Behind The Final Report · The Working

The Architecture Beneath

How the analysis was built, and how to check it.


The report is the polished surface. Underneath is a deliberately traceable evidence base, built in Obsidian so that every claim can be followed to a graded source. Source-notes feed domain-collision syntheses, which drill up into Moments, which compose into the report. This is that working, set out, including where close reading corrects the record on the evidence it relies on.

35+
Verified source-notes
6
Domains, each netted out
8
Moments
4
Traceable layers
The Method

Four Traceable Layers

Each layer feeds the next, so a reader can start at the argument and follow it all the way down to a verbatim quote with its page number.

Layer 01

Paper By Paper

One source per note: verbatim quotes with locators, a face-value appraisal, and a grade for relevance, quality and developer interest.

35+ notes
Layer 02

Domain Collisions

One synthesis per domain: where the evidence inside it agrees, contradicts, and finally nets out for this case.

7 syntheses
Layer 03

The Moments

The conceptual argument units the domains drill up into: the genesis, the background, the three challenges, the strategy, the conclusion.

8 Moments
Layer 04

The Report

The four-thousand-word essay, composed from the Moments to the brief's word budgets, in APA 7.

The surface
The Literature Review

Six Domains, Each Netted Out

The chain of claims that must each hold for the promise to hold, decomposed into six bodies of evidence. Each was steelmanned, then stress-tested, then given a single net conclusion.

1 · Attachment & The Mechanism

11 notes
does the Ideal Parent Figure protocol work, and on what?
Brown & Elliott (2016) · Baker (1981) · Ablon & Brown (1993) · Parra (2017) · Mikulincer & Shaver · the Biometric Attachment Test · the Pillar analyses.
Net: the protocol keeps the IPF's content but changes its mechanism, from a contingent, co-created, therapist-mediated process to a pre-rendered one. The link to durable repair is unestablished, not refuted. The project carries the burden of proof.

2 · Imagery & Aphantasia

8 notes
does externalising the image preserve or bypass the mechanism?
Zeman (2015, 2024) · Keogh & Pearson (2018) · Pearson (2019) · Milton (2021) · Dance (2022) · Cavazos (2025).
Net: the accessibility case is genuine, but the evidence leans toward bypass over recruit. For this group the externalised image is scaffold, not mechanism, which suggests the relational delivery is the medicine for everyone.

3 · The AR/VR Medium

verified
does the medium add therapeutic value?
Carl et al. (2019), virtual-reality exposure versus in-vivo, g = −0.07 (not significant) · Deng et al. (2019), PTSD more contested.
Net: the medium adds reach, immersion and standardisation, but no new mechanism. It equals rather than beats in-person exposure, and the evidence base is empirically silent on attachment repair.

4 · The Synthetic Caregiver

8 notes
can a synthetic figure supply corrective attachment safely?
De Freitas (2025) · Fang & Phang (2025) · Zhang (2025) · Laestadius (2024) · Huang & Huang (2025) · Fitzpatrick (2017) · Leff (2013) · the uncanny valley.
Net: one mechanism, two tails. The "feeling heard" lever that comforts is the same lever the harm literature ties to dependence, role-taking and social substitution. No trauma-specific trial of an AI attachment figure exists anywhere.

5 · The Science Bar

4 notes
do the mechanism claims survive scrutiny?
Cahn & Polich (2006) · Lutz et al. (2008) · Schoenberg (2018, developer-authored) · Grossman (2023, polyvagal critique).
Net: a permit/prohibit line. Felt safety as a clinical construct is permissible; "restores vagal tone" or "rewires the brain" as settled science is not. The burden of proof sits with the platform.

6 · Implementation & Australia

14 notes
is it adoptable, and in this real place?
Greenhalgh (NASSS) · Pear Therapeutics · Morozov · Borup · Berg · MacKenzie & Wajcman · Wyatt · Lupton · Productivity Commission (2020) · Better Access evaluation · the TGA · Indigenous data sovereignty · COORDINARE.
Net: a world congruent with the rationale but hostile to its reach. Structural non-adoption, a fragile commercial model, and a solutionist scaling narrative meet an equity inversion: the access pitch and the harm land on the same people.
Where The Argument Is Made

The Eight Moments

The conceptual units the domains drill up into. The genesis records the journey and the literature spiral; the rest map one-to-one onto the report's sections.

0.1

Genesis

The thesis, the voice, and the literature spiral, where every paper opened ten questions and thirty more papers. The backstage record that makes the foreground legible.

1.0

The Innovation & The Care Problem

What Ideal Nurture is, the Australian "missing middle" it addresses, and a balanced reading of the promise and its limits.

2.0

The Sociotechnical System & Setting

The COORDINARE setting, the direct and indirect users, and the specification for the required sociotechnical map.

3.0

Challenge A · Contingency & Clinical Judgement

Concede the personalisation in full, then relocate the wedge from content to delivery: a pre-rendered scene cannot match affect as it arises.

4.0

Challenge B · Engineered Dependence & Equity

The dependence pathway, concentrated on the underserved, with the datafication and Indigenous strands. The required equity challenge.

5.0

Challenge C · Clearance, Cost & Scale-Up

Even a working tool faces structural non-adoption, and the "click a button" scaling narrative is the clearest anti-deterministic target.

6.0

The Trilemma & Implementation Strategy

The three challenges interlock so the best defence on one worsens another, then a complexity-aware strategy that follows from them.

7.0

Conclusion & Reflexivity

The throughline restated, and an honest account of the limits, including the contested universality of attachment theory and the author's positionality.

Independent Critical Appraisal

The Rigour Trail

Independent appraisal means returning to the primary sources and grading what is actually there, rather than trusting how a body of work describes itself. Doing so surfaced real weaknesses: in the protocol's own founding studies, in the contemplative-neuroscience it leans on, and in how claims about all of it circulate second-hand. Naming them plainly is what separates a critical analysis from a promotional one.

Loose sourcing in the corpus

The Anti-Script Principle, Properly Grounded

Brown's protocol turns on a no-fixed-scripts principle, yet the corpus does not always pin its source, an avoidable looseness in a clinical manual. Returning to the primary text grounds it in Elgan Baker (1981, Chapter 6) and in Brown & Elliott at page 170.

The headline rests on one rater

The Developer Case Series

The protocol's signature result, twelve patients all reaching "earned-secure" attachment, comes from an uncontrolled case series scored largely by the senior author over a mean of about 3.4 years, with an effect size so large (about d = 6.23) that it reads as a single-rater signature rather than a clean finding. Retired in favour of what the evidence can bear.

Overstated in circulation

The One Independent Pilot

Parra (2017) is routinely overstated. A control group attached to it in secondary accounts belongs to a different study; the pilot itself is open, underpowered by its own post-hoc analysis, and uses a different attachment instrument from the developer's. Promising at pilot grade, and no further.

Developer-authored, and duplicated

The Contemplative-Neuroscience

The brain evidence the heritage leans on includes a study co-authored by the protocol's originator, carrying self-selection and no active control, and circulating in the corpus under two names as though it were two findings. Graded as suggestive at most, never as mechanism.

Figures that circulate wrongly

The Disclosure Figures

Summaries of the Zhang (2025) companion-AI study circulate figures that are not in the paper. The primary text gives emotional distress 60.8%, suicidal thoughts 18.0%, substance use 17.5%, and users adopting the bot as a "pretend parent" unbidden, the very design this project proposes.

An over-generous headline figure

Aphantasia Prevalence

The accessibility rationale leans on a "3 to 5 per cent" figure. That holds only as the broad band; near-total aphantasia, the case the pitch implies, is closer to 1 per cent (Dance 2022; Zeman 2024). Narrowed to what the data support.

Primary record, not the press

The Commercial Precedent

The digital-therapeutic flagship is often cited as proof of a pathway. Its primary record, Pear Therapeutics' Securities and Exchange Commission filing, shows the opposite: Chapter 11 on 7 April 2023, around 92 per cent of staff cut, roughly 5.7 million dollars cash against 130 million in liabilities.

Misattributed in circulation

The Indigenous Critique

The cultural-validity critique of attachment theory circulates under the wrong names. Corrected to the actual authors (Choate & Tortorelli 2022; Wright et al. 2024), with the Frontiers aphantasia paper to Cavazos et al. (2025).

Held honestly

The Paywalled Holdouts

Where a primary text could not be obtained, two paywalled sources, the notes say so and quote nothing, rather than dress an abstract up as a full reading.

The disclosure that governs all of it: the originator of the protocol was the author's late mentor, and a collaborator conducted the one independent pilot. Both relationships are held openly, and they raise the evidential bar applied here, they do not lower it.

The Throughline

Ideal Nurture's therapeutic effect is not a property of the avatar; it is produced, or lost, in the contingent relational and organisational world the platform enters. The active ingredient is the calibrated relational delivery, not the scene the avatar can render.

That's the working. Here's the report it serves.