Skip to content
finalstate.aiWho Drew the Menu?
007Atlas

007THE CARD

Who Drew the Menu?

A model offers Renew, Reprice or Rebuild for a customer-support system. The manager can make a genuine choice among them while still needing to inspect how the list was formed, and whether it can be reopened.

An illustrative menu lists A: Renew, B: Reprice and C: Rebuild; the manager selects Reprice.The same named alternatives recur through the construction review. The selection is genuine, while the list supplies no construction record. It does not depict measured user behavior.RenewARepriceBRebuildC
Illustrative customer-support-system decision. Choosing a row does not establish authorship of the list or authority over its construction.

THE DRAW CAN BE REOPENED

Choosing a row leaves an earlier question unanswered.

Renew, Reprice and Rebuild remain inside a selection boundary; Partner is retained as a cut under a cost cap, with a route to reopen the set.The cut rule precedes selection. Reopening is a proposed review route requiring authority or escalation; the figure does not approve Partner or rank it above the admitted candidates.RenewRepriceRebuildPartnerCost cap
Illustrative construction sequence. A cut can be questioned without assuming that the excluded candidate is better.

The menu admits A: Renew, B: Reprice and C: Rebuild. A fourth candidate, X: Partner, was cut under a cost cap. That rule and its application can be questioned before the choice is treated as final. Admission differs from the factual and value premises used to assess the admitted alternatives. A reviewer needs access to the construction record and authority to reopen the range, or an escalation route to someone who has it.

THE SUPREME INSTRUMENT

The definition of the alternatives is the supreme instrument of power.

E. E. Schattschneider, The Semisovereign People, consulted 1975 reissue, later HBJ printing, ch. 4, p. 66. Clause excerpt; initial capital added.

Schattschneider connects the definition of political alternatives to the choice of conflicts. The consulted 1975 reissue, in a later HBJ printing, places this clause in chapter 4, p. 66. Applying agenda control to a generated menu asks who shapes admission. It does not establish sole control of the outcome, and structural exclusions need not have an identifiable individual censor.

THE RECOMMENDATION AND THE MENU

Both formats can leave the search unexplained.

A recommendation names Reprice; a separate menu lists Renew, Reprice and Rebuild without displaying the search that produced either.This still comparison distinguishes an explicit recommendation from a choice among alternatives. Neither artifact alone establishes authorship, completeness or control of the upstream search.RecommendationRepriceMenuRenewRepriceRebuild
Illustrative formats. Rejection and selection reveal decisions about the presented material, not the adequacy of its construction.

A manager may reject a recommendation or choose another row without learning what was excluded. The feeling of authorship, a genuine selection and authority over the search are different things. An override count alone cannot establish the last of them.

A longer menu can retain the same exclusions.

A cost cap that excludes every partnership can survive the addition of more rows. Inspect the reason for that boundary and who may revise it.

NUTT’S USE-BASED OUTCOME

Nutt assessed whether decisions were put to use.

A fixed half-length bar represents Nutt's rounded failure finding, alongside a separate two-year use-assessment interval.The outcome is based principally on use, not a universal measure of business value. The paper lists 356 cases, while its publisher abstract lists 358. No exact failure count is inferred, and no causal arrow attributes all failures to limited search.About half failedUse over two years
Nutt, Academy of Management Executive 13(4), November 1999, pp. 75–90. Scan pp. 75–76: 356 decisions; publisher abstract: 358. Use-based outcomes, p. 77; nonrandom case reconstruction, endnote 5, p. 89. DOI: 10.5465/AME.1999.2570556.https://cebma.org/assets/Uploads/Nutt-1999-gecomprimeerd.pdf

Paul Nutt's 1999 paper reports that about half of several hundred organizational decisions failed. Success was assessed principally through use, including whether a decision was sustained and fully used after two years (p. 77). The cases involved senior managers in medium-to-large US and Canadian organizations. Interviews reconstructed the decision process; executives suggested cases. Limited search was among the tactics Nutt associated with poor outcomes. These observational records do not establish the causal effect of adding an option.

THREE STEERS

Set composition can change what gets chosen.

Fixed schematic rows show Rival R at $30 and score 8, Target T at $40 and score 10, and Decoy D at $42 and score 9.Lower price and higher score are preferred. T dominates D on both attributes; R is cheaper but lower-scoring than D and does not dominate it. D enters before the comparison marks. No empirical choice shares, individual switch or decoy removal is shown.Price ↓Score ↑Rival R$308Target T$4010Decoy D$429TDDominatesRDTrade-off
Huber, Payne and Puto, Journal of Consumer Research 9(1), June 1982, pp. 90–98. Method p. 92: 153 business students in simple two-attribute product choices. Schematic attributes; full-set choice probability; no empirical percentages shown. DOI: 10.1086/208899.https://people.duke.edu/~jch8/bio/Papers/Huber%20Payne%20Puto%201982%20JoCR.pdf

Huber, Payne and Puto's 1982 product-choice experiments show that adding an asymmetrically dominated alternative can increase choice of the option that dominates it. The comparison uses the full available set, including the decoy. It is not a recalculated share among the original pair alone. Two further menu diagnostics concern how alternatives were generated. Pseudo-diversity asks whether apparently different rows share one underlying assumption. The mirror asks whether the search returned the chooser's prior preference. These are review questions, not measured effects established by that experiment.

THE CONSTRUCTION RECORD

A recorded cut is different from a candidate never generated.

A, B and C enter the presented-set boundary; candidate X moves into a discard record labeled Cost cap.In the illustrative support-system case, A is Renew, B Reprice, C Rebuild and X Partner. Other dots stand for an open-ended unranked field, not a measured population or a certified set of feasible actions. The recorded exclusion differs from alternatives never generated.Candidate fieldABCMenuCost capXDiscard
Illustrative record: A Renew, B Reprice, C Rebuild; X Partner, cut under a cost cap applied by Finance. A recorded cut and a never-generated alternative are different evidence states.

Return to the support-system case. A: Renew, B: Reprice and C: Rebuild reached the menu. X: Partner was cut under a cost cap applied by Finance. Keeping that record makes the rule and its owner available for challenge. An alternative never generated would appear in neither list. A shadow option also requires a supported opportunity and resources or rights to act, alongside a recognition gap; omission alone does not establish those conditions.

A complete discard record can describe an incomplete search.

The record identifies exclusions made within the search. It cannot enumerate every alternative that might have been found by another search. A new candidate still needs evidence and assessment.

THE GATE

Keep the construction open to review.

A cut record identifies Partner, the cost cap and Finance; a reopening route remains available while review stays pending.Only the reason and accountable group are supplied by this fictional record. A route for questioning the boundary does not approve the excluded candidate or certify the menu complete.PartnerCost capFinancePending
Illustrative review route. The documented exclusion supports a question; assessment and any authorized revision still have to happen.

In this example, Partner's cut has a reason and an accountable group. Its merits have not been assessed here. Ask Finance to explain or reconsider the cost cap, and seek alternatives from a source or reviewer with evidence outside the original generator. A missing record limits inspection; it does not prove that the menu is bad or captured. Keep a route to reopen the range or escalate the question. Preserve the decision and its construction in an archive that answers with receipts.

Read the transcript

01 · THE CARD

A manager is deciding what to do with a customer-support system. A model produces three alternatives: renew the current contract, reprice it, or rebuild the system. In this illustrative case, the manager picks Reprice. The selection is genuine. Whether the manager shaped the search is a different question. The rows do not say which candidates were excluded, what rule excluded them, or who can revise it.

02 · THE DRAW CAN BE REOPENED

Before the pick, someone or something admitted candidates and applied a cut rule. In this example, Partner was removed under a cost cap. Renew, Reprice and Rebuild reached the menu. Choosing among them cannot select Partner while the list stays fixed. But the chooser can still reopen the set, if they know to ask. That requires authority to revise the boundary, or a route to someone who has it. Admission concerns what reaches comparison; premises concern the claims and objectives used to judge candidates.

03 · THE SUPREME INSTRUMENT

In The Semisovereign People, first published in 1960, Schattschneider writes: The definition of the alternatives is the supreme instrument of power. The clause appears in the consulted 1975 reissue, chapter four, page sixty-six. Its subject is political conflict. Applying it to an organizational menu asks how alternatives reached consideration. It does not make the menu author the sole decision maker. Institutional rules can exclude an issue without one identifiable censor.

04 · THE RECOMMENDATION AND THE MENU

A recommendation to Reprice makes a judgment visible. A menu of Renew, Reprice and Rebuild gives the manager alternatives to compare. Neither format, by itself, supplies the construction record. Rejecting the recommendation or choosing another row may leave the same search boundary intact. A selection, a feeling of authorship and authority to reopen the search establish different things.

05 · MORE IS NOT MORE

Adding rows does not answer a question about exclusions. If the same cost cap rules out every partnership, a longer menu can still leave that whole approach unexplored. The reviewer can ask whether the cap is justified and who can revise it. The count alone does not settle either question.

06 · NUTT’S USE-BASED OUTCOME

Paul Nutt's 1999 paper reports that about half of several hundred organizational decisions failed. The scanned paper gives three hundred and fifty-six cases; the publisher abstract says three hundred and fifty-eight. That discrepancy remains unresolved. The cases came from senior managers in medium-to-large organizations in the United States and Canada. Success was measured principally by use, with sustained and full use assessed over two years. Executives suggested cases, and interviews reconstructed the process. Nutt associated limited search with poorer outcomes, alongside other tactics. The study did not randomly assign managers to different numbers of alternatives. It cannot establish what one additional option would cause.

07 · THREE STEERS

In a separate product-choice example, lower price and higher score are preferred. Rival R costs thirty dollars and scores eight. Target T costs forty dollars and scores ten. Add the decoy D, priced at forty-two dollars with a score of nine. T is cheaper and scores higher than D, so T dominates it. R is cheaper than D but scores lower, so that comparison remains a trade-off. Huber, Payne and Puto's 1982 experiments found that adding an asymmetrically dominated option can increase choice of the option that dominates it, measured across the full available set. This illustration supplies no empirical choice shares. Two separate diagnostics ask whether the rows share one assumption or the search returns the chooser's prior preference. The experiment did not establish those effects.

08 · Advertisement · Bubble AI App Builder

Some ideas do not need another document before they become testable. With Bubble AI, you describe the app you want, and Bubble creates a working starting point: interface, data, and logic you can inspect. From there, you refine visually, connect AI models and services, and turn the first version into something real enough to use.

09 · THE CONSTRUCTION RECORD

Return to the illustrative support-system menu. A, B and C are the presented candidates: Renew, Reprice and Rebuild. Candidate X is Partner. Its discard record retains the cost cap applied by Finance. X may be weak or strong; the record supplies no ranking. It allows the reviewer to question the reason for the cut. A different search might find an alternative that the first search never generated. Such a candidate would be absent from both the menu and its discard record. Omission alone does not establish a shadow option: the opportunity and the resources or rights to act also need support.

10 · THE RECORD’S LIMIT

The discard record can be complete about the cuts it records and still describe an incomplete search. It shows which candidates were rejected under which rules. It cannot tell us every candidate that another search might discover. A newly found alternative begins an assessment; discovery does not make it the better choice.

11 · THE GATE

The record gives Partner's exclusion a reason: the cost cap. Finance is the group to ask about that rule and its application. Review remains pending. The manager can request a wider search using evidence outside the original generator, or escalate to whoever can authorize a change. A missing record is a limit on inspection, not proof that the visible options are bad. Keep the record and a route to reopen the range. The manager still has to decide whether the available evidence is sufficient to act.

01 / 11 · THE CARD0:00 / 6:18