All that glisters is not gold.
The
Merchant of Venice, Act two, Scene seven
Last week I read a recent article authored by Sedran-Price
et. al. titled Indigenous governance of data held by organisations: how to
operationalise Indigenous data sovereignty principles in Australia (link
here).
My initial reaction to the article was primarily shaped by my perception that
the entire implementation strategy proposed for each organisation was to be the
establishment of a series of committees to oversight what is best described as
layers of process. My experience of committees as the primary mechanism for
decision making is that they invariably deliver either a lowest common
denominator ‘solution’ or are dysfunctional. The broader Indigenous policy
domain is populated by innumerable committees that appear designed to produce
bureaucratic sludge and keep issues simmering in a perpetual stasis. I resolved
to write something, and this post is the initial result.
Notwithstanding the extensive impetus within academic and
even government circles (link
here)
advocating or endorsing the merits of Indigenous Data Sovereignty (IDS), to date
I have not written critically about the notion of Indigenous Data Sovereignty
on this blog. My reasons are to an extent mixed and difficult to articulate. It
might be that I have never taken the time to think hard about the merits and
demerits of IDS. It may be that I am not inclined towards the technocratic and
quantitative side of social science. Given my focus on policy outcomes over
process, I do have a view that data and data management is in some senses a
second order issue, though I am very conscious that the way data and
information is used disseminated and/or withhold has real policy consequences. Perhaps
the real reason for my reticence is that I have an intuitive sense that I
disagree with the notion that the management of key statistical metrics within
the public realm should be placed under the control of interest groups, or more
fundamentally the direct political control of governments (link
here).
That said, I acknowledge that the questions we ask and the
means we use to measure our answers can be culturally shaped and framed and
thus are not necessarily neutral. It follows that there is potentially merit in
the underlying aspiration from Indigenous interests and academic supporters of
IDS. Having spent some days thinking
about the issue, I remain of the view that the idea of Indigenous data
sovereignty is an issue that is problematic from both strategic policy and
political perspectives.
To assist my thinking about the issue, I asked Claude
Sonnet 5 to respond to four requests:
[First to] write a
short one or two paragraph summary of the [Sedran-Price] article and/or IDS, two
identify the major strengths in the argument, three identify the major
flaws/problems with IDS from a practical point of view, and four write a
short couple of paragraphs on how the implementation of IDS would actually work
if it was decided to apply it to the national census.
Below are the four responses from Claude:
1. Summary
Sedran-Price et al. argue that
Indigenous Data Sovereignty — the principle that Aboriginal and Torres Strait
Islander peoples should govern data about themselves — has stalled at the level
of aspiration in Australia despite the 2018 Maiam nayri Wingara (MnW)
principles being well established in the academic and policy literature. The
paper's contribution is meant to be operational: for each of the five MnW
principles (control of the data ecosystem; contextual, disaggregated data; data
that empowers self-determination; data structures accountable to Indigenous
peoples; data that is protective of individual and collective interests) it
proposes a concrete "mechanism" — an Indigenous Data Governance
Committee (IDGC) with delegated custodian authority, Indigenous data audits and
asset registers, contextual statements attached to published data, culturally
informed disaggregation, Indigenous Data Management Plans, metadata/provenance
standards (IEEE 2890, Local Contexts TK Labels), and free/prior/informed
consent protocols.
It illustrates each with case
studies — the Mayi Kuwayu Study, the Yarrabah Counts community survey, the
Passamaquoddy digital repatriation in the US — which are real and running, but
which operate at the scale of a discrete research cohort or a single community
partnership, not a national administrative or statistical system.
2. Major strengths
·
It correctly diagnoses a well-documented
problem: aggregated, decontextualised Indigenous data has historically produced
"deficit" narratives that read the effects of colonisation and policy
as intrinsic characteristics of Indigenous people (Walter's BADDR framework).
This is a legitimate and long-standing critique in Indigenous statistics
scholarship, not a novel grievance.
·
It moves past slogan-level advocacy into
enumerated, adoptable mechanisms — a governance committee, a data audit
template, a management plan, a metadata standard — giving institutions
something they could actually put in a policy manual, rather than leaving
"sovereignty" as an undefined aspiration.
·
It doesn't invent the wheel: it explicitly
builds on OCAP (Canada) and the CARE Principles, situating Australian practice
within a converging international movement rather than a one-off Australian
claim.
·
The case studies are genuine, bounded, and
running rather than hypothetical — Mayi Kuwayu in particular shows a
functioning Indigenous data governance committee with real veto power over a
live national cohort study, which is useful proof of concept at that scale.
·
The Indigenous Data Sovereignty /
Indigenous Data Governance distinction (the political principle vs. the
governance machinery that enacts it) and the separation of "governance of
data" from "data for governance" are useful, clarifying pieces
of conceptual scaffolding.
3. Major practical
problems
·
Representation and legitimacy at scale: an
IDGC needs "representation appropriate to the data source," but the
paper never specifies how competing claims to authority are adjudicated —
between urban, regional and remote communities, native title bodies,
community-controlled organisations, and individual voices — especially for
datasets that span all 250+ nations/language groups simultaneously. Nor does it
address how these appointed, non-elected bodies are themselves held accountable
to the people they claim to represent.
·
Scale mismatch: every worked example is a
discrete project (a bespoke health cohort, one community's survey). Scaling
that governance model to whole-of-government administrative collections —
health records, census, justice, child protection — spanning hundreds of
agencies and enormous record volumes is a different order of problem, and the
paper doesn't really engage with how authority, consent and coordination would
work across that scale.
·
Consent in non-optional systems: free,
prior and informed consent (FPIC) works for a voluntary cohort study but sits
uneasily with compulsory data collection (census, hospital admissions, welfare
and justice records) where people are legally required to be counted. The paper
doesn't resolve how FPIC applies when there is no opt-out from the underlying
collection.
·
Legal and statutory conflict: statistical
secrecy, agency independence (e.g. the ABS Statistician's powers under the
Census and Statistics Act) and privacy law are established in legislation. The
paper doesn't explain how community-body "custodian" authority would
sit relative to those existing statutory protections — complementary,
subordinate, or overriding — which is a substantial legislative gap, not just
an administrative one.
·
Resourcing without a funding model: the
paper acknowledges more resourcing is needed but doesn't quantify it or say
where it comes from, against already stretched Indigenous organisational
capacity — creating a real risk that Indigenous Data Governance becomes an
unfunded compliance burden that lands disproportionately on Indigenous staff,
which the paper itself flags as a risk without solving.
·
Authority without alternative
infrastructure: if governance bodies gain the power to restrict release or use
of data but no parallel investment builds an alternative reporting or
infrastructure, the likely near-term effect is less data available for both
government accountability and community planning — an unintended consequence
the paper only gestures at in its final paragraph.
·
Disaggregation vs. comparability: locally
defined indicators (as in Yarrabah Counts) improve local relevance but can
undermine the standardised measures used for national monitoring (Closing the
Gap), funding formulas, and international treaty reporting. That trade-off is
real and isn't resolved in the paper.
4. How would this
actually work applied to the national census?
Applying the MnW principles to
the Census would mean an Indigenous Data Governance Committee holding genuine
decision authority — not advisory input — over how Indigenous-related concepts
and questions are designed (identification, language, kinship and cultural
variables), how remote and community enumeration is conducted, how results are
disaggregated and released, and under what conditions community-level data can
be accessed or repatriated. Some of this already exists in weaker form through
the ABS's Aboriginal and Torres Strait Islander Advisory processes, but the MnW
principle requires shifting that from consultation to actual control — a
materially larger institutional change.
Several of the practical
problems above become sharp here. Disaggregation to "individual, community
and First Nations levels" runs directly into statistical confidentiality:
many remote communities have populations small enough that any fine-grained
breakdown risks re-identifying individuals, which is exactly why current
practice uses cell suppression and perturbation — a real ceiling on how far
disaggregation can go regardless of governance arrangements. A single national
IDGC could not plausibly exercise custodianship over data spanning 250+
distinct peoples in the way a 12-member committee does for one voluntary health
cohort; a federated model of a national body plus many regional or local ones
would need a decision-rights mechanism for when they disagree, which the paper
doesn't specify anywhere. And because the ABS Statistician's independence and
secrecy powers are set in the Census and Statistics Act, meaningful custodian
authority for a community body would likely require legislative change, not just
an internal policy shift — a considerably higher bar than adopting an internal
data management plan. Finally, because the census is a compulsory, universal
collection, free-prior-informed consent could only really apply to secondary
use and release of Indigenous data, not to the primary act of being counted —
meaning full application of the FPIC principle, as stated, isn't actually
achievable within a compulsory census design; at best it would be a partial,
reinterpreted version of the principle.
After giving this response some consideration, I requested
Claude Sonnet 5 to:
draft another section focused
on IDS generally (ie beyond this article) and how it interacts /relates to
mainstream institutional frameworks more generally. You may (or may not) wish
to express a view on IDS as a coherent general philosophy, and I would be happy
if you framed this as issues to be determined or requiring attention or buy in
from broader interests and or the nation’s governing institutions....
Claude responded as follows:
Addendum
Set aside this article's
specific institutional mechanisms and the harder question is whether IDS
amounts to a single, coherent philosophy at all, or whether it spans two quite
different claims that get run together under one banner. At one end sits a negotiated,
political-process account: Indigenous peoples, as citizens with distinct
collective interests and historical claims, seek enhanced authority over
decisions that affect them, won and adjusted through ordinary political and
legislative means — the kind of shared governance built incrementally through
agreement, much as other forms of devolved or co-managed authority have been.
At the other end sits a stronger, "unceded sovereignty" account,
closer to the language UNDRIP itself uses (self-determination, authority,
control) — in which Indigenous authority over Indigenous data is treated as
inherent and prior, something that doesn't need to be granted by settler
political institutions because it was never ceded to them in the first place.
These aren't cosmetic differences. The first account can be built through
statute, agreement, and administrative reform, remains answerable to and
revisable by ordinary democratic processes, and requires broad public and
institutional buy-in as a matter of legitimacy. The second implies a standing
entitlement that exists independently of, and potentially prior to, the
authority of the very institutions being asked to implement it — which is a
fundamentally different, and far more contested, constitutional proposition. The
literature, including this paper, tends to move between the two registers
without quite specifying which is doing the work, which leaves a genuine and
unresolved question about what buy-in is actually being asked for.
That ambiguity matters because
it collides directly with the architecture IDS proposes to operate inside.
Australia's data-holding institutions — the ABS, universities, state agencies —
sit within a system built on parliamentary sovereignty and the equal
application of law, where administrative and statistical independence is itself
a legislated, revisable arrangement rather than a standing entitlement
belonging to any group. Embedding "Indigenous authority" as something
delegated through negotiated agreement is compatible with that architecture;
embedding it as something that precedes and constrains that architecture is
not, at least not without a much larger constitutional conversation than a data
governance framework can settle on its own. This is, in miniature, the same
fault line that ran through the recognition and Voice debates — and IDS
advocates would be entitled to say the data question shouldn't need to wait on
that larger settlement. But the paper's silence on which model of authority
it's assuming means the practical ask (what does an agency actually need to
cede, and to whom) is left underspecified in exactly the place where
institutional buy-in would have to be secured.
Conclusions
The core issues that I would emphasise (beyond Claude’s
analysis that I largely agree with) are as follows:
There are fundamental problems with the under-
conceptualisation of the near ubiquitous aspiration for implementation of
Indigenous Data Sovereignty in academic and bureaucratic circles. As Claude
notes, the arguments for the wider implementation of processes based on
Indigenous Data Sovereignty — based on normative/aspirational principles that
are legitimate on their own terms — are potentially fundamentally at odds with
the currently accepted precepts of democratic governance in modern Australia. Moreover, much data
relevant to the Indigenous policy domain is in effect also relevant to the
mainstream policy domains, a function both of the increasing demographic hybridity
of Indigenous populations, economic activities, and interests and also of the
nature of information itself: it is not entirely ‘public’, but is best
conceptualised as akin to what economists refer to as a public good
characterised as non-rivalrous (my use does not affect your use) and
non-excludable (once released it is difficult to control its use by others). While
private information rights do exist (eg copyright), in policy domains the
importance of public access to information for democratic accountability and
good policymaking suggest that attempts to maintain what is essentially private
control over information requires clear and cogent justification. While there
are categories of Indigenous related information that are clearly able to be
justified as inherently requiring to be confidential or private, the advocacy
in favour of moving towards Indigenous data sovereignty (link
here)
is generally framed in much broader terms.
Even were we to surmount these issues, there are at least
two serious challenges: first, I doubt that governments are actually committed
to implementing these principles in full, and thus they adopt strategies of
adding layers of process that in effect divert Indigenous interests into dead
ends, while ministers and bureaucrats in effect are virtue signalling rather
than locking in institutional reforms. This has the effect of raising
expectations within First Nations interests that are never intended to be met. Telling
communities they have a right to "control the data ecosystem" are a
recipe for perpetual disillusionment. As the census example shows, attempts to
implement such an agenda would inevitably collide with mainstream privacy laws
and would potentially undermine the statutory neutrality of the ABS, changes
that could be leveraged into a political maelstrom and are thus never likely to
be seriously pursued by any government.
Second, there are strategic risks for Indigenous interests
in at least two ways: the opportunity costs involved in dedicating scarce
intellectual resources amongst Indigenous advocates towards devising ways to
measure Indigenous strengths and participating in myriad process-based
committees at the cost of ignoring the hard work of policy reform aimed at
driving achievable substantive reform in areas such as education, health,
employment and housing would be considerable. More importantly, the continued
articulation and advocacy for processes built around implicit claims of
sovereignty that are not widely supported in the wider Australian community
create the preconditions for a backlash against Indigenous interests that, should
it emerge, would extend well beyond erasing the mere control of the ‘data ecosystem’
relating to First Nations citizens.
Finally, there are two observations worth making. First, while
I am deeply sceptical about the feasibility of implementing and making
operational an Indigenous data sovereignty agenda across the span of the Indigenous
policy domain in any meaningful way, and similarly sceptical about promises by
governments and bureaucracies to implement such an agenda, I am not suggesting
that the aspiration itself for control over data and information related to
Indigenous communities is in any way illegitimate. I too wish to control and
exercise individual sovereignty over data related to my life. It is a
legitimate aspiration. But I don’t expect or trust governments to assist me in
that aspiration. First Nations interests should seriously consider what is
realistically possible in relation to the preparedness of governments to
deliver on First Nations aspirations in this area, what they would gain from
succeeding against the potential costs involved, and respond accordingly.
Second, my sense is that there has been a gradual and
inexorable shift in our policy and academic institutions and their academic
publications towards under-valuing substantive independence on issues related
to Indigenous concerns and aspirations. I raise this because (to return to where
I began this post) it strikes me that part of my reluctance to write about
these issues has been the knowledge that my views would rub up against a near
universal consensus that categorises any criticism (real or imagined) of
Indigenous Australians or their policy and political agendas as being either
racist, right wing, or somehow politically incorrect. The fact that the
Indigenous data sovereignty agenda can develop and gain the traction it has without
(to my admittedly limited knowledge) a broader debate about its conceptual
underpinnings (the literature I am familiar with never cites dissenting
perspectives) strikes me as both intellectually flawed and politically
counterproductive. The challenges facing Indigenous Australians and the nation as
a whole in relation to Indigenous policy are too important to be left hostage
to the intellectual hubris of groupthink.
Given my neglect to date in writing about this issue, I
will endeavour to think and read more about the wider issues of Indigenous data
sovereignty and if I find there is more that I can usefully say, I will share
my thoughts in due course.
9 September 2026
This post has
been written with the research assistance of Claude Sonnet 5. The core
intellectual contribution is mine alone.