Research that shows its working

Evidence-led research for decisions that can change when the evidence changes.

A public home for research questions, exact findings, limitations and practical routes to challenge or extend the evidence.

Explore research Contribute evidence

Judgement Question Evidence Learning

Repository purpose

Policy research as a revisable public record

Each publication separates what the evidence supports, what it does not support, and what would change the result.

Exact findings

Verdicts retain their scope, confidence and evidence date instead of becoming promotional claims.

Visible limits

Counterevidence, uncertainty and missing measures remain part of the reader experience.

Controlled change

Feedback can trigger review, but does not silently alter an accepted finding.

Published research

Questions, evidence and decisions

Start with the summary, then follow the result to its limits and source notes.

Case study 01 · Three-report series · evidence to 21 July 2026

Australia, AI, jobs and entry/re-entry work to 2035

Question: Will AI and robotics mainly displace or reshape entry and re-entry work—or can they open viable business-creation pathways across age groups?

What is being tested: The series tests the stronger proposition that AI- and robotics-enabled new job and business creation will become the largest and highest-quality labour-market effect for people entering or re-entering work—outweighing displacement and augmentation, aided by targeted support, and remaining viable through a major shock.

Cumulative result: That stronger proposition is not supported on current evidence. New job and business creation is real enough to investigate, but the evidence does not yet show that it is larger than displacement and augmentation, that targeted support can make it dominant and higher quality, or that it would remain viable through a major shock.

Explore the case study

Publication method

Working source. Vetted derivative. Human publication.

01
Filter
Select exact source material and record exclusions.
02
Generate
Build a reproducible candidate.
03
Challenge
Review fidelity, limitations, accessibility and leakage.
04
Decide
The Product Owner accepts one exact bundle.
05
Publish
A human transfers and verifies the accepted files.

Feedback as evidence

Contribute to the public record

GitHub issues are public. Do not submit confidential, sensitive or personal information.

Correction

Identify an exact statement, explain the issue and supply a source.

Review this form

New evidence

Submit evidence that could strengthen, weaken or qualify a finding.

Review this form

Research proposal

Propose a testable method for an accepted evidence gap.

Review this form

Accessibility

Report a barrier in navigation, reading order, contrast or alternatives.

Review this form

Reuse with attribution

Except where otherwise noted, this repository is licensed under Creative Commons Attribution 4.0 International. Cite the author, title, source and licence, and indicate changes.