The debate is being framed as biology versus ideology. That framing misses the public-governance problem.

Submissions on the Legislation (Definitions of Woman and Man) Amendment Bill are closing. The official Parliament submissions page says submissions close at 11.59 pm on Thursday 2 July 2026 NZ Time, although the Bill history page still contains a metadata line saying submissions are due on 1 July 2026. If you are submitting, the safest practical advice is simple: do not wait.

The public argument about this Bill often starts in the wrong place. It is usually presented as if there are only two options: either sex is biologically real, or everything is ideology; either women’s safety matters, or trans and intersex people matter; either law is clear, or law has been captured by language games.

That framing is too crude for legislation.

Of course sex can matter. It can matter in reproductive medicine, sport, prisons, intimate care, pregnancy care, statistics, family law, sexual violence services and safeguarding.

The real question is not whether sex ever matters, but:

I do not think it does.

The Bill seeks to define “woman” as “an adult human biological female” and “man” as “an adult human biological male”, with corresponding definitions for “male” and “female”. Parliament’s summary says the purpose is legal certainty, protection of sex-based rights, and language in law that reflects biological reality.

That sounds simple until you ask the questions any working legal system has to answer.

A law that cannot answer those questions does not provide legal clarity. It creates a new argument and calls it certainty.

One travelling definition may look clear, but the operational questions still have to be answered institution by institution.

The strongest argument for the Bill deserves to be taken seriously

There is a strong version of the pro-Bill argument. Stable sex categories can help organise privacy, sport, prisons, data, health care, safeguarding and some sex-specific services. If the law no longer has a stable meaning for “woman”, “man”, “female” and “male”, institutions may struggle to coordinate, and women and girls may lose protections built around sexed vulnerability.

That is not a silly argument. It is a recognisable social-order argument. It is also why the Bill should be scrutinised carefully rather than treated as a symbolic gesture. If the concerns are real, the legal tool needs to be real too.

The Attorney-General’s report is important here. It notes that the Human Rights Act already contains sex-based exceptions, including for separate facilities on public decency or safety grounds, some competitive sporting activities, and some highly personal counselling contexts. It also says the Bill does not affect those Human Rights Act exceptions because the Bill defines “woman” and “man” but does not mention or define “sex”. The report describes the Bill’s legal effect as “uncertain and limited”, and identifies a separate age-discrimination problem because “woman” and “man” are defined by adulthood.

That is a serious warning. The Bill is sold as a way to protect women’s spaces, sport and services. But according to the Attorney-General, the legal machinery most relevant to many of those areas is not actually changed in the way many people appear to assume.

Biology does not implement itself

A law can use biological concepts. But biological concepts do not walk into a courtroom, school, prison, pool or sports club and apply themselves.

They need translation.

In medicine, biology becomes operational through history, examination, laboratory testing, imaging, clinical judgement, consent, privacy rules and risk assessment. In law, biological facts become legal facts through evidence, procedure, statutory purpose, thresholds of proof, review rights and safeguards.

That is not pedantry. It is how complex systems avoid arbitrary decisions. Our existing systems have considerable challenges around this, in particular within the family court, criminal law and in the coroner’s jurisdiction.

Health New Zealand uses the term innate variations of sex characteristics, or IVSC/intersex, and says there are more than 40 variations that come under IVSC. It reports that the 2023 Census found 15,039 people aged 15 or older said they were born with a variation of sex characteristics, and notes that IVSC may be identified across the life course.

That does not mean every person is unclassifiable. It does not mean biology is irrelevant. It means that any law claiming to classify people biologically needs to explain how difficult cases are handled.

If the Bill means gametic sex, it should say so and explain how gametic sex is inferred (since gametic sex cannot practically be measured, at will). If it means chromosomal sex, it should say what happens when chromosomes do not align with phenotype, gonadal development or anatomy. If it means sex assigned at birth, it is not simply “biological”; it is an administrative record. If it means what an ordinary person thinks they can see, it is not a biomedical definition at all. It is social perception dressed as science.

The people claiming that they can simply “use their eyes” to differentiate a “man from a woman”, completely misunderstand the problem, what brain interprets about what their eyes are showing them. That particular proposed solution is a threat to men and women, cis, trans or intersex, and is essentially a proposal to completely rearrange how the rule of law works in New Zealand. I won’t elaborate here.

The concern may be real. The tool is still wrong.

One of the easiest mistakes in this debate is to treat every concern raised by supporters of the Bill as bad faith. Or TERF. Or far right. Some arguments are plainly hostile to trans people. Some rely on stereotypes. Some are culture-war slogans. But many people are also expressing concerns about privacy, safety, fairness, children, data quality and women-only services.

Those concerns can be taken seriously without accepting this Bill.

The practical test is not status in the abstract; it is whether the rule solves the institutional problem in front of it.

Start with the actual institutional problem. Then choose the tool that addresses it with evidence, safeguards and the least collateral harm.

Privacy, sport, prisons, health, data and schools each need tools matched to their actual function.

A definition of “woman” and “man” in the Legislation Act does not build safer changing rooms. It does not prosecute voyeurism. It does not fund refuges. It does not write sport-by-sport eligibility rules. It does not place a prisoner safely. It does not improve cervical screening. It does not train teachers. It does not create better data standards. It asks a status definition to do the work of institutional design. This is the same logical fallacy that results in the boom-and-bust change of ill-informed large-scale structural reforms such as South Australia’s Transforming Health and New Zealand’s [insert reform here, e.g. most recently Pae Ora1]

The Bill fails a basic evidence test

Good legislation should pass a basic evidence test.

What is the problem? What evidence shows it? What options were considered? Why is this option necessary? What rights are affected? What costs are imposed? Who implements it? What happens in hard cases? How will success be measured?

The Bill is weak on those questions.

The Attorney-General says its legal effect is uncertain and limited. Health New Zealand’s IVSC material shows why “biological” classification is not always administratively simple. The Law Commission’s Ia Tangata report shows a more careful pathway: separate treatment of gender identity or cultural equivalents and having an innate variation of sex characteristics, plus specific attention to how existing sex exceptions in the Human Rights Act2 should apply.

That is what serious reform looks like: define the legal interest, consult affected groups, work through exceptions, consider consequences, and amend the relevant law directly.

This Bill does the opposite. It inserts a travelling definition and leaves everyone else to work out what it means3.

What a better approach would look like

A better approach would begin with the practical domains.

For facilities, sport, prisons, health, schools, refuges and data systems, Parliament should ask what the actual problem is, what evidence supports it, which institution is responsible, what information is genuinely needed, and what safeguards prevent misuse.

One variable cannot do all jobs. One definition cannot solve every operational problem. A functional legal system needs stable categories, but not necessarily universal categories.

Stable categories can help institutions coordinate, but that does not make one universal category the right tool for every task.

That is not ideological weakness. It is institutional competence.

The one sentence test

Here is the test I would apply to every argument for the Bill:

If the answer is privacy, use privacy tools. If the answer is safety, use safety tools. If the answer is sport fairness, use sport tools. If the answer is data quality, use data tools. If the answer is clinical care, use clinical tools. If the answer is women’s services, use service-specific tools. If the answer is social discomfort with trans people, then say that honestly and do not pretend it is legal clarity. If the intention of the Bill is to set principles, then set principles- don’t create a law with substantial second- and third-order effects, many of which are yet to be fully characterised.

Closing

This Bill is not flawed because biology is irrelevant. It is flawed because biology is not self-executing in law. People are going to see this as a progressive argument. Progressives are going to see it as apologism. It is neither. This is about structural functionalism- creating a functional, least-bad system. It is not about taking sides in an ostensible culture war.

It claims certainty but does not define the test. It invokes women’s safety but does not directly legislate safety4. It invokes sport but does not write sport rules. It invokes prisons but does not improve prison placement. It invokes data but does not design data standards. It invokes children but does not build safeguarding systems. It invokes sex-based rights but, according to the Attorney-General, does not affect the Human Rights Act sex-exception framework that already governs many of the contexts being discussed.

A law can rely on biology. But then it has to say what biology counts, who decides, on what evidence, with what safeguards, and for what purpose.

This Bill does not do that. It should not proceed.

Source notes

• New Zealand Parliament, select committee submissions page: records the proposed definitions, stated purpose and closing date.

• New Zealand Parliament, Bill history page: records the Bill as a Member’s Bill, its progress and the inconsistent submissions metadata noted above.

• Attorney-General’s NZBORA report on the Bill: records the age-discrimination issue, notes that the Bill’s legal effect is uncertain and limited, and explains that the Bill does not affect the Human Rights Act sex-exception framework because it defines “woman” and “man” rather than “sex”.

• Health New Zealand IVSC/intersex page: records more than 40 IVSC variations, the 2023 Census figure of 15,039 people aged 15+ who knew they were born with a variation of sex characteristics, and that IVSC may be identified across the life course.

• Law Commission, Ia Tangata project page: records the 2025 report and recommendations to add separate prohibited grounds for gender identity or its cultural equivalents and having an innate variation of sex characteristics, with recommendations on sex exceptions.

Good law-making is not only about whether a Bill has a worthy purpose. Many Bills do.

The harder question is whether Parliament has enough information to make the law well: what problem is being solved, what evidence supports it, what options were considered, who was consulted, what rights are affected, what safeguards are included, whether implementation is ready, and how the law will be reviewed.

That is especially important for integrity legislation. When a Bill gives the State stronger powers to investigate wrongdoing, the public purpose may be strong. But the safeguards need the same attention as the power.

That is the point of using the Serious Fraud Office Amendment Bill as the series example. The submission does not say modern investigation powers are unnecessary. It says Parliament should ask whether the evidence, safeguards, implementation plan and review loop have been built with the same care as the power itself.

The checklist

A plain-language checklist can be short:

Why urgency is not enough

Urgency can be real. Fraud and corruption investigations may face genuine digital evidence problems. Old law may not fit new technology. But urgency should change the shape of scrutiny, not remove it.

If a Bill is urgent and targeted, Parliament can still ask for a minimum safeguard pack, staged commencement, post-enactment review and a departmental report that answers the hard questions. That is consistent with the Cabinet Office's 2026 expectation that good law-making should support responsible regulation, not merely quick regulation.

The choice is not between doing nothing and passing a Bill as introduced. There is usually a middle path: proceed with the objective, but add the controls needed to make the law workable and legitimate.

The public's job

Most people will not read a Bill clause by clause. But they can ask simple questions:

Those questions are enough to improve a lot of law-making.

Good law-making is not a procedural nicety. It is how public power becomes specific, visible and reviewable before harm occurs.

Takeaway: good law-making asks whether the power, the evidence, the safeguards, the implementation plan and the review loop fit together.

Sources

• Serious Fraud Office Amendment Bill submission page

• DPMC CO (26) 2: Expectations for Good Law-making

• Public filing copy of Power Needs Guardrails, Appendix AZ.

Series navigation

• Previous: Public trust needs receipts

• Preprint/submission page

When something goes wrong in public administration, the usual advice is familiar. Use the Official Information Act. Complain to the Ombudsman. Wait for audit. Go to court. Trust that internal processes will handle it.

Sometimes that advice is right. Often it is incomplete.

The Official Information Act is not an audit system. Ombudsman review is not a complete substitute for proactive reporting. Court challenges usually depend on someone knowing enough, having standing, and having the resources to bring the issue forward. Annual reports can be too high-level. Internal records may exist but never be visible to Parliament. Both the Privacy Commission and Ombudsman in New Zealand are well known for multi-year waits, even for serious complaints raised.

The result is an accountability visibility gap.

This is one of the core points in my Serious Fraud Office submission. A search power can be lawful and still be hard for Parliament or the public to understand if the system does not record enough about how the power is used, what is copied, what is deleted, what is challenged and what is learned.

What receipts look like

Visibility does not mean publishing sensitive operational details.

It means the system can answer basic questions without compromising investigations: how often a power is used, what broad type of power is used, how often urgent processes are invoked, whether privilege claims arise, whether complaints are made, whether unlawful searches are found, whether copied data is retained or deleted, and whether training or protocols change as a result. Public agencies already report some sensitive information-sharing arrangements at an aggregate level: Inland Revenue's 2025 annual report, for example, reports on information sharing with Police, Customs and the Serious Fraud Office.

For intrusive powers, visibility also means a record exists before there is a dispute. Affected people should receive notice where possible. Issuing officers should receive post-execution reports. Agencies should keep audit-ready records. Parliament should receive aggregate data. Review bodies should have enough information to identify patterns.

Without that, the system relies too heavily on litigation after the fact.

Secrecy is not the whole answer

Some secrecy is necessary. Live investigations should not be compromised. Sensitive information should not be exposed. Operational methods may need protection.

But secrecy should not swallow system visibility.

The public does not need a list of live targets to know whether a power is being used often, whether urgent routes are common, whether privilege problems arise, whether complaints exist, whether courts have found problems, or whether deletion rules are working.

That kind of aggregate reporting is how Parliament learns without damaging investigations. It also sits alongside, rather than replaces, agency transparency tools such as the SFO's published OIA responses and annual reports.

The practical test

For any coercive power, ask five questions.

If those answers are unclear, the problem is not only legal. It is architectural.

Takeaway: accountability works best when each mechanism has a clear job and the links between them are visible.

Sources

• Serious Fraud Office Amendment Bill submission page

• Official Information Act 1982

• SFO OIA responses

• IRD annual report page on information sharing with Police, Customs and the SFO

• This post is based on Section 7 and Appendix X of the public filing copy of Power Needs Guardrails.

Series navigation

• Previous: New Zealand has watchdogs. It also has gaps.

• Preprint/submission page

• Next: Good law-making is a public safety system

New Zealand does not lack accountability bodies.

We have Parliament, courts, the Ombudsman, the Auditor-General, the Privacy Commissioner, the Public Service Commission, the Chief Archivist, internal agency controls, media scrutiny and civil society.

That is a lot of watchdogs. But public accountability can still fail in the space between them. Whether they function individually, or together, is a discussion for another day.

One institution sees legality. Another sees records. Another sees privacy. Another sees money. Another sees complaints. Another sees public administration. Each has a piece of the system. No one automatically sees the whole. Because New Zealand has no personal injury torts, not a single one of them is accountable for public safety.

What the system is for

Public integrity is not only about catching wrongdoing. It is also about making sure the State's own powers are lawful, specific, proportionate, recorded, auditable, reviewable and accountable.

That matters for anti-corruption law. It also matters for health data, public records, procurement, research datasets and digital search powers.

The same basic questions keep returning:

• Who has power?

• What is the power for?

• What record is kept?

• Who checks use?

• What can affected people do?

• What does Parliament learn?

If the answers sit in disconnected places, the system can look stronger than it is whilst being the single most important failure mode.

The Serious Fraud Office example

The Serious Fraud Office supports public integrity because serious fraud, bribery and corruption damage public trust. At the same time, its own coercive powers need integrity controls.

That is not a criticism of the agency. It is a basic rule for any public body with intrusive powers- power warrants oversight.

If Parliament gives an agency stronger tools to investigate wrongdoing, Parliament should also make sure the surrounding controls are strong enough: specific warrants, privilege protection, records, minimisation, reporting, audit and review.

That is why the Serious Fraud Office Amendment Bill is more than a technical amendment. It is a test of whether the integrity system updates controls when it updates powers.

The repair

The repair is not to invent a new mega-watchdog. It is to make the links between existing institutions clearer. It is also to make accountability clearer.

The law should say what must be recorded. Annual reports should give Parliament useful aggregate information. Audit should test whether processes work in practice. Courts should not be the first place a problem becomes visible. Official information and complaints routes should be supported by proactive reporting, not treated as the only accountability tools.

This is also where New Zealand's integrity architecture has a visibility gap. The SFO is subject to the Official Information Act and publishes OIA responses, but some Law Officer material sits outside the Act: the Cabinet Manual says the Attorney-General is not subject to the OIA when performing Law Officer functions, and Crown Law explains that much of its work is outside OIA scope because of Law Officer functions or privilege. That does not make secrecy wrong. It makes substitute visibility more important. It does also raise the question of who ensures whether the use of confidentiality, privacy and privelege are being used appropriately, and what accountability mechanisms exist if it’s found not to be used appropriately.

In plain terms: the system should be able to see itself.

Takeaway: public integrity is not one office or one Act. It is a system of powers, records, rights, oversight, reporting and review.

Sources

• Serious Fraud Office Amendment Bill submission page

• Official Information Act 1982

• Cabinet Manual guidance on Law Officer functions and the OIA

• Crown Law proactive release page explaining Law Officer and privilege limits

• This post is based on the public filing copy of Power Needs Guardrails, especially Appendix L and Appendix AT.

Series navigation

• Previous: Anti-corruption powers need receipts

• Preprint/submission page

• Next: Public trust needs receipts

Dr has raised a serious concern about the Journal of the New Zealand College of Clinical Psychologists. Her post reports that NZCCP members were told that Arna Mitchell’s1 peer-reviewed article, He Wero Ano: Don’t Just Tell Me, Show Me How Science and Psychology Are Racist in New Zealand, had been removed from the journal, not because of research fraud, plagiarism, ethical misconduct, or factual error, but because NZCCP Council determined that retaining the article was inconsistent with the College’s values and could perpetuate harm to Māori. (Psychology At The Crossroads)

I left a comment here, but the issue deserves a little more sunlight. The central problem is not whether one agrees with Mitchell’s article. It is not even whether the journal may correct it. Of course it may. The journal has already published a correction notice identifying several factual inaccuracies and clarifications. The problem is whether a professional body can allow a governance decision, framed in values language, to override ordinary editorial processes after publication. (NZCCP Journal)

At the time of writing, the record is oddly ambiguous. The public JNZCCP article page I can access still shows the article, DOI, citation details, and a linked correction notice, and the Zenodo record remains live as a journal article in JNZCCP, volume 35, issue 1, pages 36–64. (NZCCP Journal) That does not make the governance problem disappear. It clarifies it. A scholarly record should not have to be reconstructed from a Substack post, an email to members, a correction notice, a live DOI, and uncertain access on the journal platform. The official record should tell readers plainly what happened.

Publication is not endorsement. A professional journal is not an organisational newsletter. It is a forum in which claims enter the professional conversation and can then be tested, criticised, corrected, rebutted, or rejected. If a published article contains factual errors, the remedy is correction. If there are unresolved serious concerns, the remedy may be an expression of concern. If the work is seriously unreliable, involves misconduct, or has an invalidating defect, the remedy may be formal retraction. If the concern is disagreement with the conclusions, the ordinary remedy is reply, commentary, correspondence, or further publication.

These distinctions are not obscure. COPE’s retraction guidance states that retraction is a mechanism for correcting the literature and preserving its integrity, not for punishing authors. It also states that once an article is posted online it is published, and that retraction is appropriate where the editor no longer has confidence in the results or conclusions because of serious unreliability, misconduct, unethical research, compromised peer review, serious legal issues, or similar defects. COPE also says that where correction would sufficiently address the concern, retraction is not usually appropriate; and where content is removed, that is a rare step, with metadata retained and the notice explaining why the content was removed.

The International Committee of Medical Journal Editors says much the same. NZCCP isn’t a medical organisation, however in instances like this they offer insights into community and professional standards. Matters of debate and evolving science are not errors; they are best handled as letters, correspondence, forum posts, or new publications- i.e with sunlight. Where correction is needed, the notice should cite the original publication, be indexed, and prior versions should be archived. (ICMJE) NISO’s CREC recommended practice makes the record-integrity point in metadata terms: retractions, removals, and expressions of concern must be clearly communicated and discoverable, because unclear status contributes to continued uncritical use of invalidated or altered publications. (NISO)

The governance issue is just as important. A professional body may own a journal. It may appoint editors, set broad purposes, and define the scope of the publication. But ownership is not the same as editorial control over individual articles. ICMJE adopts the World Association of Medical Editors’ definition of editorial freedom: editors-in-chief have authority over editorial content and timing of publication, and owners should not interfere in the evaluation, selection, scheduling, or editing of individual articles. (ICMJE) This isn’t an imagined responsibility and precedent exists, including with the Medical Journal of Australia (MJA), in the relatively recent past.

That principle matters most when the content is uncomfortable. If the owner can intervene after publication because an article is inconsistent with organisational values, the journal’s independence becomes conditional. The next Council may hold different values. The next controversy may involve different politics. A process that feels protective in one case may become coercive in another.

Values do matter. Te Tiriti, equity, inclusion, cultural safety, and the prevention of harm all matter. But values are not a substitute for process. A values-based organisation still needs rules about who decides, what standard is applied, what evidence is considered, what process is afforded to the author, how conflicts are managed, and how the public record is preserved. Without those rules, “values” become a post-publication veto.

Procedural fairness is not an administrative nicety. If an author’s published article is to be corrected, retracted, or removed, the author should know what concern is being investigated, what standard is being applied, why lesser remedies are inadequate, who is making the decision, and what opportunity exists to respond. COPE’s 2025 guidance says authors should be notified before retraction and told why the editor investigated, why confidence was lost, and why the concern could not be resolved by correction.

The JNZCCP author guidance I have seen describes the journal as an open-access forum for original research, reviews, case studies, opinion pieces, and other work relevant to clinical psychology in Aotearoa New Zealand. It describes anonymous peer review and gives submission instructions. It also says the journal retains rights to archive the work and make it available through indexing and abstracting services. What appears much less developed is the public-facing post-publication machinery: correction, expression of concern, retraction, removal, appeal, and separation between editorial and governance functions.

That is why the related Academic Renewal article on a journal code of conduct is timely. It argues that journals should be able to demonstrate genuine editorial responsibility: who makes editorial decisions, what role the editorial board actually plays, how conflicts are handled, what happens when errors are found after publication, and how the scientific record is maintained. () This should not apply only to publicly funded APC journals. It should apply at least as strongly to journals owned by professional bodies, especially where membership, reputation, regulation, and professional politics are closely entangled.

A Digital Object Identifier (DOI) is not a disposable link. It is part of a provenance trail. DataCite’s guidance is explicit that DOI names and metadata continue even if underlying content is removed, and that retracted content should resolve to a tombstone or archival page rather than simply vanishing. (DataCite Support) If a journal tries to make an article disappear while the DOI record persists elsewhere, it does not protect the record. It fractures it.

The remedy here is not complicated. NZCCP/JNZCCP should publish a clear account of who made the decision, under what policy, whether the editors made the final editorial decision, what process was afforded to the author, what status the article now has, and why correction, reply, commissioned commentary, or expression of concern were considered inadequate. If the article is corrected, say so. If it is retracted, issue a proper retraction notice. If content is removed, retain the metadata and explain the exceptional basis for removal. If the decision was made by Council rather than the editorial process, say that too.

More speech is almost always the better scholarly answer. Publish the correction. Invite a response. Commission a rebuttal. Allow letters. Let readers see the disputed claim, the correction, the critique, and the counter-critique. That is how a profession learns.

The concern is not that values and inquiry are incompatible. It is that values need institutions capable of handling disagreement without collapsing into control of the record. A profession confident in its values should be able to expose contested arguments to scrutiny. A journal confident in its editorial standards should be able to explain its decisions.

A little sunlight would help. Over the next few days I’ll do a lap and come back with a follow-up.

Anti-corruption law has an obvious appeal. If serious fraud, bribery or corruption is hard to investigate, the public interest can require stronger tools.

But the public should ask for receipts.

Not receipts as in money. Receipts as in records, limits, audit trails, reporting and review. If the State gets stronger powers, the public should be able to see, at a safe level of detail, how those powers are controlled.

The Serious Fraud Office Amendment Bill is a useful worked example. Parliament describes it as dealing with search warrants and evidence admissibility. Its practical concern is modern investigation: digital records, search-site control, warrants, assistants and the relationship between Serious Fraud Office powers and other search and evidence laws.

I support the objective. The Government's own explanation is that the Bill is directed at digital and cloud-based evidence. Serious fraud is rarely a box-of-paper problem now. Evidence may sit in cloud accounting systems, email accounts, shared drives, phones, bank records, audit files and messaging platforms.

But stronger tools should not mean vague tools.

The bargain

The bargain should be simple.

If Parliament gives the Serious Fraud Office modern digital search and evidence powers, the Act should also give affected people, courts, Parliament and the public modern integrity controls.

That means at least six things.

First, warrants should be specific. They should identify the system, account, device, data class, date range and offence connection as far as practicable.

Second, copied material should be minimised. Digital search can create durable state-held datasets. The issue is not only entry into a place. It is what happens to copied data afterwards.

Third, privilege and confidential material need practical protection. Fraud investigations can touch lawyers, accountants, auditors, employers, agencies and cloud providers.

Fourth, assistants and contractors need controls. Technical help may be necessary, but the Act should record who assisted, what they accessed and what obligations applied.

Fifth, the public needs aggregate reporting. Operational secrecy may protect live investigations, but it should not prevent Parliament from seeing how powers are used. The Serious Fraud Office already publishes annual reports and corporate documents; the question is whether new coercive powers are reported with enough precision to support legislative learning.

Sixth, the law should include review. Urgent targeted reform should be followed by a statutory check once the new powers have been used.

This is not anti-enforcement

It is easy to frame safeguards as a brake on enforcement. That is the wrong frame.

Good safeguards can make enforcement more legitimate and more durable. A specific warrant is easier to defend. A privilege protocol reduces litigation risk. A deletion rule reduces privacy risk. Audit trails help agencies show that they acted lawfully. Public reporting helps Parliament decide whether the law is working.

The point is not to weaken the Serious Fraud Office. The point is to make sure stronger powers sit inside a stronger public integrity system.

The bigger pattern

This pattern appears across public administration. A new power is proposed because a real operational problem exists. The power is urgent. Safeguards are treated as detail. Reporting is deferred. Review is vague.

That is how gaps become architecture.

The better habit is to ask, every time: what is the matching control?

Takeaway: anti-corruption powers and integrity controls are not opposites. Strong powers need visible limits.

Sources

• Serious Fraud Office Amendment Bill submission page

• Bill text

• Government release on additional SFO tools

• SFO corporate documents and annual reports

• Public filing copy of Power Needs Guardrails, recommendations and Appendix AG.

Series navigation

• Previous: When the State copies too much

• Preprint/submission page

• Next: New Zealand has watchdogs. It also has gaps.

The old search image is physical: an officer enters a place, looks for specified things, seizes them, and leaves. Digital search is different.

The evidence may not be in the place at all. It may be in a cloud account, a software platform, a shared drive, a phone image, a messaging system, an accounting database or a third-party provider. That is the digital evidence problem the Serious Fraud Office Amendment Bill is trying to address.

The search may involve previewing, filtering, exporting, copying, imaging, hashing and later review. That changes the accountability problem.

This is the digital-search bridge back to the Serious Fraud Office Amendment Bill and my submission. Parliament describes the Bill as updating search warrants and evidence admissibility, but the practical problem is broader: once the State can copy and review modern digital material, safeguards need to follow the evidence through its full life cycle.

A copy is not a moment

When paper is seized, the State holds the paper.

When digital material is copied, the State may hold a dataset. It may include responsive material, irrelevant material, privileged material, third-party material, metadata and information about people who are not suspected of anything.

That dataset can be searched again. It can be retained. It can be shared within an investigation. It can create disclosure issues. It can create security and records obligations. It can outlive the urgency of the original search. Overseas surveillance and search examples show why this matters: the United States Department of Justice Inspector General's 2025 report on FBI Section 702 querying emphasised continuing internal controls and oversight, while the United States Supreme Court's 2026 geofence warrant argument showed how location datasets can pull non-suspects into an investigation.

So the integrity question does not end when the warrant is executed. It continues through storage, access, review, retention, deletion and reporting.

The risk is over-collection

A digital warrant should not become a general authority to look through everything available to an account, device or platform. That is the common lesson from the European Court of Human Rights' Big Brother Watch judgment on bulk interception safeguards and the Court of Justice of the European Union's Digital Rights Ireland decision on indiscriminate data retention.

The warrant should be as specific as the investigation allows: account, system, device, data class, date range, offence connection, search method, remote-access method, privilege plan, minimisation plan, retention and deletion plan.

Specificity protects rights. It also helps investigators. It gives the search team a clearer plan, the issuing officer a clearer basis for authorisation, and the court a clearer record if the search is later challenged.

The people in the background

Digital searches can affect people who are not suspects.

An accounting platform may include clients. A law-firm portal may include privileged material. A cloud provider may hold data across jurisdictions. A health or research system may include sensitive information collected for a different public purpose. An employer's records may include staff who have nothing to do with the suspected offence.

That does not mean such material is always off-limits. It means the law should recognise the problem before the search happens.

Applications should identify likely third-party data and explain why the proposed method is necessary. Warrants should include conditions where needed. Post-execution reports should record what happened. Retention rules should require review and deletion of non-responsive material.

The public checklist

For a digital search power, ask:

• What exact data source is being searched?

• What offence connection justifies the search?

• What data classes and date ranges are in scope?

• How will privileged or confidential material be protected?

• What non-responsive material may be copied?

• Who can access the copy?

• When will material be deleted or returned?

• What will be reported to the issuing officer and Parliament?

If the law cannot answer those questions, it is not ready for modern digital search.

Takeaway: digital search creates lasting datasets, so the safeguards must cover the full evidence lifecycle.

Sources

• Serious Fraud Office Amendment Bill submission page

• Bill text

• DOJ OIG report on FBI Section 702 querying practices

• Associated Press report on the 2026 United States Supreme Court geofence warrant argument

• This post is based on the public filing copy of Power Needs Guardrails, especially Sections 8 and 13 and Appendices AD, AJ and BA.

Series navigation

• Previous: Your data was collected for one reason. Who else can use it?

• Preprint/submission page

• Next: Anti-corruption powers need receipts

There is a kind of constitutional change that arrives without a headline saying “constitutional change”. No statute is repealed. No court announces a new doctrine. No official says an independent regulator has been captured. Instead, an appointment is not renewed, a public explanation is given, and the rest of the system understands the message.

That is why the dispute over the Medical Council of New Zealand matters. It is not only about Dr Rachelle Love1. It is not only about cultural safety. It is not only about one Minister, one Council or one draft statement. It is about the quiet constitution of healthcare: the settlement by which Ministers are politically accountable, professional regulators remain arm’s-length, and patients are protected by standards that do not swing with the ideological weather.

On the public record, the first discipline is precision. Te Ao Māori News reported that Health Minister Simeon Brown declined to reappoint Dr Rachelle Love and Simon Watt2 when their terms expired, despite both remaining eligible, and said the Medical Council had become distracted by politics and an ideological agenda. The Medical Council later announced Dr Kenneth Clark as chair and Ms Ming-Chun Wu as deputy chair, while acknowledging Dr Love’s leadership and Simon Watt’s service3. The 2022 Gazette notice confirms the earlier three-year terms.

That means the technically safe formulation is not that the Minister formally “removed” the chair. Under the Health Practitioners Competence Assurance Act 2003, Council members are appointed by the Minister, members may be reappointed but not indefinitely, and removal for misconduct or inadequate performance is a distinct statutory mechanism. The Council elects its chair and deputy from among its members. So the decision can be a non-renewal in law while still ending leadership in fact.

This distinction should not make the controversy disappear. It should sharpen it. Appointment power is not a standards power.

The Council’s statutory purpose is public protection through ensuring practitioners are competent and fit to practise. Its functions include setting standards of clinical competence, cultural competence, including competencies for effective and respectful interaction with Māori, and ethical conduct. That phrase is not activist gloss. It is statutory text4.

The Council’s 2026 consultation on cultural competence, cultural safety and hauora Māori may deserve criticism5. Standards should be clear. They should not require doctors to confess political sins or recite fashionable language. They should regulate professional conduct, communication, judgement and safety. They should not become a loyalty test.

That is the strongest version of the Minister’s defence, and it deserves to be taken seriously. ACT welcomed the decision as a reset and urged the Council to withdraw the draft cultural safety guidance6. Democracy Action similarly argued that respectful care is legitimate but that draft wording about privilege, colonialism and systems risks ideological compulsion7. Those objections cannot be dismissed merely because they are politically inconvenient. A regulator that writes unclear standards can chill doctors just as surely as a Minister can chill regulators. That is why they consult.

But the other side of the ledger is stronger. The subject matter itself is not outside medicine. The HDC Code of Rights gives consumers rights to respectful services that take account of cultural, social and ethnic needs, including Māori needs 8; to services of an appropriate standard; and to effective communication. The Pae Ora Act still speaks of equity, culturally safe and responsive services, and a representative workforce. The Government’s own Policy Statement on Health 2024-2027 includes a skilled and culturally capable workforce9.

The literature is also more serious than the slogan “ideology” suggests. Curtis and colleagues’ 2019 review argued that cultural safety is required for health equity and must be linked to power, bias, organisational accountability and patient-defined safety. Their 2025 refinement records how regulators and training bodies are increasingly incorporating cultural competence and cultural safety. Tipene-Leach and colleagues’ 2024 NZMJ viewpoint traces cultural safety from nursing and medical education into Medical Council standards and specialist training.

Mannes, Thornley and Wilkinson’s work on international medical graduates shows why culture is not abstract: it affects adaptation, vulnerability, complaints, support and retention. And, as I will show in coming weeks, this is not solely a New Zealand perspective nor New Zealand evidence. The focus of Medical Regulation in all high-income countries is public safety. Cultural safety is cited equally in each of them. As a Medical Administrator I see the real consequences, including avoidable harm and death, from a lack of cultural safety. Not as a rare event, but regularly.

The policy problem is therefore not “culture versus patients”. Culture is “what we do around here”. Culture is one of the ways patients experience safety, trust, explanation, consent and follow-up. The problem is how to write standards that are rigorous enough to matter and restrained enough not to become coercive.

That is why the method of intervention chosen by the Minister matters. If the draft guidance was poor, the Minister could have submitted to consultation. If the Council was underperforming, the Act contains review, information and audit mechanisms. If the Government wants to narrow cultural competence obligations, it can amend the Act. If it wants greater power over regulators, it can make that case openly through the HPCA Amendment Bill. Ministry materials on that reform already show the central tension: streamlining and patient focus on one side, regulator independence and appointment-power concerns on the other, as summarised in Putting patients first. In a narrower, specific and different (medicines) regulatory sense, I have also made the argument for balancing upstream with downstream regulation.

Public appointments are not meant to be opaque political lightning strikes. CabGuide sets out appointment and reappointment processes, and Public Service Commission guidance contemplates briefings on expiring terms, eligibility for reappointment, skills and succession planning. Those materials may ultimately vindicate an ordinary governance explanation. But until the appointment papers, recommendations and reasons are released, the public is left with the Minister’s rhetoric and the timing.

The reputational question should be framed carefully. The public record supports institutional disparagement with foreseeable personal spillover more clearly than it proves a personal shaming campaign. Still, when named leaders are not reappointed and the regulator is publicly described as politically distracted, a reputational cloud falls somewhere. Ministers have every right to criticise regulators. They also have a duty to be accurate, evidenced and proportionate, because their office gives their words institutional force. On the evidence, the minister was not only inaccurate, he has sanctioned the Medical Council of New Zealand for executing their core statutory function- not by intent, not interpretively, but by statute:

118 Functions of authorities

(1) The functions of each authority appointed in respect of a health profession are as follows:
…
(i) to set standards of clinical competence, cultural competence (including competencies that will enable effective and respectful interaction with Māori), and ethical conduct to be observed by health practitioners of the profession

That is unambiguous. Parliament decided to include cultural competence in 2003. Minister Seymour Brown in 2026 decided to sanction the Medical Council for executing its statutory function. If I put a red-face emoji here, I’m not sure people will get it. But there should be a red-face emoji here, labelled “Yes, Minister”.

The deeper economic issue is trust. Independent regulation is a commitment device. It tells patients, doctors, overseas regulators, insurers and the public that competence standards are not simply electoral artefacts. If regulators learn that politically exposed standards may cost leaders their positions, they will adapt. This effect is called “crowding out”. They will avoid difficult subjects. Good people may decline appointment. Māori patients and Māori doctors may hear that cultural safety is optional until politics changes10.

The final judgment should be calibrated. The Minister may have acted within the narrow law of non-reappointment. There is no automatic entitlement to another term. Regulators should not become self-perpetuating guilds. Draft standards must be scrutinised, including for clarity, conscience and proportionality.

But a lawful appointment power can still be used in a constitutionally damaging way. Just because something is within the rules, doesn’t mean it was a good idea. The danger is not only what happened to two people. It is the signal sent to every regulator: do not set standards the executive dislikes. That is not consumer or public-centred government. It is standards by appointment threat.

That is to say, the Minister sanctioned Medical Council of New Zealand for undertaking their statutory duty, whilst under consultation about the matters for which he has taken issue with, whilst his Ministry is preparing to submit legislation related to this, and therefore contradicting an existing decision by Cabinet, that unless or until this legislation is passed, this is a matter for parliament to decide.

The better settlement is demanding and democratic: publish the reasons, release the appointment advice, debate the draft standards on their merits, protect doctors from ideological compulsion, protect patients from culturally unsafe care, and preserve the independence of the body Parliament entrusted with competence standards.

Cultural competence should not become an oath of political loyalty. Ministerial appointment power should not become a quiet weapon against statutory standards.

10

This effect and problem is even worse in New Zealand, because health practitioners can be sanctioned by MCNZ, by HDC, by HPDT and in the worst cases, by police. But because New Zealand has no personal injury torts, and because in 1961 New Zealand effectively removed serious maladministration and misfeasance (the only high income country to do so), non-clinical managers, executives, boards and ministers, have virtually no mechanism for recourse. This is the ultimate definition for moral hazard (an economic, not a “woke” term).

Some public datasets exist because people trust the purpose.

Newborn blood spot screening is a good example. A small sample can help screen babies for serious conditions. The public value is obvious: early detection, care, public health and research.

But the same thing that makes the sample valuable also makes it sensitive. People need to know why it is collected, who can use it, how long it is kept, what secondary uses are allowed, and what happens if another arm of the State wants access for a different purpose.

This is not an argument against retaining important health samples. Health New Zealand's current programme guidance treats informed consent, leftover blood spots and storage as real governance questions, not paperwork. The point is simpler: retention and access are different questions.

The trust problem

Public systems depend on participation. The whole is greater than the sum of the individual parts. The difference between the individual parts and the whole, are referred to by economists as “positive externalities” or “spillover benefits”.

People disclose information to health services, schools, researchers, regulators and public agencies because the purpose is limited. If those boundaries become unclear, trust can weaken.

That is not only an individual privacy problem. It is a system problem.

If people fear that information collected for care, screening, research or public administration may later be used for unrelated enforcement purposes without clear limits, they may become less willing to participate, disclose, or support retention. That is not hypothetical. Western Australia had a public backlash after police obtained access to newborn screening cards in a criminal investigation, and New Jersey litigation later exposed police use of baby blood spots in criminal investigations.

That is the chilling effect.

It matters in health because trust is not a soft extra. It is part of how the system works. Here is my go-to example:

Whatever the mechanism, the effect is, people stop telling them things, errors stop being detected and eventually serious adverse events happen.

I’ve now seen this pattern in public health, health care, medicine, finance, human resource, worker health and safety, education, engineering, software, you name it.

Why this links to integrity law

At first, newborn blood spots and Serious Fraud Office search powers look unrelated.

They are connected by a public-administration question: when the State holds sensitive information for one public purpose, what rules govern access for another?

That is why the Serious Fraud Office Amendment Bill matters beyond fraud law. It is the current worked example in my submission: a Bill about modern investigation powers that also tests whether New Zealand updates purpose limits, access controls and public accountability at the same time.

Digital search powers can reach information held by third parties. They can affect people who are not suspects. They can involve medical, professional, financial, research or public-sector records. They can create durable copies that live beyond the initial search.

So a search-power law should not only ask whether investigators can access data. It should ask what data, whose data, under what threshold, with what notice, what minimisation, what confidentiality protection, what retention rule and what reporting.

A better rule of thumb

The better frame is not "law enforcement versus health". It is purpose, authorisation and accountability.

Some access may be justified in exceptional cases. But exceptional access should be exceptional in law, not only in internal practice.

For sensitive repositories, the rules should be clear before a crisis: independent authorisation, narrow scope, minimisation, recordkeeping, notice where possible, retention and deletion limits, and public reporting at a safe aggregate level.

That protects the repository as well as the individual.

Why this belongs here

A Little Sunlight is about public administration, not only misconduct. The quiet machinery matters: records, trust, consent, audit, access rules, review and public explanation. That machinery is called “governance”.

Rare Insights is where I would put a technical health, genomic or evidence-methods analysis. Here, the health example is doing a different job: it shows how public trust depends on visible purpose limits and accountable access rules.

Health systems, research systems and integrity systems all depend on the same foundation. People will tolerate public power when they can see its purpose and its limits.

Takeaway: sensitive public repositories can be worth retaining, but trust depends on clear rules for any access beyond the original purpose.

Sources

• Serious Fraud Office Amendment Bill submission page

• Health New Zealand newborn metabolic screening guidance

• Western Australia newborn screening card legal controversy

• This post is based on Appendix AW of the public filing copy of Power Needs Guardrails, including the newborn blood spot discussion and health/public trust framing.

Series navigation

• Preprint/submission page

• Next: When the State copies too much

That is not a bad thing. A public dashboard can make performance visible, reduce unnecessary secrecy, help people understand what is happening, and give journalists, officials, advocates and citizens a common object to argue about.

But a dashboard is not accountability by itself.

Accountability begins when a public measure can answer four questions:

1. What is the public duty being measured? 2. Who is responsible for acting when the measure worsens? 3. What explanation is owed to the public? 4. What changes if the answer is not good enough?

Without those questions, transparency can become theatre. A metric can be updated every month and still leave nobody answerable. A target can be met while the burden shifts to the people least able to absorb it. A red cell in a table can announce failure without identifying the decision, resource constraint or institutional design that produced it.

The best public dashboards do not merely count activity. They describe a system of responsibility. They show the service promise, the population affected, the decision point, the agency with authority, the uncertainty in the data, and the action that follows.

The weakest dashboards create the feeling of openness while protecting the institution from judgment. They are full of indicators but thin on consequences. They name demand, pressure and complexity, but not responsibility.

For A Little Sunlight, this is the central distinction. Public governance is not only integrity, scandal or misconduct. It is the everyday architecture of public responsibility: how institutions make decisions, allocate attention, explain trade-offs, correct mistakes and earn trust.

So the test for a dashboard is simple:

If the public can see the number but cannot see the responsibility, the dashboard is not yet doing democratic work.

The next step is not fewer dashboards. It is better-governed dashboards: measures tied to duties, duties tied to decision-makers, decision-makers tied to explanations, and explanations tied to correction.

That is where transparency becomes governance.