Holagram
APPLICATION NAME FIXED ADDRESS DATE TICKET THE COUNTER WAITING TIMES THE REPORT Drawing, invented · the form, the counter and the report

Decisions: what a decision looked at, kept with the accounts it rests on

Kind :
A practice, for one decision
First study :
Closed decisions, then one live
Needs :
Records, owners, a keeper, readers
Status :
Proposed, not run

Closed decisions first · then one live one

Proposal · October 2026

A public decision is made on what the deciding body has looked at, and the selection is seldom written down. This proposal is a practice for one decision that keeps the accounts it rests on, each under its owner, and a first study that would show whether it adds anything.

The seven parts

  1. 1 · The opportunity
  2. 2 · The accounts
  3. 3 · The practice
  4. 4 · The first study
  5. 5 · The process
  6. 6 · What it needs
  7. 7 · How we will know

Also in the drawer

Cities · Post-AI planning · Strategy making · Mandates

1 · The opportunity

A decision could also see what its forms have no room for.

The applicant’s account and the counter team’s, kept beside the report the decision is made on.

The report · one case, invented

The form, and the report on waiting times

Fixed address
asked for
Another box
none
The report
waiting times, exact

A year later a change to the service is decided on the report and a consultation summary.

Kept
beside

The accountsEach under its owner

The applicant
has no fixed address
The counter team
knows this and works round it every week
The report
exact, and still leaves out most of what the applicant holds together

What the decision could gain

The one point the decision turned on, written down.

Nobody wrote down that the applicant’s account and the team’s account differed from the report on the one point the decision turned on.

A decision could also see what its forms have no room for, and those it concerns could learn what was done with their accounts. What a deciding body looks at is set by its forms, its categories and its indicators and, more and more, by summaries a model hands it. Those left out cannot say so, and those included cannot tell what became of what they said.

One case. The form asks for a fixed address. The applicant has none, and the form has no other box. The counter team knows this and works round it every week, and nothing it knows reaches the report on waiting times, which is exact and still leaves out most of what the applicant holds together. A year later a change to the service is decided on the report and a consultation summary. Nobody wrote down that the applicant’s account and the team’s account differed from the report on the one point the decision turned on.

The case is invented, as on the Hola page. The pattern is not. The best-documented public failures studied for Holagram, among them an inquiry into damage from gas extraction (2023), a report on how bereaved families were treated after a stadium disaster (2017) and a study of advice to hill farmers after a nuclear accident (Wynne 1992), are cases in which an institution’s account stood as the account of record. How often this happens is not known.

2 · The accounts

In place of levels, accounts under their owners. The people the decision concerns, the staff who carry it out, the deciding office and any model used each give an account of the situation, by name or by role, with where and when it was drawn from. The office says why it asked for each, and a reader from outside the office may add one. Giving a name is never a condition of being heard.

Accounts under their owners, in place of levels

N°OwnerWhat it holdsIn noteAnswers
  1. 01People concerned (in note, answers)By name or by role; a name is never a condition
  2. 02Staff (in note, answers)By role
  3. 03Deciding office (in note, answers)Says why it asked for eachWhat the decision could gain · the one point a decision turns on, written down
  4. 04A model, if used (in note)One more account, with what it was given and asked
  5. 05Added by a readerA reader from outside the office may add one

Who gives an account

  • each with where and when it was drawn from
  • two more agents: a keeper of the accounts, an outside reader

In place of levels

accounts under their owners, in place of levels

a name is never a condition of being heard

  • an account in the note
  • a model’s account
  • answered by its owner

3 · The practice

The noteSigned and dated before the decision, by whoever keeps the accounts

Who

Who was asked, who chose and who was not asked.

Agree

Where the accounts agree.

Differ · over what can be checked

Where they differ over what can be checked.

Differ · in reading

Where they differ in reading.

Turns on

What each takes the decision to turn on.

Answer · each owner

Each owner can say whether it is fair to theirs, and that answer stays beside the note. No model writes or signs the note.

Each owner answers the note; a model’s reading enters as one more account. The decision stays with those who have the authority to make it.

2 · The accounts, in full

A model’s reading enters as one more account, with what it was given and asked stated. Two more agents take part: the keeper of the accounts and an outside reader. No model writes or signs the note.

3 · The practice, in full

The text is the record of the decision: the report it was made on, the consultation report, minutes, the reasons given. The ground is what the people concerned and the staff say of the situation, and what the record shows was done with what they said. For one decision, before options are drawn:

  1. Accounts, each under its owner, as above.
  2. The difference written down first. A short note, signed and dated before the decision by whoever keeps the accounts, says who was asked, who chose and who was not asked; where the accounts agree; where they differ over what can be checked; where they differ in reading; and what each takes the decision to turn on. Each owner can say whether it is fair to theirs, and that answer stays beside the note.
  3. The decision says what it looked at. It names what it turned on, what it did not look at, and gives a reason to each account it does not follow. The decision stays with those who have the authority to make it.
  4. Each owner is told what was done with theirs, including nothing, and why.

The Gram page describes the same practice in general terms; this proposal is how it would first be tried.

4 · The first study

Closed decisions first, then one live one.

Stage one, a retrospective, six to eight weeks: documents and consenting people only. Stage two, one live decision, eight to twelve weeks.

THE CODING

Stub 1 of 4

Choose and fix

Weeks :

1–2

Scope :

Eight to ten closed public decisions

What it gives :

The list of decisions, the coding and the falsifiers, dated.

Choose eight to ten closed public decisions with public records that no inquiry or review has already read. Fix the coding and publish it before any account is read: what counts as something the decision turned on, and as something the record lacks. Write down what would show the practice adds nothing.

Stub 2 of 4

Accounts

Weeks :

3–5

Scope :

People it concerned and staff, with their agreement

What it gives :

Accounts, one note per decision, each owner’s reply on whether the note is fair to theirs.

For each decision, ask people it concerned and staff who carried it out for their account, by name or by role, with their agreement. Read the record. Write the note for each decision afterwards.

AB

Stub 3 of 4

Blind reading

Weeks :

6–8

Scope :

Each note beside a competent consultation report

What it gives :

The comparison, the share of decisions where an account names something the record lacks, and a written result, published whatever it shows.

Readers who do not know which is which compare each note with a competent consultation report on the same decision: can they find what the decision turned on, whose account it leaned on and what it did not look at?

THE NOTE DATED

Stage two

One live decision

Weeks :

8–12

Scope :

A decision not yet taken; four or five accounts before options are drawn

What it gives :

A signed note attached to the record of the decision, an answer log, the reader’s statement, the follow-up, and a result on whether any account changed anything.

The note signed and dated before the decision. An outside reader, appointed and paid from outside the deciding office, who sees the list of accounts, may add one and publishes a statement in every case. An answer log for each owner. A follow-up with the owners three to six months later.

Holagram

First study · Decisions

Closed decisions first · then one live one

A host gets

Retrospective

  • How often the accounts of the people concerned and of staff name something a decision turned on that its record lacks
  • A note for each decision
  • A comparison with the consultation reports
  • A method note that others can repeat

Live decision

  • A signed note attached to the record of the decision
  • An answer log
  • The reader’s statement
  • The follow-up
  • A result on whether any account changed anything

Documents and consenting people only in the retrospective. The decision stays with those who have the authority to make it.

Status : proposed, not run

Stage one, a retrospective, six to eight weeks. Documents and consenting people only.

Stage two, one live decision, eight to twelve weeks. Four or five accounts taken before options are drawn. The note signed and dated before the decision. An outside reader, appointed and paid from outside the deciding office, who sees the list of accounts, may add one and publishes a statement in every case, including that the practice was not followed. An answer log for each owner. A follow-up with the owners three to six months later. A comparable decision in the same body, handled as usual, read beside it.

Care

In the retrospective no live decision is at stake, so no one is exposed by the outcome. Before any host is approached:

  • The outside reader is designed as a role: who appoints, who pays, for how long, how the reader can be removed, and what is published in every case.
  • A legal view, for each jurisdiction, on disclosure, data protection and contacting people again about a closed decision.
  • An owner of record: a named person, or a host institution that signs each study and takes the data-protection role.
  • Ethics review wherever accounts come from people in vulnerable positions, and payment for owners’ time.

5 · The process

The first study, step by step.

The stages as the first study sets them; dates are set with the host.

The first study, step by step · dates [to be set with the host]

  1. Weeks 1–2

    choose and fix · [to be set with the host]

    1. Decisions chosen
    2. Coding published, before any account is read
    3. What would show it adds nothing, written down

    Gives the list, the coding and the falsifiers, dated

  2. Weeks 3–5

    accounts · [to be set with the host]

    1. Accounts asked for, with agreement
    2. Record read
    3. A note for each decision

    Gives accounts, notes, each owner’s reply

  3. Weeks 6–8

    blind reading · [to be set with the host]

    1. Notes compared with consultation reports
    2. Result published, whatever it shows

    Gives the comparison and a written result

  4. Stage two

    8–12 weeks · [to be set with the host]

    1. Four or five accounts, before options are drawn
    2. Note signed and dated, before the decision
    3. Reader’s statement
    4. Answer log

    Gives the note on the record, the answer log, the statement

  5. Then

    three to six months later

    1. Follow-up with the owners

    Gives a result on whether any account changed anything

The text names no workshop: each step is shown as it is written.

6 · What it needs

Aside from budget, and only what the proposal states.

It needs

Torn from the receipt in 4 · the first study

  • public records of closed decisions · weeks 1–2
  • people who agree to give an account · weeks 3–5
  • a keeper · ?
  • readers for the blind comparison · weeks 6–8
  • six to eight weeks
  • a decision not yet taken · stage two
  • four or five owners, paid for their time · stage two
  • a keeper outside the line management of the staff whose accounts are kept · stage two
  • an outside reader · stage two
  • eight to twelve weeks

7 · How we will know

What would show it adds nothing, and what is not on the list.

Written down in weeks 1–2, before any account is read.

The tests that would sink it

  1. 1

    In the closed decisions, the ordinary record already contains the differences that the people concerned and staff name.

  2. 2

    Blind readers find what a decision turned on as well from the consultation report as from the note, and owners find the note no fairer to theirs.

  3. 3

    Across decisions, no decision, plan or account changes once accounts have been answered, and the outside reader never adds an account.

  4. 4

    Owners and readers judge a fluent model summary as fair and as useful as named accounts.

  5. 5

    The more powerful accounts are followed more often, and the note does not change that.

  6. 6

    Taking part exposes people to harm, through disclosure or retaliation, more than it adds anything.

Not on the list

Trust in the deciding body

Trust in the deciding body not rising is not on the list. The practice aims at conduct others can check; trust is given by others, and may move slowly or not at all.

Where it goes wrong

  • The keeper’s sorting is power. Placing a difference under fact, reading or what matters decides how it is treated. Any owner can contest the sorting, the reasons are published, and the outside reader may sort it again. A claim that fails an agreed check is entered as checked and not found.
  • A note as cover. A signed note can be added to a decision already taken. The note is dated before the decision, and the decision gives a reason to each account it does not follow.
  • The office chooses the accounts. It says why it asked for each, the reader sees the list and may add to it, and the note opens with who was not asked.
  • Equal space, unequal power. A ministry’s account beside a resident’s can suggest equal standing where one decides and holds the money. The note says who decides, who holds the resources and which accounts were drawn with help or paid for.
  • Names keep the least safe away. Giving a name is optional, anonymous accounts stay and are marked as such, and staff accounts are entered by role.
  • Exposure. Accounts held by a public body can be reached by requests for information or in a dispute. Owners are told in writing how before they give one, and nothing is promised that the law does not allow.
  • Positions harden. Setting accounts side by side can entrench them. Differences that can be checked are checked together first.

Beside other work

Every part has a precedent. Written evidence to a parliamentary committee is published under its author’s name. An inquiry report sets out each party’s case. A planning committee report summarises the representations and the officer’s answer, and an inquiry may open with a statement of the ground the parties share. Appraisal can be opened up to show how the answer changes with the framing (Stirling 2008); four questions can be put to any expert assessment about what it leaves out (Jasanoff 2003); the circle of those who judge evidence can be widened beyond experts (Funtowicz and Ravetz 1993); and trust can be treated as a matter of relationships and not of delivery alone (Mulgan 2012).

What is claimed is the combination, in a plain form for one decision: accounts kept apart under their owners, the difference written down before anyone decides, a model’s reading received as one account, and the office’s own account treated as one of them. It is a bet that this shows differences a competent consultation report with a minority note would not. The first study is designed to settle that.

Read next