Governance innovation

Labs, experiments, portfolios, anticipatory governance and their critics.

Governance-innovation practice already holds most of the vocabulary Holagram needs, and its own documents state the limits. The OECD portfolio approach treats innovation as a spread of activities so that some can fail, with learning at portfolio level; UNDP's portfolio pages work with interconnected options, accommodate failures, and call the approach a good start in changing ways of thinking, not a management tool. Experimentalist governance in the EU already has peer review between units. Critics say labs are small, short-lived and dependent on patronage, and that pilots judged yes or no lose their lessons. What the sources do not show is anyone testing a reading of a document at another level or by another author, or keeping a log of readings that broke. The nearest cases are a lab failure report written for a funder, whose interviewee says it depended on trust, and an Estonian programme record that gives a failure share without reasons. For Mandates the useful anchors are Moore's authorising environment, the capability literature (isomorphic mimicry, premature load bearing) and the UAE's layered instruments: the constitution's free-zone clause shows that levels can be carved out, not only nested. The strongest objections are that portfolio logic aggregates by design, that documents-first reading misses political support, and that publishing failure is unsafe where stakes are high. Introduce the method through what practitioners already do, with two additions: a test by someone else at another level, and a log recording the level of each broken reading.

Sources

What it implies for the proposals

Ideas to take

Make the Mandates Trace a pair: the text and the act. Each Entry records the instrument (what it authorises, in which verbs), who holds the next step, and one observable act in the public record; the Test asks whether that act exists.

Changes: Defines the Trace for Mandates and ties the definition of a mandate as ability to act to something checkable; separates authorisation (the text) from capacity (the act).

Risk: The public record is thin, so many Entries will end unchecked; the pair can read like an audit of a named unit, which needs consent and the outside reader's stop right.

In Mandates studies, replace the single outside reader with a reciprocal peer unit: two units in different authorities each test the other's Entries.

Changes: Gives the outside-reader rule an institutional form practitioners already recognise (peer review) and puts the test at unit level, never with the author or the author's parent.

Risk: Reciprocity can become mutual comfort, and units in one jurisdiction share blind spots; needs rotation, a pre-agreed stop right and some outside check on the pair.

Offer a pilot Entry template: a level tag (idea, unit, authorising environment), three Test states, and a line on who owns the next step after the pilot.

Changes: Gives labs and offices a small, usable form for pilots, makes 'unchecked' respectable, and ties a failure to the level where it happened.

Risk: Funders still want pass or fail; the template may be seen as extra paperwork; level tags need worked examples to avoid sounding like jargon.

Add an exnovation Entry type: what was stopped, by which instrument, and what the unit could and could not do afterwards.

Changes: Extends the failure log from broken readings to deliberate endings (labs closed, mandates lapsed, indicators dropped), which this literature treats as part of innovation.

Risk: Closures are politically loaded and records are thin; needs a publication delay and care about naming.

Write one short layered-jurisdiction case in the condensed-novel format: the constitutional allocation, a sandbox rule and one licence, with the Fiction labelled and kept separate.

Changes: Gives the Mandates method a worked example for audiences with a federation, emirate, authority and free zone, and tests whether the five levels hold or need a carve-out rule.

Risk: Legal accuracy; any Reading about named institutions can be taken as a finding; needs a legal reader and strict labelling of Fiction.

Treat a portfolio-level reading as its own Entry: written at authority level, tested by someone outside the portfolio team, and not supported by adding up project Entries.

Changes: Applies rule 2 in the medium practitioners already use (portfolio sensemaking) and gives the steering room a rule for who tests its synthesis.

Risk: Clashes with portfolio logic, which aggregates by design; needs a worked example so that it does not look like a refusal to manage portfolios.

Pilot the Adversary role on an existing sensemaking output: give a synthetic agent a portfolio sensemaking summary and the cited documents, ask for the strongest Test, and log which Tests a human then ran.

Changes: A first concrete use of the two-role protocol with a practitioner setting; the log shows whether Adversary Tests ever break a Reading.

Risk: Agreeableness and flattening in language models (held critic-01, agents-07); data-handling limits inside agencies; the agent must not supply Traces.

Add a standing self-test to every Holagram study: has any Test broken any Reading? If none has, mark the study as possible isomorphic mimicry.

Changes: Makes the failure log accountable to its own function and not only its form.

Risk: Could push authors to stage breaks; the outside reader should judge whether a break is real.

Counter-arguments

Gaps