely by its having been formed. A rule affects others; its makers therefore need to state what task it serves, how adverse consequences are handled, and under what circumstances it should be reopened for discussion. The more binding the enforcement, the less these questions can be waved away with slogans.
When the spreadsheet becomes order, the key is not how many columns it has but how wide a range of consequences it has acquired. A column of data used only to remind a worker to check further has relatively limited significance; if it determines all resources, cancels appeals, and defines participants' overall worth, it has assumed an evaluation power far exceeding its original purpose. Judging whether excessive expansion has occurred requires tracing these consequences one by one. Conversely, that an institution touches far-reaching consequences does not mean it necessarily constitutes a cage from which there is no escape. Whether the grounds of decision can be stated, whether disputes receive appropriate independent review, whether misjudgments can be remedied, and whether those affected can take part in rule revision — each of these alters the conclusion. A critique of obedience should be able to recognize the real differences between institutions.
The thought experiment of this chapter has not proven some universal social trend. What it offers is a path of analysis: start from the real problem a tool solves, track how it connects resources and behavior, examine whose inputs it omits, and then observe whether deviation can in turn change the standard. This path should also be applied to the critic's own explanations, to prevent critique from admitting only the material that proves control.
Letting Order Preserve Answers from the Field
In the end, the repair station still needs a spreadsheet. Residents need to know whether their requests have been received, staff need to know what to do next, and managers need to understand whether resources suffice. Abolishing the shared representation would not automatically increase anyone's available margin. But this spreadsheet should not be a one-way device that permits only reality to bend toward it. Residents' experiences, staff's difficulties, recurring faults, and uncounted inputs should all have the chance to change how it is used. The common order continues its becoming through these responses; it is not already finished at the end of the first design.
A spreadsheet can help people collaborate because it gives finite experience a shape that can be linked to others. It can confine people because this shape has been granted the power to refuse other experience. We need to protect the former function while continually checking for the latter transformation.
To judge whether an institution deserves trust, one cannot ask only whether it makes everyone fill out a tidy row. One should also ask: when a life cannot fit into that row, does the institution have the capacity to hear the reasons, and to make corrections together with those who bear the consequences?