Mandates and the authorising environment
Authority, capacity and practice, and nested jurisdictions.
The literature agrees that an institution's ability to act differs from what its founding text authorises, and divides on how to read the difference. Moore's strategic triangle (value, authorising environment, operational capacity) and the Harvard working paper on managing an authorising environment give practitioners the working vocabulary; the paper treats authority as something built, and spread across offices. Pritchett, Woolcock and Andrews give the sharpest warning for Holagram: forms that imitate function, and tools asked to carry too much, apply to a mandate map as much as to the units it maps. Meyer and Rowan, Brunsson, Pressman and Wildavsky, Lipsky, and Hill and Hupe describe separate gaps: between text and structure, decision and act, layer and layer. Crawford and Ostrom's institutional grammar is the nearest method for turning a decree into checkable statements. For testing a claim at another level the best material is primary: the UAE and Italian constitutions both contain procedures by which one level challenges another's competence, and Italy's sets a sixty-day window. The most direct objection is Rhodes and Wanna's: reading and stretching a mandate is not an official's job. The 19 June 2026 Dark Matter Labs piece on Lund argues, for climate planning, that visioning fails without grounding in cost and policy change; its five questions do not ask who may decide, which is where a Mandates Entry would start. A public UAE sandbox page is proposed as the first-study document.
Sources
- Mark H. Moore (1995). Creating Public Value: Strategic Management in Government. Harvard University Presscheckedmust read
Supplies the base question of the Mandates method: can this unit take this step? Moore's strategic triangle (recalled from the book, not from the catalogue page) asks whether a proposed action is valuable, whether it is supported and authorised (the authorising environment), and whether the organisation can deliver it (operational capacity). That maps onto Holagram's text, means and act columns and gives practitioners a vocabulary they already use. It sharpens the next-step framing. It threatens the method in one respect: it casts the authorising environment as something a manager reads and manages, where Holagram would make it contestable by others, and Rhodes and Wanna (mandates-14) challenge that framing for Westminster-type systems. The blurb frames the problem as "inconsistent and fickle political mandates".
- Matt Andrews, Lant Pritchett, Michael Woolcock (2016). Managing Your Authorizing Environment in a PDIA Process (CID Working Paper no. 312). Center for International Development, Harvard Universitycheckedmust readopen access
A practitioner working paper that treats authority as something to be built and managed. The abstract says authority is often hard to obtain, is rarely concentrated in one office or person, and is harder to secure when a challenge involves high risk, uncertainty and many actors; it argues that every effort should carry an explicit strategy for its authorising environment, and offers exercises and examples. It is the nearest existing practice to a mandate map, so it is the baseline Holagram has to improve on. What Holagram adds: a Trace for each claimed authoriser, a separation of text, means and act, and a test by a reader at another level. It also supports starting from a specific problem or next step instead of the institution in general.
Read first: Read the whole; the abstract says it ends with practical exercises and examples.
- Lant Pritchett, Michael Woolcock, Matt Andrews (2013). Looking Like a State: Techniques of Persistent Failure in State Capability for Implementation. Journal of Development Studies 49(1) (online 2012; also Harvard CID working paper 239)checkedmust readopen access
Names two failure modes Holagram needs in its own failure log. 'Isomorphic mimicry': the outward forms of functional states and organisations are adopted to camouflage a persistent lack of function. 'Premature load bearing': highly unrealistic expectations are placed on fledgling systems, undercutting learning, legitimacy and support. For the Mandates method they describe the units under study and also the mandate map itself, which can look like analysis and be asked to settle disputes it cannot. The paper's setting is state capability in developing countries, so use it as a mechanism, not as a diagnosis of the audience's governments.
Read first: Abstract: the two techniques and the closing suggestions for undermining them.
- Sue E. S. Crawford, Elinor Ostrom (1995). A Grammar of Institutions. American Political Science Review 89(3): 582-600checkedmust read
The nearest existing method for the step from Trace to Reading. The abstract describes a syntax that sorts institutional text into three types of statement (rules, norms, shared strategies) and outlines methods for operationalising it. The components, recalled from the paper and not itemised in the abstract, are who is addressed, a deontic (required, permitted, forbidden), the aim, the conditions and the sanction. A decree parsed this way becomes a set of checkable statements, each tied to a passage someone can open. Later versions extend the grammar to constitutive statements, the kind that create offices and confer powers (Frantz and Siddiki, Institutional Grammar 2.0 codebook, arXiv 2008.08937; see cross_leads). Tension with Holagram: the grammar presumes statements can be coded; Holagram keeps ambiguity and treats disagreement between coders as data.
Read first: Abstract: the syntax, the operationalisation, and the discussion of pragmatics in complex institutional settings.
- Meggan Collins (written by), for Dark Matter Labs, with Lund Municipality (Lund Kommun) and NetZeroCities (2026). Visioning without execution is just expensive dreaming.. Dark Matter Labs, 'Provocations' (Substack)checkedmust readopen accessdated 2026-06-19
A practitioner provocation, 19 June 2026. What it argues: visioning workshops end with appealing scenarios and no groundwork (what policy must change, who must be in the room, what the future costs against what not acting costs). In Lund, which is in the EU's climate-neutral cities mission with a 2030 target and already had a climate contract, an action plan and six priority areas, 48 staff from almost every municipal department built scenarios from existing initiatives and voted; the top scenario in each of the six areas was then tested against the same five questions: cost of acting, cost of not acting, policy changes needed, where data exists or is missing, and unintended consequences. Its claim is that attaching cost, timeline and policy barrier turns an idea into "a decision someone has to make", and that the most consistent barrier was the lack of a shared evidence base that the right people had built together. The scenarios went to politicians, the public and colleagues; the authors say they are waiting to see what happens and that the method is still evolving; no outcomes are reported. Scope note, not a criticism: the five questions concern cost, policy change, data and side effects, and do not ask who may decide. That is where a Mandates Entry would begin. The piece also carries an AI-use disclosure line (parts drafted with AI tools; judgement stays with the named humans), a coarse version of Holagram's per-register hands note.
Read first: The whole piece (about 8,000 characters), then the five grounding questions and the closing section on what the authors are learning.
- Francis Fukuyama (2013). What Is Governance?. Governance 26(3)checked
A short commentary that reads the quality of the executive branch on two axes, capacity and autonomy, rejecting output measures; it notes that low-income countries are advised to reduce bureaucratic autonomy while high-income ones seek to increase it. For Holagram it separates what is delegated (autonomy) from what can be done (capacity), two things mandate talk usually fuses, and it shows that the same grant of discretion means different things at different capacity, which supports 'no level sums up another'. The same plane appears earlier as despotic against infrastructural power in Mann (1984, European Journal of Sociology; Crossref and OpenAlex abstract seen, not listed). Threat: any such plane invites scoring and ranking, which the method should avoid.
Read first: Abstract: the four approaches to measuring state quality and the two-axis framework.
- Nils Brunsson (1989). The Organization of Hypocrisy: Talk, Decisions and Actions in Organizations. Wiley (Chichester); later edition Copenhagen Business School Presschecked
Separates what an organisation says, decides and does as three outputs that can diverge, and treats divergence as a way of coping with conflicting demands rather than a lapse (argument recalled; the subtitle is confirmed by the review records). For Holagram it sharpens the reading of gaps: a Reading that takes a gap as concealment has to be tested against incapacity, delay and conflicting demands. It also supports saying, for each mandate document, whether it is talk, a decision or an action. It threatens the assumption that gaps exist to be closed.
- John W. Meyer, Brian Rowan (1977). Institutionalized Organizations: Formal Structure as Myth and Ceremony. American Journal of Sociology (September 1977)checked
The abstract describes formal structures that take on institutional rules as myths in order to gain legitimacy, with structures decoupled from each other and from ongoing activities and, in place of inspection and evaluation, "a logic of confidence and good faith". That is the opposite of Holagram's rule that no reading judges itself, and it explains why a published mandate may be a legitimacy signal and not a statement of ability. Use it as the root source for 'ceremony' in the failure log. Caution: the reading is easy to over-apply and can become the paranoid reading Holagram guards against; test it against simpler explanations.
Read first: Abstract: formal structure as myth, legitimacy, and the decoupling of structure from activity.
- Jeffrey L. Pressman, Aaron Wildavsky (1973). Implementation: How Great Expectations in Washington Are Dashed in Oakland. University of California Presschecked
The Oakland case of a federal programme, argued (from memory) from the number of separate agreements joint action requires, so that each extra clearance lowers the chance of completion. For Holagram it supplies a quantity a unit-level Entry can record: how many separate consents the next step needs, each traceable to an MoU, an approval or a delegation letter. Hill and Hupe (mandates-11) criticise the 'dashed expectations' framing for assuming control from above, and the independence of clearances is an assumption. The audience will know this book; the useful part is the count.
- Michael Lipsky (1980). Street-Level Bureaucracy: Dilemmas of the Individual in Public Services. Russell Sage Foundationchecked
Street-level workers decide and ration under resource pressure, and their routines in effect make the policy people meet (argument recalled). It grounds the 'act' level of the Mandates method: the act is where a mandate is exercised, not only where it ends. It supports the rule that no level sums up another, because a unit's terms of reference are not the sum of its acts. Method suggestion: read refusals, queues and referrals as evidence of ability, not only approvals.
- Michael Hill, Peter Hupe (2003). The multi-layer problem in implementation research. Public Management Review 5(4): 471-490checked
Argues that implementation research under-treats the fact that several layers of government take part, and names two pitfalls: reading 'dashed' expectations at one layer as a failure of control or as illegitimate intervention, and assuming the relation between layers is simple and uniform. This is the implementation-research version of Holagram's rule 2, and a reason to treat each upward link between mandate levels as its own Entry. It also answers the cleanest objection to a levels scheme: layers exist, but how they relate has to be found in each case, not assumed.
Read first: Abstract: the multi-layer problem and the two pitfalls.
- Mark Bovens (2007). Analysing and Assessing Accountability: A Conceptual Framework. European Law Journal 13(4): 447-468checkedopen access
Defines accountability narrowly as a relationship between an actor and a forum: the actor has an obligation to explain and justify, the forum can pose questions and pass judgement, and the actor may face consequences. Those four fields (who answers, to whom, about what, with what consequence) are what a mandate map needs for each claimed mandate in order to be contested, and they decide who counts as 'another level' for a Test. Threat: Holagram's tests carry no consequence, so they are not accountability in this sense and should say so.
Read first: Abstract: the definition, then the dimensions used to describe accountability arrangements.
- Archon Fung (2006). Varieties of Participation in Complex Governance. Public Administration Review 66 (supplement 1): 66-75checked
Maps participation mechanisms in a three-dimensional design space: who participates, how participants communicate and decide, and how the discussion is linked to policy or public action. The third dimension is the participation counterpart of the mandate question: what a process is allowed to change. For Holagram it gives a way to classify what a participation or visioning clause actually carries, and a vocabulary for the 'visioning without execution' failure (mandates-05). The better-known antecedent is Arnstein's ladder (Journal of the American Institute of Planners, July 1969; Crossref-verified, listed in cross_leads).
Read first: Abstract: the three dimensions and the design-space idea.
- R. A. W. Rhodes, John Wanna (2007). The Limits to Public Value, or Rescuing Responsible Government from the Platonic Guardians. Australian Journal of Public Administration 66(4): 406-421checkedopen access
The scholarly objection to treating the authorising environment as a management problem. The authors challenge public value in Westminster-type systems, with strong ministers and tight authorising regimes, identify five assumptions including public servants as Platonic guardians, and defend the primacy of party politics. For the Mandates method this is a design constraint: describing a mandate is not advising a unit on extending it, and the body that authorises a unit should be the first reader of that unit's Entry. The setting is Westminster systems, so it applies less directly to others.
Read first: Abstract: the five assumptions and the defence of party politics.
- Jenny M. Lewis (2020). The limits of policy labs: characteristics, opportunities and constraints. Policy Design and Practice 4(2): 242-251 (online 22 December 2020)checkedopen access
A review of what policy labs are and what constrains them: forms from government-controlled to independent, typically small and short-lived, with flexibility as a strength but easy to shut down, defund or ignore, and dependent on political patronage. This is the audience's own position. It implies that a lab's mandate is weak by construction, that the lab's own Entry is the natural first one, and that publishing it is a risk the lab may not be able to take (care rule: outside reader, and the authoriser's knowledge).
Read first: Abstract: forms of labs, then the opportunities and constraints.
- James Mohun, Alex Roberts (OECD Observatory of Public Sector Innovation) (2020). Cracking the Code: Rulemaking for Humans and Machines. OECD Working Papers on Public Governance (12 October 2020)checked
Rules as Code: governments publish an official machine-consumable version of rules alongside the text. The OPSI page lists six principles (transparency, accountability, traceability, appropriateness and appealability, availability and interoperability, security) and flags open questions: the legal status of machine-readable law, version governance, and how coding errors are found and corrected. For Holagram it is a second reader of mandate texts: where two codings of one clause diverge, the text under-determines, and the divergence is data. The principles of traceability and appealability can be borrowed as checks on the map itself. Tension: Rules as Code aims to reduce ambiguity, Holagram keeps it.
Read first: On the OPSI page: the overview, 'Why has RaC emerged?', and the six principles.
- Sofia Ranchordas (2021). Experimental Regulations and Regulatory Sandboxes: Law without Order?. Law and Method (2021); SSRN 3934075checkedopen access
Argues that poor design and implementation of experimental regulations and sandboxes limit their internal validity (unclear whether results come from the intervention) and external validity (results do not generalise), weaken evidence-based lawmaking, and can breach the principle that experimental rules follow objective, transparent and predictable standards. Sandbox conditions are one of the Mandates document types; this supplies the questions to put to them, and a caution that even formal experiments struggle to be testable, which bears on how strong a Holagram Test can be. A companion paper, Ranchordas and Vinci, 'Regulatory Sandboxes and Innovation-friendly Regulation: Between Collaboration and Capture' (SSRN 4696442, 2024; Crossref-verified), treats capture.
Read first: Abstract: the three adverse effects of deficient design.
- United Arab Emirates (constitutional text) (1971; text revised 2009). Constitution of the United Arab Emirates (1971, text revised 2009): Arts 99, 116, 120-123, 125, 149-151. Constitute Project (English text)checkedopen access
A primary text that sets out the five Mandates levels in one document for a federation. Art. 120 lists matters of exclusive federal legislative and executive jurisdiction (it includes issuing licences to aircraft and pilots); Art. 121 lists matters of exclusive federal legislative jurisdiction, including regulating free financial zones, how they are established and how far they are excluded from federal legislation; Arts 116 and 122 leave the Emirates everything not conferred on the federation; Art. 125 requires the Emirates to implement federal laws and lets federal authorities supervise; Art. 99 gives the Federal Supreme Court power to hear disputes between Emirates and the federation, to review a federal law challenged by an Emirate and an Emirate's legislation challenged by a federal authority, and to interpret the Constitution on request, with binding effect. Art. 151 gives the Constitution and valid federal laws priority over Emirate legislation, makes the inconsistent part of the lower provision void to the extent of the inconsistency, and refers disputes to the Federal Supreme Court; Art. 149, an exception to Art. 121, lets the Emirates legislate on Art. 121 matters without violating Art. 151, so even an 'exclusive' list has to be read together with its exceptions (it sits beside Art. 150, on replacing existing legislation, so it may be transitional; that is a question for a legal reader). It gives Holagram a worked example of enumerated against residual competence, of legislating separated from executing, of a conflict rule between levels, and of built-in tests of a mandate claim by another level. Limits: it covers financial free zones only; the legal basis of other free zones is not in this text and was not checked.
Read first: Arts 99, 116, 120, 121, 122, 125, 149 and 151.
- Italian Republic (constitutional text) (1947; text revised 2012). Constitution of the Italian Republic (1947, text revised 2012): Arts 117, 118, 127, 134. Constitute Project (English translation)checkedopen access
A second primary text with the same structure and a time-limited contest. Art. 117 lists exclusive State matters and concurrent matters, which include land-use planning; Art. 118 gives administrative functions to municipalities unless they are attributed elsewhere under subsidiarity, differentiation and proportionality; Art. 127 lets the Government challenge a regional law, and a Region a State or regional law, before the Constitutional Court within sixty days of publication, when a competence is exceeded or infringed; Art. 134 gives the Court jurisdiction over conflicts of attribution. It is the model for the contest window (idea 5) and shows overlapping mandates in planning, which also serves the Cities method. Italian route: the Constitution is the primary text; the English translation used here is the Constitute Project one.
Read first: Arts 117, 118, 127 and 134.
- UAE Government (Official Portal, u.ae) (2024). Regulatory sandboxes in the UAE. u.ae (page 'Updated on: 30 Dec 2024')checkedopen accessdated 2024-12-30 (page update date given on the page)
A government list of six sandboxes with different issuers and instruments: the Cabinet's RegLab, the TDRA's ICT Regulatory Sandbox, the Global TradeTech Sandbox (with the Ministry of Economy, the Abu Dhabi Department of Economic Development and the World Economic Forum), the ADGM RegLab, the DFSA's Innovation Testing Licence programme and the Central Bank's insurance sandbox. For RegLab it gives a start date (January 2019), a basis ("pursuant to a federal law issued in 2018", no number given) and dated acts, including a March 2023 licence for an electric cargo aircraft. It does not say how the sandboxes relate to each other or to federal and emirate law, which is the gap a chain-of-title Entry fills. A suitable first-study document: public, bounded and dated. No view is taken here on any sandbox's performance.
Read first: The RegLab section and the list of sandboxes at the top.
What it implies for the proposals
- Start from one next step, then map backwards. Take a specific step a unit might take (issue a licence, publish a rule, release data, admit a second cohort), not the institution in general, and put Moore's three questions to that step: is it valued, is it authorised, can it be done. Map the documents backwards from the step. This keeps the study to the working definition of a mandate as the ability to take the next step, and to a three-week block. (mandates-01, mandates-02)
- Levels and what each is read from. Act: a dated decision, licence, refusal or instruction (Lipsky's street-level acts; the dated licences on the UAE sandbox page). Unit: the office and its terms of reference, including a lab's own. Authority: the body that issues, funds and can withdraw the unit's instruments. Jurisdiction: the legal order that decides which rules apply (federal, emirate, free zone; State, region, municipality). Polity: the constitutional settlement and the court that rules on competence. Each level carries its own Traces; an authority-level Entry is not the sum of unit-level ones, which is Hill and Hupe's point about layers that do not reduce to one control relation. (mandates-10, mandates-11, mandates-15, mandates-18, mandates-19, mandates-20)
- Documents to read and how. The invisible literature for this method is decrees, statutes, charters, terms of reference, delegation letters, MoUs, sandbox conditions and KPI sheets. For each, the Trace is the clause with issuer, number, date, version and a link, and its legal force is recorded (binding act, programme page, announcement). The first Reading is a statement parse in the institutional grammar: who is addressed, what is required, permitted or forbidden, the action, the conditions, the sanction; later versions (Institutional Grammar 2.0, see cross_leads) separate constitutive statements, those that create an office or confer a power, from regulative ones. Rules as Code suggests a second reading: two readers, one possibly the Adversary, code the same clause independently; where the codings diverge the text under-determines, and the divergence is logged, not resolved. The parse and the code are Readings, never the Trace. Each Entry carries the date and version of the text read, because the OECD primer lists version governance as an open question. (mandates-04, mandates-16)
- Record the competence regime and the upward link at each level. The UAE text gives the federation an enumerated list (Art. 120 for legislation and execution, Art. 121 for legislation only) and leaves the Emirates what is not conferred (Arts 116 and 122); the Italian text lists State and concurrent matters and places administrative functions with municipalities unless attributed elsewhere (Arts 117, 118). The same silence therefore means opposite things under an enumerated and a residual regime, and 'not forbidden, so allowed' is a safe first Reading only under the second; the UAE text also separates legislating from executing, so an Entry names which one it concerns. Even the list has exceptions elsewhere in the text: Art. 149 lets the Emirates legislate on Art. 121 matters without violating Art. 151, which gives federal law priority and voids the inconsistent part of Emirate legislation (Art. 149 sits beside a provision on replacing existing legislation, so it may be transitional). One article is never enough, and a legal reader checks each regime note. Each Entry also links upward to the instrument it derives from, marked traced, unchecked or broke, with two parents or none allowed. The UAE sandbox page shows why: it lists six sandboxes with different issuers or partners (the Cabinet's RegLab; the TDRA; a joint initiative with a ministry, an Abu Dhabi department and the World Economic Forum; ADGM; the DFSA; the Central Bank) and says nothing about how they relate, while the Constitution leaves to federal legislation how far financial free zones are excluded from federal law (Art. 121). (mandates-11, mandates-18, mandates-19, mandates-20)
- The ground to check a mandate against has three columns. Text: what is authorised. Means: money, staff, and the number of separate consents the step needs, each traceable to an MoU, approval or delegation letter (the Pressman and Wildavsky count). Act: dated instances of the unit acting or being blocked, including refusals (Lipsky). Name the authorisers and give a Trace of consent for each, as the authorising-environment exercise asks. A Reading about ability that rests on one column is labelled text-only or act-only. (mandates-02, mandates-07, mandates-09, mandates-10)
- Design the Test at another level from constitutional procedure. UAE Art. 99 lets an Emirate challenge a federal law, lets a federal authority challenge an Emirate's legislation, and gives the Federal Supreme Court binding interpretation on request, while Art. 151 sends disputes over conflicting federal and Emirate provisions to the same court; Italy's Art. 127 lets the Government challenge a regional law, and a Region a State or regional law, within sixty days of publication on grounds of competence, and Art. 134 names the court for conflicts of attribution. Common features: a claimant at the other level, a named forum, a ground (competence), sometimes a window. A Holagram Test of a unit-level claim is run by a reader at authority level, of a jurisdiction-level claim by a reader from the other jurisdiction; Bovens's four fields (actor, forum, duty to explain, consequence) decide who counts. The result stays 'unchecked' until a competent forum has answered; 'held' means only that a test was run and the claim survived it. (mandates-12, mandates-18, mandates-19)
- Failure modes for the log. Mimicry: a tidy map that looks like analysis. Premature load bearing: an Entry used to settle an authority dispute before it is tested. Ceremony: formal structure accepted on good faith without inspection. Legalism: text read without any act. Capture: an Entry written with the unit it describes. Talk standing in for action. A vision or scenario with no instrument and no owner. Non-lawyer misreading of statute, which a synthetic reader may repeat. And Holagram's own paranoid reading, in which silence or ambiguity is read as concealment: test it against incapacity, delay and conflicting demands. (mandates-03, mandates-05, mandates-07, mandates-08)
- A claim that lacks (b), (e) or (k) cannot be contested by a stranger. (mandates-02, mandates-04, mandates-07, mandates-08, mandates-09, mandates-10, mandates-12, mandates-16)
- First study (three weeks, public material only): RegLab on the UAE sandbox page. Traces: the page's statement that RegLab began in January 2019 under a federal law of 2018 authorising the Cabinet to grant temporary licences (no number is given, so finding the law is the first task); the dated act it lists, a March 2023 licence for an electric cargo aircraft; and Art. 120 of the Constitution, which names licensing of aircraft and pilots as exclusive federal legislative and executive jurisdiction. Reading (illustrative): the licence is an exercise of a federal competence through an experimental channel. Test: the text of the 2018 law, read by a legal reader. Expected first-pass result: unchecked. The exercise shows a chain of derivation and takes no view on whether it holds. (mandates-18, mandates-20, mandates-12)
- For labs and offices, and for participation clauses. Make the unit's own mandate the first Entry, written by someone outside it: labs are described as easy to shut down, defund or ignore and as dependent on patronage, and Moore's book is organised around inconsistent and fickle political mandates. Publish an Entry only with the knowledge of the unit's authorising body, and let that body read it first (Rhodes and Wanna's constraint). When a document promises participation or visioning, classify what the process is linked to (who takes part, how they decide, how the discussion connects to policy or action) and ask for an instrument and an owner for the output, using the Lund grounding questions. (mandates-01, mandates-05, mandates-13, mandates-14, mandates-15)
Ideas to take
Chain of title. Write each mandate as a chain of derivation from act to unit to authority to jurisdiction to polity, and test it one link at a time, each link by a reader from the level above. Each link records the competence regime it sits in (enumerated or residual, legislative or executive). A broken link is logged 'broke'; a link with no document found is 'unchecked'.
Changes: The five Mandates levels become links with an upward 'derives from' field instead of five separate lists; the diagram shows chains and their gaps; the failure log records broken links; the regime field blocks the 'not forbidden, so allowed' error.
Risk: It imposes a hierarchy on arrangements that may overlap or have two parents (a zone regulator may derive from both federal and emirate texts); the chain image can hide that. Allow forks, carve-outs and 'no parent found', and keep Hooghe and Marks (held holarchy-14) in view.
Three Traces, one Reading. A Reading about ability must stand on a text Trace, a means Trace and an act Trace; with fewer it is labelled text-only or act-only and cannot be used as a finding.
Changes: Adds a column tag to Traces in the Mandates Entry template and a completeness flag; it operationalises the difference between authorised, able and does.
Risk: Act Traces are scarce in public material and will be filled with press releases; a complete-looking Entry can itself be mimicry.
Grammar parse as the Adversary's first job. A human parses a clause into institutional statements; the Adversary parses it independently; the disagreements are the Test. The passage stays the Trace and stays human-sourced.
Changes: Gives the synthetic Adversary a bounded, checkable task on legal text without letting it supply Traces, and makes coder disagreement visible as data (rule 3).
Risk: A legal reader is needed.
Count the clearances. For each next step, list the separate consents it needs, where each is documented, and how long a comparable step took.
Changes: Adds a recordable quantity at unit level and gives MoUs and delegation letters a defined role as Traces; the comparable step is an act-level check on the count.
Risk: Assumes the clearances are independent; informal routes (Lipsky's discretion) bypass paper ones; a count invites false precision.
Contest window. Publish each mandate Entry with a dated invitation to named actors at other levels to flag it wrong, modelled on Italy's sixty-day challenge window (Art. 127) and on Bovens's four fields. After the window the status is 'held, not contested', which is not 'held, tested'.
Changes: Turns rule 5 into a standing procedure and adds a status distinct from 'unchecked'; defines the 'other level' reader by role, not by availability.
Risk: An invitation can read as an accusation; silence is weak evidence; staff, and labs that depend on patronage, are exposed. Needs the authoriser's knowledge and an outside reader who can call a stop.
Vision-to-instrument Entry. For any vision or scenario document, open a Mandates Entry that asks the five Lund grounding questions and one more: which instrument lets which body decide?
Changes: A standard follow-up Entry for strategy and foresight outputs, which links the Mandates method to the Strategy-making method (move, unit, organisation); 'vision without instrument' becomes a failure-log category.
Risk: Turns imaginative work into compliance and removes useful vagueness; the Fiction register must stay exempt.
Read sandbox conditions as a mandate in miniature: an exemption (which rules are suspended), a scope, a duration, a supervisor and an exit rule. Put Ranchordas's validity questions to each: what would show that this sandbox tests what it says it tests? Use the UAE page's six sandboxes as the sample, one Entry each, within three weeks.
Changes: Gives the Mandates method a bounded, public document type with a built-in time limit and a defined question, which makes a workable first study.
Risk: A reading of a government programme can be taken as criticism of it; stay with public material, describe what each page says, and make no claim about any sandbox's performance.
Name the method's own failures in the field's words. Add 'mimicry', 'premature load bearing' and 'ceremony' as failure-log tags, and flag any map with no act-level Entry after a set time as probable ceremony.
Changes: Gives the failure log a vocabulary the audience already knows and ties the safeguards to evidence from the field.
Risk: Jargon from development studies and organisation theory; the tags can drift into labels applied to other people's work instead of the method's own.
Counter-arguments
- We already do this. Moore's triangle and the authorising-environment exercises in PDIA are in use among practitioners (mandates-01, mandates-02). The objection: what do the Trace, the other-level test and the failure log add that justifies the overhead? Holagram needs one case in which a Test changed a decision; until then the added value is a claim.
- The map is the mimicry. A clean diagram of mandates can become what a unit shows in order to look diligent; Pritchett, Woolcock and Andrews call this isomorphic mimicry, and Meyer and Rowan describe organisations accepting formal structure on good faith (mandates-03, mandates-08). Premature load bearing is the matching risk: asking a new instrument to settle disputes about authority. And a map that makes a unit legible to outsiders can also be used to control it, which is the warning of Scott's Seeing Like a State (held documents-13).
- It is not a manager's task. Rhodes and Wanna argue that in systems with strong ministers and tight authorising regimes the authorising environment belongs to elected office, and that tools helping officials manage it risk displacing it (mandates-14). A method that teaches a unit to read and test its mandate can be used to stretch it. Holagram needs a rule that Entries describe and do not advise on extension, and that the body that authorises a unit reads that unit's Entry first.
- Ambiguity is for courts, not research teams. Lawyers will say that competence disputes are settled by a competent forum, that interpretation follows rules a non-lawyer misses, and that 'ambiguity stays' is not how a legal order works; both constitutions read here name a court for competence conflicts, and the UAE text shows how one article can be qualified by another (Art. 121's exclusive list against the exception in Art. 149) (mandates-18, mandates-19). The labels held, broke and unchecked will be read as legal findings, so they must say 'unchecked by the competent forum'.
- The act is the mandate. Lipsky's point reverses the method's order: practice at the street makes policy, so a text-first map misdirects effort (mandates-10). The reply that no level sums up another cuts both ways, because act-level evidence is anecdotal and cannot stand for the unit.
- Gaps are not defects. Brunsson treats divergence between talk, decisions and actions as a way of coping with conflicting demands, and closing it may remove what keeps the system working (mandates-07). A map that ranks gaps as failures will be read as an audit, and units will not cooperate.
- The audience cannot use it on itself. Labs are easy to defund and depend on patronage (mandates-15), so publishing a contestable map of one's own authorising environment is a risk they may not be able to take. Fukuyama's two-axis point adds that how much autonomy is wise depends on capacity (mandates-06), so a map cannot be compared across jurisdictions without misleading.
- The tests are too weak to count. Even formal experiments such as sandboxes struggle with internal and external validity (mandates-17), and a three-week reading with an outside reader will be weaker. Bovens's definition of accountability requires consequences (mandates-12); Holagram's tests have none, so they are commentary and should say so.
Gaps
- 99; Italy Arts 127 and 134) and the accountability literature (Bovens), but no empirical work on research-style or adversarial tests of mandate claims by someone outside the system.
- Delegation and principal-agent theory: no source is listed for it. Fukuyama (mandates-06) touches delegation as autonomy, and McCubbins and Schwartz and Hadfield-Menell and Hadfield are in cross_leads, but the standard review, Bendor, Glazer and Hammond (2001), 'Theories of Delegation', Annual Review of Political Science (Crossref record seen; not read), is the place to start. Weber's types of authority and Beetham on legitimacy are also only in cross_leads or unsourced.
- A search for 2023 to 2025 journal articles that map an innovation unit's mandate in this way returned none.
- Whether a synthetic Adversary can parse a decree reliably is untested.
- The Constitution text read here covers financial free zones only (Art. 121); the legal basis of other free zones was not checked, nor whether Art. 149 is transitional and still has effect. The u.ae page gives no number for the 2018 federal law behind RegLab.
- Check the 2022 paper directly (OECD Working Papers on Public Governance, DOI 10.1787/52389006-en).
- A search result listed a 12 June 2026 UAE Cabinet item on an Artificial Intelligence and Data Authority; both attempts to open it returned 404, so nothing is claimed. If the item exists as listed, its founding instrument would suit the Mandates method: a new authority beside existing ones.
- The public record is thin where the method needs it. Delegation letters, MoUs and KPI sheets are mostly not published, so a public-material-only study will lean on statutes, decrees and programme pages, which say least about ability. Act-level evidence (refusals, delays) is the scarcest.
- No Finnish, Estonian or Italian-language scholarly items: the Italian addition is a primary text only. Leads not checked: the Constitution of Finland on municipal self-government (s. 121), the Finnish Local Government Act, and the Estonian Local Government Organisation Act. The European Charter of Local Self-Government, Art. 4, is in cross_leads.