The first round of contention runs aground on evidence. The opponents demand that the constructor prove on what grounds he claims that the bridge connects shared life. We continue the small-town thought experiment of Chapter 1. Residents want to know how the space beside the bridge is being used, so someone records the flow of people crossing it. The first record says the bridge is very crowded in the evening; the second says it is not crowded most of the time. Supporters of the renovation adopt the first record, opponents adopt the second, and each side believes it already possesses the facts. Yet the two statements may not conflict at all. The first records a short interval after the market closes; the second summarizes a longer period. Only by stating the object and scope of the statistics can we judge where the disagreement lies. If materials are first selected according to position, and the selected materials are then called the complete reality, public records themselves become grounds for mutual isolation.
This round of debate concerns the stability of public facts, but it does not require facts to wait for universal recognition before they exist. The point at issue is this: how do factual judgments about common affairs gain support that can traverse individual experience, divergent positions, and change over time? Such support requires observation, but it also requires the public conditions of definition, recording, review, and revision.
First, Give Statements Comparable Objects
"There are many people on the bridge" can express lived experience, yet it is not yet a statistical statement with a clear scope. Person-trips or unique persons, the whole bridge or a particular entrance, the whole day or a certain interval: different objects yield different records.
Imagine the same person crossing back and forth several times in one day. If the question is the number of crossing activities, each passage may deserve to be recorded; if the question is how many different people use the bridge, another way of counting is required. We cannot conclude from the difference between the two results that one side is fabricating; they may be answering different questions.
Definitions determine what enters the picture, and therefore deserve public discussion. But definitions also cannot be exchanged at will according to preference. If users want to use data to support a conclusion, they must show that the definition is relevant to that conclusion. Crossing counts cannot automatically establish all of residents' public-space needs, and the number of people lingering cannot by itself determine traffic arrangements.
RC uses the bidirectional representation of language to illustrate the selectivity of expression. Here the definition is the entrance to common observation, and at the same time the place where omissions arise. Making omissions visible does not abolish the definition; it prevents the definition from masquerading as the totality of the object's properties.
The first layer of stability in public facts comes from statements having relatively determinate objects. People need not endorse the same policies, but they can at least know what this record counts and what it does not, and on that basis check whether conclusions exceed their scope.
Uncertainty Can Be Stated Rather Than Hidden
In real metrology, measurement results are not meaningful because they claim absolute infallibility. Section 3.1.2 of JCGM 100:2008, the Guide to the Expression of Uncertainty in Measurement issued by the Joint Committee for Guides in Metrology, notes that a measurement result is generally an approximation or estimate of the value of the measurand, and that a complete statement must be accompanied by a specification of the corresponding uncertainty. This requirement comes from the field of metrology. This chapter does not transplant its mathematical treatment as a general formula for social surveys; it borrows it only to state a limited fact: acknowledging bounded uncertainty can be part of rigorous statement, and is not equivalent to conceding that arbitrary claims are equally tenable.
Returning to the fictional statistics, the recorders can state which intervals were under-recorded and which positions were blocked from view. If the effect of the omissions cannot be estimated, that limitation should be retained honestly rather than fabricating a precise error. Incomplete data can weaken the strength of conclusions, but it does not require discarding every part that was reliably recorded.
An integer with no stated method is not necessarily more reliable than an estimate that candidly states its range. Precision in the outward form of a number and precision supported by evidence are not the same thing. If the public rewards only expressions free of doubt, recorders may become less willing to expose the boundaries of their methods.
RC's processual completeness supports placing judgment within a process that can be updated. The value of stating uncertainty lies in helping decide what must be supplemented next, not in creating a pocket from which conclusions can be indefinitely deferred. Which parts suffice for action and which parts still require verification should be discussed separately.
Ten Retellings Are Not Ten Observations
A resident sends one record of pedestrian flow to several discussion groups. Soon many people cite the same figure. The figure appears more and more frequently, and it easily creates the impression that it has been confirmed by many parties. But what is added is first of all the number of transmissions, not independent observations. The distinction here does not turn on whether retellers are honest. Everyone can faithfully cite the same document, and this still cannot automatically reduce the omissions that may exist in the original record. Distinct speakers and distinct sources of evidence must be counted separately.
Imagine now that another group records independently. If the two results are close, that adds some support, but we must still check whether the same erroneous definition was used, and whether both omitted the same entrance. Independent collection does not mean that all sources of error are independent. Being able to state the relations among sources is more useful than simply counting how many institutions concur.
RC distinguishes observational consensus from conceptual consensus. In the social analysis of this book, wide acceptance of a claim is a public condition that requires explanation; it cannot be taken directly as proof that multi-level observation has already achieved agreement at the ontological level. Still less can we infer, from the fact that the powerful within a group have declared unanimity, that the observations of other subjects have lost their significance.
The stability of a public fact therefore consists not in the echo growing louder, but in its support withstanding examination under different conditions. Examination has costs and limits of its own, but it provides an increment that transmission itself cannot replace.
Public Records Must Preserve Their Provenance
If residents receive only a beautiful chart, without knowing the time of recording, the definitions, and the history of modification, they can forward the conclusion but can hardly re-examine it. That much material has been made public does not by itself mean that the process of fact formation has gained a public entrance.
In the thought experiment, users should at least be able to know which set of records the chart corresponds to, what criteria were adopted, and which parts were later corrected. In this way a specific objection can be traced back to the relevant material instead of propagating without limit through the entire public controversy.
Publicity also has boundaries. To discuss traffic across the bridge, there is no need to publish all the identifiable personal information of every person who crosses. Allowing appropriate re-examination and unnecessarily exposing individuals are not the same goal. If certain materials cannot be made directly public, the reasons and the alternative means of verification must be stated. This bounded traceability does not promise that every resident can single-handedly redo the entire investigation. People can trust recorders of the relevant competence while retaining channels for raising specific questions, obtaining methodological explanation, and correcting errors. The legitimacy of professional division of labor cannot rest on others never being permitted to ask further.
Facts Do Not Complete Value Choices on Our Behalf
Suppose everyone eventually agrees that pedestrian flow is concentrated in a certain interval. This helps the discussion, but it does not by itself determine how the space beside the bridge should be transformed. Among the goals of faster crossing, preserved space for lingering, and limited expenditure, trade-offs still have to be made.
If supporters say that the numbers prove this is the only thing to do, they extend the scope of factual support into value choice. Data can weaken certain claims about the status quo, yet they may not exclude every alternative. The reasons for choosing ends and means still have to be stated.
Opponents, likewise, cannot declare statistics meaningless merely because the data cannot yield a unique plan. Facts constrain the discussion: if a congestion problem genuinely exists, a plan must respond to it or explain why other goals take precedence here. Value judgment is not a privilege exempt from answering to facts.
RC's theoretical dimensional reduction here constrains both policy interpretation and critical interpretation. An interpretation may propose a useful angle of observation, but it cannot, by virtue of the complete appearance of its conceptual system, package value choices that have not been made as conclusions necessarily derived from the facts.
The Two Closures of the Entrance to Fact
This analysis also contains a dark possibility that must be faced. Since public judgment depends on conditions of definition, recording, and verification, whoever controls these conditions may influence which experiences can attain the status of public facts. This does not require changing what actually happens on the bridge; so long as certain consequences never enter visible records, public discussion can remain structurally skewed. This power runs deeper than merely advancing a prejudice. A person who advances a prejudice may still encounter counterevidence; if the formation, preservation, and consultation of counterevidence are all controlled from the same position, dissenters may be unable even to obtain the materials with which to demonstrate the problem. Within such a relation, "no evidence" must be examined further: is it because the claim lacks support, or because the conditions for obtaining support have been blocked?
But this latter possibility in turn cannot prove any evidence-free accusation. If a claimant says that the inability to find records precisely proves that the records were deliberately destroyed, relevant materials for the destruction must still be produced. A closed official explanation and a closed anti-official explanation can alike treat absence as something that forever supports their own side.
RC's constructive language may also be used as a more radical excuse: since certainty is a secondary construction, controlling the narrative people accept seems equivalent to controlling the facts themselves. This inference skips over the distinction between "social recognition of a person" and "convergence of multi-level observation". Silencing a person does not remove the material consequences he bears, nor does it prove that the other levels of observation are running in accordance with some narrative.
The political consequence of this skipped step deserves to be stated plainly: power may shrink the reality that can be acknowledged to the range it itself supplies, and then require the affected to appeal only within that range. The public qualification of a fact then becomes a license rather than the outcome of common re-examination. Those harmed face not only a particular erroneous conclusion but also a restriction on which experiences deserve to be called evidence.
To criticize this path, it is not enough to say that the theory has been misused by outsiders. Precisely because the theory emphasizes the conditions of observation and the rights of evaluation, it has already revealed that these positions hold a potency that can be abused. It must explain why it opposes certain subjects monopolizing these conditions, and it cannot blur differences of authority with "the unity of subject and object". Common participation does not mean being able to stipulate on equal terms whose experience is valid.
Public facts can also be damaged from the opposite direction. A form of criticism keeps insisting that records have standpoints, definitions involve selection, and measurements carry uncertainty, while refusing to state what evidence deserves provisional acceptance. In the end, every material is excluded for being imperfect, and no new material can change the judgment.
Superficially, this posture keeps its distance from all authority; in effect it may strip people of the ability to distinguish more reliable materials from arbitrary assertion. When public verification loses its meaning, decision-making need not stop, and may instead lean more heavily on identity, volume, and immediate mobilization. Thorough skepticism does not automatically weaken power; sometimes it merely removes the requirement that power justify itself factually.
If a person declares on the one hand that all histories and records are untrustworthy, and on the other demands that others accept his own definitive judgment about the entire situation, he is reserving an exception for his own interpretation. What must be pursued here is not the intensity of the attitude but whether standards of evidence are symmetric.
If RC's theoretical dimensional reduction is interpreted to mean that all theory is useless, even this interpretation itself loses its priority. The more consistent understanding lets the boundary of theory become a reason for continued observation. Finitude demands revision, not the abolition of all comparable judgment. Otherwise the language of critique provides shelter for assertions that can never be re-examined.
Does Correction Make Facts Weaker, or More Reliable
The fictional recorders later discover that a certain interval was double-counted and publish a corrected version. Some conclude from this that the entire prior investigation was a fraud. This conclusion is likewise too strong: the existence of one error shows that the corresponding results need correction; whether it constitutes deliberate deception requires additional evidence about the course of conduct.
Nor is a correction automatically worthy of trust. If the chart is quietly replaced without any statement of what changed, people cannot judge how much the old citations are affected. If a trivial error is acknowledged while examination of the core questions is refused, the correction may supply only the appearance of being responsible.
A meaningful revision should state what was changed, why it was changed, which conclusions are affected, and which are still supported by the original material. In this way the public need not choose between believing everything and discarding everything, and can update its judgments in a targeted way.
Revisions must also reach the places where the old conclusions are used. If planning discussions still cite the old chart, the existence of a new chart on the website does not mean that public judgment has been updated. Between correcting the record and correcting actual decisions, someone must carry out the transmission and the explanation. There are costs here as well. We cannot demand that every minor revision trigger a maximal campaign of explanation, but conclusions with a wide reach require correspondingly accessible correction. The convenience with which old conclusions circulate is not a reason for refusing to bear the work of correction.
The Division of Labor in Fact Formation, and Who Constrains It
Not everyone can carry out the whole of verification, which means that public knowledge inevitably requires a division of labor. Some record, some preserve, some interpret, and some decide on action on the basis of the material. The division of labor can carry public judgment beyond individual capacity, and it also creates dependence on intermediary positions.
If this dependence is dismissed wholesale as blind faith, we cannot explain how people use reliable knowledge they have not personally produced. Requiring every resident to re-investigate the whole bridge is unrealistic, and it may give those with time and resources an even stronger discursive advantage. Complete self-verification is not a freedom everyone can afford. But the division of labor also cannot be used to declare that ordinary participants deserve only to receive results. Recorders can answer questions about method, preservers can state the versions of materials, and interpreters must distinguish what the data support from their own inferences. Each position bearing its own burden of explanation is more feasible than requiring one person to master the entire process.
Nor must these roles be held by different people in every small affair. One person can record and organize the data, but when conclusions have wide reach and controversy is sharp, review must be made to depend less on that person's self-confirmation. Independence is not a matter of job titles but of whether contrary judgments can form under conditions not fully controlled by the original decision-maker. This also shows that professional authority should be neither deified nor abolished merely because professional differences exist. A professional position worth relying on is one whose competence and claims can be examined within the relevant range. A questioner may lack equivalent technical ability and still reasonably demand an explanation of whether the conclusion exceeds the population, the timespan, and the questions the data cover.
Before the Argument Begins, Selection Has Already Occurred
The question of power over public facts arises not only when conclusions are interpreted but also when what is to be investigated is chosen. If the town counts only crossings and does not record lingering, the need for crossing will possess tidier data, while the need for lingering appears mostly as personal testimony. The former may then be called an objective need, the latter a sentimental opinion. We should not assert outright that the crossing survey is malicious. A survey has a limited aim from the start and cannot cover the whole of life. The question is whether the limited aim is acknowledged, and whether its results are extended into the sole yardstick for evaluating all public space.
What has not been measured does not thereby lack a factual component. When a particular resident lingered beside the bridge, and whether the original space was available for use, can be checked against suitable material. Its significance for value must be discussed separately, but it should not lose its standing to enter factual discussion merely because the existing form has no column for it.
If the position that determines the scope of investigation also has the power to declare that only what has already been investigated counts as grounded, it may delimit the visible world before the discussion begins. The dark usage appears here as the mutual support of the entrance to evidence and the entrance to agenda: what has not entered the agenda has no record, and the absence of a record in turn becomes the reason for keeping it off the agenda.
The practical demand of critique is to make the scope of observation itself open to challenge. Not every challenge must immediately add a new survey; what must be stated is whether the omission affects the decision being made, and whether a proportionate form of supplementation exists. Otherwise the real constraint of limited resources may be used as a reason for perpetually refusing experience from the same direction.
Uncertainty of Fact and Uncertainty of Decision
Suppose the record of pedestrian flow is still incomplete, yet the town must decide whether to adjust passage temporarily. That the facts remain uncertain does not mean all action must stop; but it must be stated what a provisional decision rests on, and which questions remain open.
For arrangements that can be corrected at low cost, allowing action and observation to proceed together may be reasonable; for changes that are hard to reverse, the key conditions call for closer examination. The distinction derives from consequences and reversibility, not from which side of the dispute one favors. One cannot demand absolute evidence for decisions one dislikes while accepting the weakest clues for decisions one endorses.
A provisional decision should also have an object of feedback. If the trial aims to reduce congestion in a certain interval, the change in that interval should be observed, and an indicator that more easily displays success cannot be substituted after implementation. When the criteria of evaluation must be adjusted, it should be stated why, and how the old and new results cannot be directly compared. This point connects with the dark usage: when decision-makers can redefine success at will after results appear, every result can be redescribed as having met expectations. The interpretive space for failure keeps expanding, and facts can no longer constrain action.
An arrangement open to error-correction must permit a limited but definite conclusion: under the conditions and goals originally stated, this measure did not achieve what was expected. Acknowledging such a conclusion need not negate the whole program of reform, yet it prevents the legitimacy of the goal from granting permanent immunity to every particular means.
Narratives Can Change Action but Cannot Exempt It from the Verification of Consequences
Stories can organize scattered material and can help people understand why a certain fact concerns them. Without narrative, public discussion would not necessarily consist of pure data alone; it might consist only of disconnected fragments. Criticizing the power of narrative therefore does not require society to renounce narrative. Yet precisely the organizing power of narrative means it can produce dark consequences. A story of common prosperity can make certain losses appear a matter of course; a story of total ruin can make any improvement appear a disguise. Both can subordinate concrete consequences to a total meaning settled in advance.
RC's analysis of how conceptions influence experience should allow us to face the actual power of narrative without treating that power as a reason to stop checking facts. That a person becomes more willing to support a decision because he has accepted a story is one possible change in conduct; whether the improvement in life the decision promised actually occurs is another matter that still has to be examined.
If the two are run together, success in mobilization may be called success in reality. That people are more inclined to believe does not by itself show that passage beside the bridge has become more convenient; nor does the decline of opposing voices by itself show that the inconvenience of the affected has disappeared. Public attitude can itself become an object of study, but it cannot replace all the actual consequences it points to.
Critique must also avoid simplification in the reverse direction. That a person accepts a certain interpretation does not necessarily mean he has been deceived; the interpretation may genuinely help him organize experience he could not organize before. To judge its worth, we need to see whether it admits counterexamples, states its limits, and is willing to let its promises be tested by subsequent life, not merely whether the expression is infectious.
Who May Continue to Speak after Correction
Once a public record is bound to a group identity, admitting a local error may be understood as helping the opponent. Recorders who discover problems may still fear that correcting them will cost them all their credibility. This cost affects whether errors can surface at all, and is not merely a question of personal honesty.
A public environment that permits correction needs to allow local failures to be handled according to their scope. Deliberate concealment, repeated irresponsibility, and voluntary correction after reasonable verification should not be treated as the same kind of conduct. Otherwise, having no path to recover reliability through correction, people find it easier to maintain their original claims to the end. This does not require the public to forgive every error promptly. Some failures cause grave consequences and require corresponding accountability. Distinguishing responsibility from permanent revocation of the standing to observe is meant to keep subsequent material admissible, not to abolish accountability in the name of openness.
After the town's recorders correct the pedestrian data, they can still take part in supplementary surveys; the residents affected by the old data can likewise demand an account of whether the earlier decisions need to change. Only when both hold together do public facts have both people who correct errors and reasons to demand correction.
There is a further distinction: not having confirmed a conclusion and having confirmed that the conclusion is false are not the same. If the town temporarily lacks sufficient material to judge whether a certain interval is persistently congested, this unknown cannot be written down as never having been congested; equally, one cannot, because congestion cannot be ruled out, announce the fact as established in its strongest version. The unknown should keep its own place until new material changes it. This place is especially liable to be lost in mobilization. Supporters and opponents alike may find uncertain expression insufficiently forceful, but writers cannot, for the sake of momentum, supply what the evidence has not provided. What public facts require is not only the willingness to correct errors but also the refusal to answer for reality in advance of knowing.
The Stability of Public Facts Is an Explicable Relation
After all these distinctions, the town still will not possess a master table exhausting life beside the bridge. The record of pedestrian flow can help us understand crossing; it cannot replace all understanding of lingering, memory, and everyday need. But it can become a ground with a clear scope that different positions can use in common. This stability requires neither unanimity of opinion nor numbers that never change. It depends on objects being clear, sources being traceable, inferences having bounds, contrary material having an entrance, and corrections being able to change actual use. Each of these conditions demands investment, and each can become a position through which power exerts influence.
Precisely because the public formation of facts has these conditions, we can neither reduce facts to the command of the strong nor mistake the existence of the conditions for evidence that facts do not exist. Critique must face both temptations at once, the monopoly of interpretation and the abolition of verification. The foundation of a common world is that different people remain able to bring a specific claim back to comparable material, without first having to accept one party's approval of the whole of reality.