Seats
席位
Paper 3 said that whether misrouting can last depends on where the function of questioning the ought lies, and how large it is. This paper looks at seats. A seat means a function, not an institution: within one institution, several committees each have their own powers and touch the same kind of decision; which of them is questioning which line on the normative ledger is registered by powers and by the actual record of handling, not sorted by name. Comparison must align levels and also functions: at the same level, compare at the same stage; the several functions of one institution are registered separately (S21). In the rule-making cases this paper examined, one can see: the evidence function is standing, routine, and has decision-making power; the function of questioning normative premises, in the public participation arrangements examined, is advisory only, and the committee that bore it was set up and then stopped; the jurisdiction of appeals is narrowed to procedure. Three further types, absent, set up ad hoc for one topic, and set up only after something goes wrong, are registered only as types; this paper has no rule-making instance of them. This is existential and typological, not a claim about numbers (S22). Ethics review at the project level has real power; as to the long-range effects of applying knowledge, the text excludes them from this assessment of risk (S23). Comparison is registered column by column by condition; "advisory power only" is not by itself judged to be toothless (S24). An arrangement in which the four conditions hold jointly and which is standing cannot be pointed to as an instance of their not being met; it may still take an advisory or other organizational form; existential claims can also be refuted (S25). Five propositions in all (S21 to S25), stated at the end of each section and listed together at the end.
Writing Declaration: This paper was co-drafted with Claude (Anthropic). All intellectual decisions, framework design, and final editorial judgments were made by the author.
Introduction
When the UK National Institute for Health and Care Excellence (NICE) appraises a new drug or technology, a technology appraisal committee looks at the evidence: clinical effectiveness, cost, cost-utility; it also makes trade-offs according to the thresholds and methods the Institute uses. The committee's recommendations have consequences: they decide whether the technology can be used in the National Health Service.
The same Institute also had a Citizens Council, made up of ordinary citizens, which gave views on questions of social value in the Institute's work. In its principles document Social Value Judgements, the Institute drew on the Citizens Council's reports (NICE 2008). The Citizens Council could only advise. It did not meet after 2015. The Institute later set up NICE Listens, a deliberative public engagement programme, whose first project was in 2021; the Institute's annual report states: "NICE Listens replaces our Citizens Council." (NICE 2022, print edition p.24). Its recommendations, too, are only for the consideration of the Institute's executive team and board. Whether this kind of questioning had any other bearer in the interval, this paper has not checked; what is registered here are the powers and continuance of the two arrangements, and it is not concluded from them that this function of the Institute ever stopped as a whole.
The two committees touch the same kind of decision. The technology appraisal committee assesses evidence and also makes trade-offs according to the thresholds and methods the Institute uses, and its recommendations have consequences; the Citizens Council gave views on the social values behind these trade-offs, could only advise, later stopped, and was succeeded by NICE Listens. Which of them is questioning which line on the normative ledger is not sorted by name; it is registered item by item by the columns of §4. This paper starts from here. It does not judge whether NICE does well or badly; what it asks is how arrangements like this should be compared.
The reading follows the Prequel. This paper concerns the second domain, standards that act on persons (Prequel §5). §1 says that comparison must align levels and functions; §2 registers several forms at the rule-making level; §3 speaks of narrowing at the project level; §4 gives the columns for comparison; §5 gives conditions for counterexamples; §6 lists several contrasts on the government side.
§1 Aligning Levels and Functions
The reviewer of a single paper can reject a manuscript that used the wrong statistical method, but cannot change the threshold of "significance." What can change a discipline's premises is at the rule-making level: the working groups of a diagnostic manual, the working groups of evidence grading, standards committees, the revision groups of a guideline's methods manual.
So comparison must first align levels. Compare a reviewer with a standards committee, and the difference that comes out is only a difference of level.
Once levels are aligned, stages must be aligned too. From its formation to its taking effect, a standard passes through several stages: development, recommendation, adoption, implementation, review. Institutions at the same level are responsible for different stages. Comparison must be made at the same stage: compare a committee that develops standards with a body that reviews their implementation, and the difference that comes out is again only a difference of stage.
One institution often carries several functions. In NICE, the technology appraisal committee and the Citizens Council have different powers, yet both touch lines on the normative ledger: appraising cost-utility against a threshold is itself a trade-off (Paper 6, §3). So when this paper says "the evidence seat" and "the ought seat," it means functions, not institutions: an institution can carry both functions, or only one; in registering, the several functions of one institution are recorded separately. The main comparison is placed within one construct, because there levels and stages are easiest to align.
S21 Aligning levels and functions. The reviewer of a single paper cannot change a discipline's premises; what can is at the rule-making level: the working groups of a diagnostic manual, the working groups of evidence grading, standards committees. Comparison is made at the same level and the same stage (development, recommendation, adoption, implementation, review); the several functions of one institution are registered separately. "The evidence seat" and "the ought seat" mean functions, not institutions.
§2 Forms at the Rule-Making Level
In the rule-making cases this paper examined, the evidence function is standing, routine, and has decision-making power: the committees that assess evidence meet regularly, work by their manuals, and their conclusions enter decisions. For the function of questioning normative premises, two forms can be seen in the cases examined:
- Advisory only. There is a position that can voice views, but the decision lies elsewhere. NICE's Citizens Council, and NICE Listens which succeeded it, are both of this type.
- Set up and then stopped. It was set up, it ran, and later it stopped. NICE's Citizens Council, as an arrangement, is of this type; its work was later taken up by NICE Listens, and what is recorded here is the continuance of this arrangement, not the continuance of this function of the Institute.
Three further types are registered only as types; this paper has no rule-making instance of them:
- Absent. No stage in the procedure deals specifically with that line on the normative ledger.
- Set up ad hoc for one topic. A question arises, a committee is set up, the question is answered, the committee is dissolved. There is one instance on the government side; see §6.
- Set up only after something goes wrong. Only once a dispute has blown up is a position set up to deal with it.
There is also an instance of "set up and then stopped" on the government side, the US Congress's Office of Technology Assessment; see §6.
There is one more form, which falls on appeals. Standard development has a seat for appeals, but its jurisdiction can be narrowed to procedure. The Board of Standards Review of the American National Standards Institute "does not evaluate the content of standards and does not hear appeals with regard to technical issues"; the IEEE Standards Association provides that it "shall not provide an appeals process to resolve a disagreement about a technical decision" (ANSI; IEEE SA). There are appeals, and the appeals ask only about procedure. The contrast is the International Organization for Standardization and the International Electrotechnical Commission: their Directives allow appeals on technical matters (ISO/IEC Directives, Part 1).
The Red Book cited in Paper 3 is a close forerunner at this level: it called for the assessment and management of risk to be separated in concept, but did not recommend placing the two in different organizations (National Research Council 1983). The functions were separated in concept; the seats were not.
These are existential and typological: the two types that have instances are written as they are; the other three are only types. It is not said that "most rule-making bodies have no function of questioning normative premises." To use a quantifier such as "most," one must first fix a sampling frame, for example all the bodies that write standards in a field, and then check them one by one; this paper has not taken that step.
The type "set up only after something goes wrong" joins Paper 5: a remainder covered by a default value is still on the ledger; when someone takes it up, newly established questioning is one shape that taking up can have.
S22 Forms at the rule-making level. In the rule-making cases this paper examined, one can see: the evidence function is standing, routine, and has decision-making power; the function of questioning normative premises, in the public participation arrangements examined, is advisory only, and the committee that bore it was set up and then stopped (NICE's Citizens Council, and NICE Listens which succeeded it); the jurisdiction of appeals is narrowed to procedure (ANSI, IEEE SA). Three further types are registered only as types, and this paper has no rule-making instance of them: absent; set up ad hoc for one topic (one instance on the government side); set up only after something goes wrong. This is existential and typological, not a claim about numbers.
§3 Narrowing at the Project Level
Below the rule-making level is the project level: a particular study, a particular protocol. At this level there is a seat with real power, the ethics review board. It can approve, require modifications, and disapprove.
Its jurisdiction is drawn very clearly. Paper 1's Introduction cited the provision: the board "should not consider possible long-range effects of applying knowledge gained in the research (e.g., the possible effects of the research on public policy) as among those research risks that fall within the purview of its responsibility" (45 CFR 46.111(a)(2)). The text says should not: not "need not," but should not. As to the long-range effects of applying knowledge, which this paper asks about, the provision excludes them from this assessment of risk. An ethics review board's jurisdiction is not only risks and benefits: the equitable selection of subjects, informed consent, and the protection of privacy are in the other paragraphs of the same section; this paper asks only about the one item of long-range effects.
Methodology is drawn the same way. In 2024 the US Department of Health and Human Services' Secretary's Advisory Committee on Human Research Protections issued a recommendation on how research of no informational value should be reviewed. It says: "It is not the role of the IRB to either design or improve research methodology, but to identify research that it finds is unlikely to achieve its stated goals." The board's authority to disapprove research or require modifications is limited to cases where risks are unreasonable, or where modification is needed to meet the balance of risks and benefits (SACHRP 2024).
So this seat at the project level has teeth, but on the two matters this paper asks about, what it can bite is marked out: the long-range social effects of research, the text states, do not count among the risks it assesses; as to methodology, it neither designs nor improves it, only identifies research unlikely to achieve its stated goals, and its power to disapprove or require modifications is tied to risks and benefits. Paper 1 said that the federal regulation states "this is not handled here," without stating "who handles it." Stanford University's additional Ethics and Society Review makes completion of the Ethics and Society Review process, including consideration of the proposal's possible risks to society, a requirement for funding from participating programs (Bernstein et al. 2021): it takes up part of what is left, as a supplementary arrangement with its own scope, with funding as its gate.
S23 Narrowing at the project level. Ethics review at the project level has real power: it can approve, require modifications, and disapprove. As to the long-range effects of applying knowledge, the text excludes them from this assessment of risk; as to methodology, it neither designs nor improves it, only identifies research unlikely to achieve its stated goals; its authority to disapprove or require modifications is limited to cases of unreasonable risk or an imbalance of risks and benefits. This does not summarize the whole jurisdiction of ethics review.
§4 Comparing by Condition
With levels and stages aligned, what is compared? Column by column, by the four conditions of Paper 2, with several columns added. The columns of a comparison table can be laid out like this:
| Column | What is recorded |
|---|---|
| p | The premise being questioned |
| J | Jurisdiction |
| Stage | Development, recommendation, adoption, implementation, review |
| Jurisdictional power | Whether it can put the scope of warrant itself in question |
| Right of response | Whether it can require core practitioners to respond publicly |
| Right of review | Whether it can initiate a review, a failure test, or an evaluation of alternatives |
| Condition of independence | Whether its continued existence and evaluation depend on p |
| Record of actual initiation | Whether these powers have been used, how often, with what result |
| Standing | Standing, temporary, stopped |
| Trigger | Self-initiated, mandated, after something goes wrong |
| Coverage | Which standards, which objects it covers |
Take the two committees in the Introduction as a trial fill, to see which cells can be filled. First, a note: this table is a demonstration of registering scopes of work, not yet a comparison of conditions on the same premise; the premises the two touch must first be brought down to the same p before they can be compared column by column.
| Column | Technology appraisal committee | Citizens Council |
|---|---|---|
| p | Not aligned | Not aligned |
| J | Adoption of technologies in the National Health Service | Questions of social value the Institute put to it |
| Scope of work | Assesses clinical effectiveness, cost, cost-utility, by the Institute's thresholds and methods | Gives views on the questions of social value put to it |
| Stage | Recommendation (recommendations have consequences) | Recommendation (advisory) |
| Jurisdictional power | Not checked | Not checked |
| Right of response | Not checked | Not checked |
| Right of review | Not checked | Not checked |
| Condition of independence | Not checked | Not checked |
| Record of actual initiation | Routine | Reports exist; none found after 2015; succeeded by NICE Listens from 2021 |
| Standing | Standing | Stopped |
| Trigger | Mandated, routine | Mandated |
Most cells read "not checked." That is not a defect of the table but its use: it sets out what is not yet known, and does not let the sentence "one has decision-making power, the other can only advise" stand in for registering the four conditions item by item.
One point must be held to especially here: "advisory power only" is not by itself judged to be toothless. Can the advice require the other side to answer? Can it initiate a review? Can its views be set aside without reasons? Only once these three have been checked can one say whether advice has teeth. A position that can only advise but has the right to require a written answer point by point is not the same as one whose advice, once given, is the end of it.
S24 Comparing by condition. The comparison table aligns these columns: p; J; stage; jurisdictional power; right of response; right of review; condition of independence; record of actual initiation; with standing, trigger and coverage added. "Advisory power only" is not by itself judged to be toothless: check whether the advice can require an answer, whether it can initiate a review, whether its views can be set aside without reasons.
§5 Conditions for Counterexamples
This paper's generalizations must be open to refutation.
For the same p, the same jurisdiction and the same stage, if there is an arrangement in which the four conditions for the corresponding function hold jointly and which is standing, it cannot be pointed to as an instance of the four conditions not being met, and it limits the range over which this paper's generalizations apply. It may still take an advisory or other organizational form: a standing advisory position can require responses, initiate reviews and stay independent, while not holding the final decision. Form and conditions are registered separately. A judgment about a field as a whole must be tested against the same sampling frame and coverage. Existential claims can also be refuted: if a particular case pointed to as "not met" in fact has the corresponding powers, protections and record of initiation, that identification does not stand.
There is already an example that limits this paper's generalizations. GRADE, the system for grading evidence, rates which outcomes count and how important each is before rating the certainty of evidence (GRADE Handbook). Its Evidence to Decision framework lists how people value the main outcomes as one of the judgments to be made in forming a recommendation (Alonso-Coello et al. 2016). This shows that, within this procedure, the importance of outcomes is not left unevaluated: this line on the normative ledger is placed at the start of the procedure. So the generalization "the function of questioning normative premises is absent everywhere" does not hold here.
Whether it meets the four conditions must be registered item by item: who rates the importance of outcomes; whether those affected can get in; whether the results of the rating can require a response, whether they can initiate a review. These items have not all been checked. Until they are, "not checked" is kept, and this section judges neither that it is a counterexample nor that it is not. It serves first as a positive example: a line on the normative ledger can have a place in the procedure.
Stanford's Ethics and Society Review is another kind: a newly established position that takes up part of the questions left at the project level, with completion of its review required for funding from participating programs (Bernstein et al. 2021). Its four conditions, too, are recorded separately item by item.
S25 Conditions for counterexamples. For the same p, J and stage, an arrangement in which the four conditions for the corresponding function hold jointly and which is standing cannot be pointed to as an instance of the four conditions not being met; it may still take an advisory or other organizational form, and the two are registered separately. Such an arrangement limits the range over which this paper's generalizations apply; judgments of coverage about a field must be tested against the same sampling frame and coverage. Existential claims can also be refuted: a particular case pointed to as "not met" may in fact have the corresponding powers, protections and record of initiation.
§6 Contrasts on the Government Side
The government side has also set up seats that question the ought. They are not part of this paper's main comparison, because they do not sit at the rule-making level of the disciplines; they serve as contrasts, to see how seats can be set up.
- The US Congress's Office of Technology Assessment. Established by the Technology Assessment Act of 1972, with the statutory task of providing Congress with assessments of the consequences of technology (2 USC §472). In 1995 Congress voted to withdraw its funding, and the Office closed on 29 September of that year (OTA archive page); what stopped was the funding, and the provisions establishing it are still in the Code. Standing, mandated, set up and then stopped.
- The German Ethics Council. Under the Ethics Council Act, it follows the ethical, social and legal questions raised by research in the life sciences, informs the public, issues opinions and makes recommendations; the Act states that it is independent in its work and bound only by the tasks the Act sets (Ethikratgesetz §§2–3). Standing, independent, able only to advise: of the "advisory only" type.
- The German Ethics Commission on Automated and Connected Driving. Set up by the Federal Ministry of Transport to work on one topic, it delivered its report in 2017, proposing a set of ethical rules (Ethik-Kommission 2017). Of the "set up ad hoc for one topic" type.
- Article 27 of the EU Artificial Intelligence Act. Certain deployers of high-risk AI systems must, before deployment, assess the system's impact on fundamental rights and notify the market surveillance authority of the results (Regulation (EU) 2024/1689, Article 27). For high-risk systems listed in Article 6(2) and Annex III, this obligation applies, under the 2026 amending regulation, from 2 December 2027 (Article 113 as replaced by Regulation (EU) 2026/1744): the law already provides for it, and the obligation has not yet begun to apply. It places a normative assessment before deployment, on the adopting side; it does not belong to the five types of §2, and Paper 6 speaks further of the adopting side.
The first three belong to three of the forms in §2; the fourth belongs to none of the five types. What they show is that seats can be set up in many forms; this paper does not judge which is good.
Where This Paper Stops
This paper does not say where the function of the ought should be set up or what it should look like; that is the business of Paper 9, and there too it is offered only as suggestion; it does not judge which institution does badly.
What it hands to Paper 5: how a premise that the using side treats as absent goes on acting.
The remainder it leaves: the trial table in §4 has not yet been aligned to one premise, and most cells are "not checked"; GRADE's four conditions have not all been checked; the three further types in §2 have no rule-making instance, none of the five types has a sampling frame, and no claim about numbers is made.
The Five Propositions of This Paper
- S21 Aligning levels and functions. Comparison at the same level and the same stage; the several functions of one institution registered separately; "the evidence seat" and "the ought seat" mean functions. (Notation: method of comparison)
- S22 Forms at the rule-making level. In the cases examined: the evidence function is standing, routine, and has decision-making power; the function of questioning normative premises, in the public participation arrangements examined, is advisory only, and the committee that bore it was set up and then stopped; appeals are narrowed to procedure. The three further types (absent, ad hoc for one topic, set up only after something goes wrong) are types only, with no rule-making instance. Existential and typological. (Descriptive; typological registration)
- S23 Narrowing at the project level. Ethics review has real power; as to the long-range effects of applying knowledge, the text excludes them from this assessment of risk; as to methodology, it neither designs nor improves, only identifies research unlikely to achieve its stated goals, and its power to disapprove or require modifications is tied to risks and benefits; this does not summarize the whole jurisdiction of ethics review. (Descriptive; 45 CFR 46.111(a)(2); SACHRP 2024)
- S24 Comparing by condition. The columns of the comparison table; "advisory power only" not by itself judged toothless. (Notation: method of comparison)
- S25 Conditions for counterexamples. An arrangement in which the four conditions hold jointly and which is standing cannot be pointed to as an instance of their not being met, may still take an advisory or other form, and the two are registered separately; it limits this paper's generalizations; existential claims can also be refuted. (Notation: conditions for counterexamples, a falsifiable way of writing)
References
I · This system
- Qin, H. (2026). Parochialism (SAE Theory of Disciplines, Paper 1) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24177748
- Qin, H. (2026). Standard-Setting (SAE Theory of Disciplines, Paper 2) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24177754
- Qin, H. (2026). Misrouting (SAE Theory of Disciplines, Paper 3) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24177759
II · Rule texts and institutional documents
- 2 U.S.C. §472 (Technology Assessment Act of 1972).
- 45 CFR 46.111(a)(2). Criteria for IRB approval of research. eCFR (current text).
- American National Standards Institute. Description of the responsibilities of the Board of Standards Review (current version).
- Office of Technology Assessment. Archive page (archived 2008, UNT Digital Library). webarchive.library.unt.edu.
- Ethik-Kommission Automatisiertes und Vernetztes Fahren. (2017). Bericht. Bundesministerium für Verkehr und digitale Infrastruktur.
- Gesetz zur Einrichtung des Deutschen Ethikrats (Ethikratgesetz), 2007, §§2–3.
- IEEE Standards Association. Standards Board Bylaws (current version), the provisions on appeals.
- ISO/IEC Directives, Part 1 (current version), the provisions on appeals.
- NICE. (2022). Annual Report and Accounts 2021/22 (print edition p.24; p.23 in the accessible web version).
- NICE. (2008). Social Value Judgements: Principles for the Development of NICE Guidance (2nd ed.).
- Regulation (EU) 2024/1689 (Artificial Intelligence Act), Article 27 (consolidated text of 27 July 2026).
- Regulation (EU) 2026/1744 of 8 July 2026, amending Regulation (EU) 2024/1689 (Article 113).
- Secretary's Advisory Committee on Human Research Protections. (2024, October 22). Recommendation on Assessment Considerations on Uninformative Research.
- Schünemann, H., Brożek, J., Guyatt, G., & Oxman, A. (Eds.). (2013). GRADE Handbook.
III · External literature
- Alonso-Coello, P., et al. (2016). GRADE Evidence to Decision (EtD) frameworks: A systematic and transparent approach to making well informed healthcare choices. 1: Introduction. BMJ, 353, i2016.
- Bernstein, M. S., et al. (2021). Ethics and society review: Ethics reflection as a precondition to research funding. PNAS, 118(52), e2117261118.
- National Research Council. (1983). Risk Assessment in the Federal Government: Managing the Process. National Academy Press.
Interfaces
Main upstream: Paper 2, S12 (the four conditions); Paper 3, S16.
Also used: NICE 2008 and the records of the Citizens Council; the OTA statute and archive page; the appeals rules of ANSI, IEEE and ISO/IEC; 45 CFR 46.111(a)(2); SACHRP 2024; the GRADE Handbook and Alonso-Coello et al. 2016; Bernstein et al. 2021; the two German ethics bodies; Article 27 of the EU Artificial Intelligence Act; the Red Book 1983.
This series: S21 to S25.
Handed to Paper 5: how a premise that the using side treats as absent goes on acting.