Rights You Have to Ask For: Autistic Adults and the Universal Declaration of Human Rights

In 1948 the United Nations adopted the Universal Declaration of Human Rights. Thirty articles, written a few years after the world had finished demonstrating what happens without them. ¹

Read them and something about the drafting stands out. Everyone has the right to work. Everyone has the right to a standard of living adequate for health and well-being. Everyone has the right to recognition everywhere as a person before the law. There is no clause about producing evidence. Not one article is conditional on a document.

In 2017, Simon Baron-Cohen spoke at the United Nations for World Autism Awareness Day and walked through a handful of those articles, showing where autistic people were still falling outside them. ² He was right, and the speech has held up. But it was mostly about children, it was mostly built on British numbers, and it treated the failures as a list.

This argument does the same job for adults, and it arrives somewhere else.

The rights are not being denied to us. They are being made available on terms we cannot meet. And they do not fail one at a time.

Part 1:
The paper that comes first

Article 6 says everyone has the right to recognition everywhere as a person before the law. Article 7 says everyone is equal before the law and entitled to equal protection against discrimination. Article 8 says everyone has the right to an effective remedy when their rights are violated. ¹

Now apply those three to a forty-seven-year-old who has never been assessed.

They have no adjustment at work, because an adjustment requires a disclosed and evidenced disability. They have no protection under discrimination law, because that law protects a person with an attribute and they cannot demonstrate the attribute. They have no remedy, because a remedy requires standing and standing requires the attribute they cannot demonstrate. They are not being treated worse than an autistic person. They are being treated as though there is no such person present.

That last one deserves more than a piece of logic, because it is where this argument is most likely to be waved away as a figure of speech, and it is not one.

Australia’s Disability Discrimination Act does not require a diagnosis. Its definition is deliberately wide: disability that presently exists, that previously existed, that may exist in the future, and disability merely imputed to a person by whoever is discriminating against them. ³ Read as written, an undiagnosed autistic adult is covered. The autism presently existed. Nothing in the statute says a clinician must have noticed it first.

But the onus of proof sits with the complainant. ⁴ You must establish the attribute, and then establish that the treatment happened because of it. And the practical way to establish an attribute like this one is a report - the report you did not have at the time, because you did not know, because nobody had told you, because it costs three thousand dollars.

So a claim of this kind does not fail because the discrimination did not happen. It fails before anybody reaches that question. What gets decided is whether you are the sort of person the Act was written for, and it stops there.

Sit with the difference, because from outside the two look identical. A right refused has at least been considered. A claim that fails on the attribute has not - the substance is never examined, never recorded and never counted. Nobody is saying no to you. They are saying that the question of whether somebody else should have said no is not a question you are entitled to ask.

This is the first thing that separates adults from children, and it separates them completely. A child sits inside a system that is obliged to look at them. There are teachers, developmental checks, a school with duties. Nobody is obliged to look at a forty-seven-year-old. They have to go and find someone, and pay them.

In Australia, adult autism assessment is a private purchase. There is no public adult diagnostic pathway in most of the country; you find a private psychologist or psychiatrist, you wait, and you pay somewhere between two and three and a half thousand dollars for the assessment and report. ⁵ My own wait, in Sydney, was eight months.

The research says this is normal rather than unlucky. Huang and colleagues surveyed Australian adults on how they actually got diagnosed and titled the paper Choose your Own Adventure, which tells you most of it: there is no pathway, there are many, and which one you get depends on where you live and what you can pay. ⁶ Cost and the fear of not being believed came up as the two barriers people named most. De Broize and colleagues, interviewing Australian adults about the same process, found it described as confusing, expensive and long. ⁷ Arnold and colleagues, looking more broadly at Australian autistic adults and healthcare, found the same three obstacles running through everything: waiting times, too few clinicians who know anything, and cost. ⁸

You might think a public system fixes this. It is worth looking at what happens when the money is removed and only the queue is left.

In England, autism assessment is free at the point of use through the NHS, and the National Institute for Health and Care Excellence recommends that someone referred for assessment be seen within thirteen weeks. In March 2026 there were 270,701 people in England with an open referral for suspected autism. Of those, 242,708 - 89.7% - had been waiting at least thirteen weeks. Of the people who had already passed thirteen weeks, 3.7% had been seen inside the recommended window. ^{9 10}

So the two systems fail differently and arrive at the same place. Australia rations by price, England rations by queue, and in both countries the median autistic adult is outside the system looking in.

Then there is the part that is easy to miss. The gate does not only decide who gets support. It decides who exists in the data.

O’Nions and colleagues estimated life expectancy for autistic people in the UK using a matched cohort drawn from nearly ten million primary care records, and found reductions of around six years for autistic men and women without a co-occurring intellectual disability, and considerably more for those with one. ¹¹ It is a careful study, and the authors state its main limitation themselves: very few autistic adults have been diagnosed, so the cohort is a fraction of the autistic population, and probably the fraction with the greatest needs.

Read that again with Article 6 in mind. We cannot say what autistic adults die of, because we do not know who they are, because they could not get through the gate that would have told us.

So somebody went and looked in the one place where undiagnosed autistic adults are guaranteed to have been counted.

Cassidy and colleagues obtained the coroners’ inquest records of 372 people who died by suicide in two regions of England, and read them for evidence of autism. Ten per cent showed elevated autistic traits indicating likely undiagnosed autism - eleven times the rate of autism in the UK population. The team then interviewed the families of twenty-nine of the dead, to check what the paperwork had captured. With the families’ evidence included, the figure rose to 41%. Nineteen times the population rate. ¹²

The subject of that study is not autistic people who died. It is autistic people who died without anyone having established that they were autistic.

Two things in it should be sat with rather than read past.

The first is the distance between 10% and 41%. That gap is not a fact about autism. It is a measurement of how much a coroner’s file leaves out, and it closed only because researchers telephoned the families and asked. Autism is not routinely recorded at inquest in England. So the official account of why these people died contains no mention of the thing that, on this evidence, was present in four out of ten of them. They were undiagnosed in life and they stayed undiagnosed in the record of their death.

The second is what the same group had already published. In a specialist diagnostic clinic, they found that 66% of autistic adults had thought about taking their own life, and 35% had made plans or attempts. ¹³ That was 2014, in The Lancet Psychiatry.

Asked about the coroner study eight years later, Baron-Cohen said this: they had published the 2014 data as a wake-up call to governments, and nothing had been done. ¹⁴

That is the man whose address I am answering here, reporting on the result of his own alarm.

Which is the argument of this part at its hardest. The gate is not an administrative inconvenience that delays support by a year or two. Its output is a population of adults who are undiagnosed, unsupported, uncounted, and over-represented among the dead - and who stay uncounted afterwards, because the form has no box for them.

Everything after this part assumes the gate has been passed. Most autistic adults alive today have not passed it. Some of the ones who did not are in that study.

Part 2:
Nobody hands you a right

Here is the second engine, and it is the one I think matters most.

Look at how each article in the Declaration actually reaches a person.

The right to work arrives through a job interview. The right to an effective remedy arrives through a complaint you have to make, in the right forum, in the right words, within a time limit. The right to health arrives through a fifteen-minute appointment in which you must describe what is wrong with you in a register the clinician recognises. The right to social security arrives through an assessment interview. The right to education arrives through an enrolment process and a request for adjustments. The right to a fair trial arrives through your ability to be understood by a court.

Every one of them is delivered over a counter. And in every case, walking up to the counter and asking is a social performance.

That is not an accident of administration. It follows from the kind of society the Declaration was written for and by. Henrich, Heine and Norenzayan gave that kind of society a name - Western, Educated, Industrialised, Rich and Democratic - and their argument was that WEIRD populations are unusual rather than typical, and unusual in a specific direction: individualist, analytic, and organised around the self-directing person rather than the group. ^{15 16} The Declaration is a WEIRD document in exactly that sense. Its subject is an autonomous individual who knows what they are entitled to, notices when they are not getting it, and says so to the right person.

Rights in such a society are not distributed. They are claimed. They sit on a shelf and you are expected to come and get them.

And claiming is the disability.

Not a consequence of the disability. The thing itself. The diagnostic manuals define autism by persistent difficulty in social communication and social interaction - initiating, sustaining, reading, adjusting. Every clause of that definition is a description of what you must be good at to walk up to a counter and ask well.

The evidence on how that goes is unusually blunt. Sasson and colleagues found that non-autistic observers form negative impressions of autistic adults within seconds, from thin slices of behaviour, and become less willing to interact with them - before content, before anything is said. ¹⁷ Milton’s double empathy argument explains the mechanism without making anyone the villain: two groups with different styles failing to read each other, mutually. ¹⁸ Crompton and colleagues showed that information transfer between autistic people is as efficient as between non-autistic people, and degrades specifically in mixed pairs. ¹⁹ The failure is at the join.

Now put that join at every counter where a right is handed over. The interview panel. The GP. The claims assessor. The magistrate. The manager who decides whether your request is reasonable. Each is a mixed-pair interaction in which one party is deciding what you get, and the transfer is known to degrade.

Shaw, Doherty and colleagues gave that its own name in the clinical setting: a triple empathy problem, where the ordinary double empathy gap is widened further by the power imbalance of the consultation and by the specialist language and conventions the patient is expected to already know. ²⁰ The counter is not a neutral version of a conversation. It is the hardest version.

Now add what happens after the interaction has gone badly enough times, because this is the step that is usually left as an assumption and does not need to be.

Doherty and colleagues surveyed 507 autistic and 157 non-autistic adults about getting to a GP. ²¹ Eighty per cent of the autistic respondents reported difficulty visiting one, against 37% of the non-autistic. Then look at which barriers they ranked highest. Difficulty making appointments by telephone, 62%. Not feeling understood, 56%. Difficulty communicating with the doctor, 53%. The waiting room environment, 51%.

And above all of them, at 72%: deciding whether the symptoms warranted a visit at all.

That is the highest-rated barrier in the study, and it happens before any contact with the health system whatsoever. It is not a barrier at the counter. It is a barrier to walking towards it.

The consequences the same study reports are not administrative. Autistic respondents reported untreated physical and mental health conditions, not attending specialist referrals or screening programmes, needing more extensive treatment or surgery because they presented late, and untreated conditions that were potentially life-threatening.

There is no mystery about why the approach stops. van Asselt and colleagues, interviewing autistic adults about rejection sensitivity, recorded the experience as feeling constantly kicked down - an expectation of rejection built from a history of it, which then shapes what you are prepared to attempt. ²² Büttner and colleagues examined whether repeated exclusion makes people hyposensitive or hypersensitive to further exclusion, and the answer is not that you get used to it. ²³ And the oldest description of the pattern is still the plainest: Abramson, Seligman and Teasdale’s reformulation of learned helplessness, in which what is learned from repeated uncontrollable outcomes is not the specific failure but the general uselessness of trying. ²⁴

None of that is unreasonable behaviour. It is accurate forecasting. A person who has been disbelieved by four doctors has good evidence about the fifth.

And here is the trap that closes. When you stop approaching the counter, no record is made of a right denied. A record is made of a person who did not apply. Non-claim is filed as non-need, and the statistics that would show the problem instead show an absence of demand - which is then read, reasonably enough, as evidence that no service is required.

Doherty’s paper ends on the comparison that makes the point better than I can: adjustments for autism-specific needs are as necessary as ramps for wheelchair users. The difference is that nobody has to be persuaded a ramp was needed by counting the wheelchair users who did not come.

That is why I do not think “we are failing to protect the rights of autistic people” quite covers it. Nobody has to fail to protect anything. The rights sit on the shelf, formally universal, genuinely available, and reachable by a mechanism built for a mind that works differently to ours.

What follows is what that does across seven areas of adult life, and what happens when the failures start feeding each other.

Part 3:
What the numbers here are not

Before any of the figures, a correction that has to come first, because without it everything after this reads as an argument that autistic people cannot manage.

Here is what autism actually is, as defined by the people who define it. The DSM-5-TR requires persistent deficits in social communication and social interaction across multiple contexts - all three sub-criteria, no partial credit - together with restricted, repetitive patterns of behaviour, interests or activities, of which sensory hyper- or hypo-reactivity is one. Present from early development. Causing clinically significant impairment. ²⁵ The ICD-11 says substantially the same thing in its own words. ²⁶

Read those criteria again and notice what is not in them. Nothing about employment. Nothing about income, housing, mood, self-harm or life expectancy. Not one clause about being unable to work.

So every number here is measuring something the diagnosis does not describe. That is worth saying out loud, because the shorthand does the opposite. Autistic people are unemployed at six times the general rate is a true sentence that arrives in the reader’s head as a fact about autistic people, when what it records is an outcome that happened to them.

Three things wearing one name

What gets bundled together as “autism outcomes” is at least three different kinds of thing, and they have different causes and different owners.

The constitutional part is the criteria themselves. Communication that runs on a different protocol, sensory equipment calibrated differently, a mind that works by depth rather than breadth. Present from the start. Not anybody’s doing, not curable, and genuinely disabling in a world built to other specifications - I am not going to soften that, and the whole of Part 2 depends on it being real.

The co-occurring part is the set of conditions that travel alongside at elevated rates, some of them sharing biological origins with autism rather than being caused by it. Lai and colleagues pooled 96 studies and put attention deficit hyperactivity disorder at 28%, anxiety disorders at 20%, sleep-wake disorders at 13%, and depressive disorders at 11% - all well above general population rates. ²⁷

The acquired part is what happens to a person who lives as the first thing, in a world arranged for neither. Anxiety built from being wrong-footed daily. Depression built from failing at things everybody else finds automatic. Burnout. Trauma. Botha and Frost’s minority stress work is precisely about this route: harm arriving through social position rather than through neurology. ²⁸

Almost everything measured in these pages sits in the third category, and some of it in the second. Very little of it sits in the first.

The employment figure, taken apart

Take the one that started this - the employment figure, which arrives properly in the next part. What it shows is that a very large share of autistic adults of working age have left the labour market entirely.

The historical record is against reading it as a capacity: autistic people have worked, and worked well, for as long as there have been jobs. What the participation figure records is a behaviour - a large number of people who have stopped presenting themselves for work.

And Part 2 already explained that behaviour without any reference to neurology. Rejection sensitivity, repeated exclusion, learned helplessness - three well-described mechanisms, and not one of them is autism. Withdrawal after enough adverse encounters is a trauma-shaped response. It is what happens to anybody, of any neurology, who is knocked back often enough.

Which changes what the figure is evidence of. Not autistic people cannot work. Something closer to autistic people stop trying to, and we already know who taught them.

I want to be careful about how hard to push that, because the evidence is thinner than I would like and the gap is instructive. Rumball and colleagues found autistic adults reporting PTSD symptoms following a much wider range of life events than the diagnostic criteria for trauma anticipate. ²⁹ But look at what happened in Lai’s meta-analysis: of eleven categories examined, eight had enough studies to pool, and three did not. One of the three was trauma and stressor-related disorders. ²⁷ The category with the best claim to explaining these outcomes is the one nobody has counted properly, which means I can point at it and cannot prove it.

Why this belongs in an argument about rights

It would be easy to file all of the above as pedantry. It is not, and here is the reason.

Naming autism as the cause converts a rights question into a medical fact.

Autistic people don’t work is a statement about a population’s capacities. It closes. There is nothing to be done about it, no one to ask, no obligation attaching to anybody, and the correct response is sympathy. Autistic people are not employed is a statement about employers, interviews, adjustments and the labour market. It opens. It has a subject who is doing something, and that subject can be asked to stop.

The same sentence, differently framed, is either a diagnosis or a complaint.

And this is the version that survives the argument in Part 11 rather than undermining it. If these outcomes were constitutional, no obligation would attach to anybody. Because they are largely acquired, they belong to whoever built the conditions that produced them. The medical framing does not just describe the harm. It quietly relocates responsibility for it, from the environment onto the person, and it does so in language nobody can object to because it sounds like clinical fact.

One honest limit before moving on. I cannot cleanly separate the three layers and neither can anybody else. They interact, they compound, and the research designs that would tell us how much belongs where have largely not been run on adults. What I am claiming is the direction, not the proportions. ³⁰

Part 4:
Work, and the rest you are supposed to get from it

Article 23 gives everyone the right to work, to free choice of employment, and to protection against unemployment. Article 24 gives everyone the right to rest and leisure, including reasonable limitation of working hours. ¹

Take Article 23 first, because the numbers are the strongest anywhere in this argument.

The Australian Bureau of Statistics, in its 2022 disability survey, put the unemployment rate for working-age autistic people at 18.2%. For people with disability generally it was 7.5%. For people without disability, 3.1%. Labour force participation was 50.2%, against 84.9% for people without disability. ³¹ So half of working-age autistic Australians are not in the labour market at all, and of the half who are, close to one in five cannot get work.

Those two rates are usually quoted separately, and quoting them separately hides the answer to the obvious question. Participation tells you who is in the labour market. Unemployment tells you what share of those people cannot find work. Multiply them out and you get the number nobody states: about 41% of working-age autistic Australians are actually employed. ³²

Do the same arithmetic for people without disability - 84.9% participating, 3.1% of them unemployed - and you get about 82%.

Two in five of us are in work. Four in five of everyone else.

Read that with Part 3 in hand. The half sitting outside the labour market entirely are not a clinical finding - they are a withdrawal, with a history behind it.

I want to be clear that the 41% is my calculation, not a published Bureau figure. It is derived by combining two rates the Bureau does publish, and it carries whatever assumptions sit underneath them. ³² But the arithmetic is not controversial, and the reason it is worth doing is that the two published numbers are individually survivable in a way their product is not. An 18.2% unemployment rate sounds like a bad labour market. A 50.2% participation rate sounds like a lifestyle statistic. Together they say that the majority of autistic adults of working age are not working, and that is a different sentence entirely.

The United Kingdom produces the same shape. Its official statistics show autistic people with the lowest employment rate of any disability group, by a wide margin - a figure that has been reported around the low twenties and, on more recent definitions, in the mid thirties, against roughly half for disabled people overall and around eighty per cent for non-disabled people. ^{33 34}

The peer-reviewed work agrees and adds the time dimension. Bury and colleagues followed 2,449 autistic adults across eight annual waves and found that the single largest employment trajectory was stable unemployment: 1,189 people, staying there. ³⁵ Baldwin, Costley and Warren surveyed 130 Australian autistic adults and found people with the capacity and the willingness to work, facing significant labour market disadvantage, working below their qualifications. ³⁶

That last finding is the one that matters for the article as written. Article 23 is not a right to be employed. It is a right to work and to freely choose that employment. A qualified person parked three rungs below their qualification has not been denied a job. They have been denied the choice, which is the part the Declaration actually names.

Now Article 24, which nobody thinks of as an autism issue.

Rest is not the absence of work. It is what has to happen after work for the person to be usable tomorrow. Autistic burnout is the failure of that recovery, and it is now defined rather than merely described. Raymaker and colleagues characterised it as exhaustion of internal resources beyond measure, with loss of function and reduced tolerance to stimulus; Higgins and colleagues, working with autistic people through a Delphi process, produced a similar definition anchored in lived experience. ^{37 38} The consistent feature is that it follows sustained masking and unaccommodated demand, and that ordinary rest does not fix it.

Which produces something the Declaration did not anticipate. Reasonable limitation of working hours is calibrated to a body that recovers at the standard rate. If eight hours of unaccommodated open-plan work costs you what twelve costs someone else, then a legally compliant week is already an overrun, and the limitation that is supposed to protect you has been set at the wrong number. There is no violation to point at. The protection simply does not fit.

And this is where the counter reappears. The fix for both articles is an adjustment, and an adjustment has to be asked for - which requires disclosure, which requires the diagnosis from Part 1, and which then requires you to conduct the conversation with the person who controls your income. Pryke-Hobbes and colleagues found that autistic adults describe work as the setting where masking is least optional, precisely because you cannot choose your colleagues, cannot leave when it becomes too much, and have your income riding on the performance. ³⁹ Cage and Troxell-Whitman found the same: masking is heaviest where the stakes are highest. ⁴⁰

So the mechanism that would deliver the right can only be triggered by the behaviour the disability most impairs, in the room where failure is most expensive.

Part 5:
Health, housing, and the standard of living

Article 25 gives everyone the right to a standard of living adequate for health and well-being - food, clothing, housing and medical care, and the right to security in the event of unemployment, sickness or disability. ¹

Start with medical care, because Part 2 predicts exactly what happens and the evidence obliges.

Arnold and colleagues compared 263 autistic and 70 non-autistic Australian adults on a checklist of barriers to healthcare. The autistic group reported an average of 4.58 barriers each. The non-autistic group reported 0.76. ⁸ The barriers themselves are the ones Part 2 predicts: fear and anxiety about the appointment, difficulty communicating with the provider, and difficulty navigating the system to get to them. None of that is a denial of care. It is a delivery mechanism - a short verbal consultation with a stranger, in a bright room, under time pressure - that is close to the worst possible design for the population using it.

The consequence shows up in the diagnostic record before it shows up anywhere else. Fusar-Poli and colleagues catalogued the missed and mistaken diagnoses that autistic adults accumulate on the way to the right one; Kentrou and colleagues found perceived misdiagnosis of psychiatric conditions to be common in autistic adults. ^{41 42} Years of treatment for the wrong thing is not a failure of access. It is access, working as designed, on a mind the design did not consider.

Then housing, which is where Article 25 stops being abstract.

Churchard, Ryder, Greenhill and Mandy assessed 106 people experiencing homelessness in London and found 12.3% met criteria for autism, with a further 8.5% marginal - having autistic traits falling just short of the threshold. ⁴³ Against a population prevalence of roughly one to two per cent, that is a large over-representation, and the authors note that the autistic group looked different from the rest: more socially isolated, less likely to be using substances. Which is to say the usual explanatory story for homelessness does not fit them, and the housing system is unlikely to have a category for what does.

Put Part 4 next to it and the sequence is not hard to see. Half out of the labour market, one in five of the rest unemployed, and nothing much in between that position and this one.

Part 6:
The instrument built to deliver it

Australia has an answer to all of that, and it is the National Disability Insurance Scheme - the largest disability support system this country has ever built, and worth looking at closely, because it delivers Article 25 and demonstrates the gate at the same time.

Autism is the single most common primary disability in the NDIS. On the Agency’s 2026 figures it accounts for around 35% of participants, some 214,880 people - the largest group in the scheme by a distance. And roughly 78% of them are aged eighteen or under. ^{44 45}

So the scheme that stands for Australia’s commitment to Article 25 is, for autistic people, overwhelmingly a children’s scheme. That is not a scandal in itself - early support is good and I am not arguing against it. But it means the adult autistic population, the one with the employment figures in Part 4 and the mortality figures in Part 11, is largely outside the main instrument built to help. To get in you need a diagnosis, which is Part 1, and then you need to demonstrate permanent and significant functional impairment, in an assessment, by explaining yourself to an assessor - which is Part 2, at a counter, with the whole thing riding on it.

And it is not only that there are fewer adults in it. The supports themselves are shaped for the children. The autism offering Australians recognise - speech pathology, occupational therapy, early intervention - is developmental in design: it exists to alter the trajectory of a child who is still forming one, and it is delivered to the under-nines through a distinct early childhood pathway that has no adult counterpart. An adult can be funded for therapy, but only from inside the scheme, only where it is judged reasonable and necessary, and against a model that was not built with them in mind. For the late-diagnosed there is no trajectory left to intervene in, which is precisely the group’s defining feature and precisely what the instrument has least to say about.

Then there is how you get in, which is where it becomes specific to the people I am writing about.

The Agency assesses access through two lists. List A holds conditions taken to meet the disability requirements of the Act automatically. List B holds permanent conditions where functional capacity varies and further assessment is generally required. ⁴⁶ Autism diagnosed at Level 2 or Level 3 sits on List A. Level 1 sits on List B, which means the applicant must demonstrate substantially reduced functional capacity in at least one prescribed domain before anything else happens.

Now consider who gets diagnosed at Level 1 in their forties and fifties. By definition, the people whose support needs were never obvious enough to be noticed in childhood - which is to say the people who compensated, at the cost documented in the masking literature and in Part 4’s burnout research. ^{39 40}

So the evidence required is evidence of not coping, and the late-diagnosed adult’s history is a record of coping. Thirty years of holding down employment, of raising children, of getting through, reads to an assessor as capacity. It was capacity, in the sense that the tasks got done. It was also the thing that produced the exhaustion, the misdiagnoses and the eventual collapse that sent the person for assessment in the first place - and none of that is legible in the form.

The better you masked, the weaker your claim. That is the whole of it in one administrative rule: the mechanism that delivers the right penalises the adaptation that the absence of the right made necessary.

And even for those who get in, Part 7 will show what tends to arrive is not what was asked for. ⁴⁷

And the instrument is being reshaped as I write this. The National Disability Insurance Scheme Amendment (Securing the NDIS for Future Generations) Bill, introduced in 2026, proceeds from the stated premise that the scheme “was set up to support people with permanent and significant disability, but has grown too much and now covers many Australians with less significant support needs”. ⁴⁸ From 1 January 2028, eligibility will turn on a standardised assessment of functional capacity, and everyone already in the scheme - adults included - will be reassessed against the new criteria progressively over the following three years.

Most of the public argument about that reform has concerned children, and most of it should. The largest single change removes children aged eight and under with developmental delay or autism and low to moderate support needs from the scheme altogether, moving them to a new foundational supports programme. That is not my subject and I am not going to borrow its numbers. ⁴⁹

What does belong here is the sentence underneath it. Adults already in the scheme will have to establish, again, on a new standard, to an assessor, that their functional capacity is reduced enough to qualify. That is Part 7’s transaction repeated on a timer. The gate in Part 1 was one you passed once. This rebuilds it as a gate you stand in front of periodically, for as long as you continue to need what is on the other side.

Part 7:
What you hand over to be helped

Article 12 says no one shall be subjected to arbitrary interference with his privacy, family, home or correspondence. Article 22 gives everyone the right to social security. ¹

These two are in the same section because in practice they are the same transaction. The second is purchased with the first.

Consider what an adult autism assessment actually requires. Your developmental history, which means your childhood reconstructed in detail. School reports, if they still exist. An informant - usually a parent, sometimes a sibling, or a partner - interviewed separately about what you were like as a child and what you are like now, including things you would not say about yourself and may not know they think. Your psychiatric history in full. Screening for everything the clinician has to rule out.

Then, if you want support, you do it again for the support system: functional capacity assessments, reports, evidence of what you cannot do, gathered and submitted and periodically renewed.

In Australia it works differently, and I think worse. You are not usually asked to assemble a second body of evidence. You are asked for the first one - the diagnostic report itself, in full. Not a letter from the clinician confirming the diagnosis, not a summary of the findings. The document. ⁵⁰

Sit with what is in it. The informant interview, which is to say your mother’s or your partner’s account of you, written up in the third person by somebody else. The developmental history, reconstructed. The psychiatric history entire, including everything that was raised and ruled out, which stays on the page as a list of things a clinician once thought worth considering. The clinician’s observations of how you presented in the room - your eye contact, your posture, your manner of speaking. Test scores. Often enough, direct quotation of things you said during the assessment, which you said because you had been told to be honest.

None of that was written for an administrator. All of it goes to one.

And the scheme is the gentler case, because that administrator is a stranger in a building you will never enter. Now run the same transaction at work.

The adjustment in Part 4 has to be asked for, and asking means evidencing, and the document you have is the one described above. So it goes to human resources - to a person with a desk, a lunch room, and a working relationship with the manager who decides what you are worth. What gets passed on is supposed to be only the functional part. What was handed over was all of it.

No procedure anywhere sets out to put your mother’s account of your childhood in front of the person who sets your salary. None has to. It is simply the shortest route between an autistic adult and a quieter desk.

And this is the version that makes the point better than repetition would. A second round of paperwork is a burden; you at least choose what goes into it. Handing over the report is not a burden, it is a disclosure, and it is total. You cannot redact your mother’s opinion of your childhood in order to qualify for support as an adult.

I want to be careful here, because none of that is arbitrary in the ordinary sense. Every item has a defensible clinical reason. Informant history genuinely improves accuracy. Functional evidence genuinely prevents misallocation. Nobody designed this to be intrusive.

But look at the totality from the position of the person supplying it. To obtain a right under Article 22, you surrender, in stages, most of what Article 12 protects: your private history, your family’s account of you, your home arrangements, your correspondence with clinicians. There is no version of the transaction where you keep any of it. And the burden is not evenly distributed - a physical disability is largely evidenced by examination, while ours is evidenced by narrative, which means the currency we pay in is disclosure.

There is a second cost, subtler and worse. To qualify you must present yourself in deficit terms. The assessment does not ask what you are good at except to rule things out; the funding application requires a description of yourself as impaired, sustained at length, in writing, repeatedly.

Article 1 of the Declaration - the very first line - says all human beings are born free and equal in dignity and rights. ¹ The support system is not indifferent to your dignity. It requires you to spend it. And it requires the spending from people already carrying documented levels of internalised stigma: Botha and Frost showed minority stress operating on autistic people through exactly this route, with internalised stigma predicting worse mental health outcomes. ²⁸

Huang and colleagues interviewed nineteen Australian autistic adults and four of their support people about what arrived after the diagnosis. Formal support was hard to access, particularly around education and employment; informal support helped, but put strain on the relationships it came from. ⁴⁷

So the full transaction reads: hand over your history, your family’s testimony and your account of yourself as a deficient person, and receive, in the main, something other than what you asked for.

Part 8:
Family, and the fear of being watched

Article 16 says men and women of full age have the right to marry and found a family, and that the family is entitled to protection by society and the State. ¹

This is an adult article. There is no childhood version of it, which is why it is missing from most autism rights writing, and it contains a harm I have not seen named clearly enough.

Pohl and colleagues compared 355 autistic mothers with 132 non-autistic mothers on their experience of pregnancy, birth and parenthood. ⁵¹ The autistic mothers were less likely to feel that professionals had explained things to them adequately, more likely to report anxiety and depression, more likely to find the sensory and social demands of the perinatal system overwhelming, and more likely to feel misunderstood by the people meant to be helping. Two participants had experienced a referral to social services.

Two is a small number and I am not going to inflate it. The finding that matters is not the referral rate. It is that the fear of referral was widespread, and that it changed behaviour - mothers reporting reluctance to be fully honest with professionals because of what disclosure might trigger.

Follow that through, because it is a closed loop. You are less able to explain yourself to professionals in the way they expect. That difficulty is itself sometimes read as evasiveness or flat affect or poor insight. Knowing it may be read that way, you disclose less. Disclosing less makes you harder to read. And the whole thing plays out in the one domain where the consequence of being misread is that your child is taken.

The Declaration says the family is entitled to protection by society and the State. For an autistic parent, society and the State arrive in the same person - the health visitor, the caseworker, the school - and that person is simultaneously the source of protection and the source of the risk. You cannot approach one without exposing yourself to the other.

Everything above is about mothers, and that is not a choice I made. It is the shape of the literature.

Until 2026 there were no peer-reviewed studies at all of how autistic fathers experience parenting. Not few. None. Systematic reviews of autistic parenthood confirmed it; the only prior empirical work was an unpublished doctoral dissertation with nine fathers in the United Kingdom. ⁵² Article 16 concerns founding a family, roughly half the people founding one are not mothers, and the research had not asked them.

Edwards and colleagues asked, this year, and the study is worth reading properly rather than mining. Nineteen autistic fathers across Australia, aged twenty-nine to sixty-one, interviewed by an autistic researcher and free to answer by video, phone or in writing; the whole thing co-produced with four autistic fathers who shaped the questions before anyone was recruited. ⁵² Thirteen of the nineteen had at least one autistic child.

The first thing to report is what the fathers actually said, because it is not what a rights argument would predict. They described joy. Ordinary, specific joy - walking with a daughter and the dogs, inventing silly games, watching your own traits surface in a child. Several described fatherhood as the thing that finally explained them to themselves; one said plainly that he learnt he was autistic through the process of parenting his children and their autism. Several pursued a diagnosis only because becoming a parent made the thing visible. This is a study about being a good father, and I am not going to strip that out to make a point about the state.

But the fifth theme the authors identified is titled parenting in systems stacked against us, and it belongs in this writing.

Fathers described being disbelieved by professionals in the specific idiom autistic adults will recognise. One recalled GPs offering “judgey, archaic views of, you know, well you don’t look Autistic. Or, you know, you can speak.” Another described raising autism with teachers and being able to tell instantly he had been dismissed. One named the cost of the emotional labour that follows: you feel like you’ve got to make them understand… that’s not my job… You don’t ask the person suffering from the thing to educate everyone about it.

And one father, describing the institution that decides who a child lives with, said that had he known he was autistic at the time, he would have dealt with his son’s mother and the family court a great deal more assertively and confidently. That is Part 1 arriving in Article 16: the paperwork gap, showing up in the one venue where the family is adjudicated.

Then there is the finding that closes the loop with everything else here. Fathers described school events, sports days and parent groups as close to unendurable - one calling the prospect of sharing personal stories with dad-strangers in an unfamiliar room “more like multi-layered torture than informal peer support”. And another said what happens next:

Autistic fathers who avoid their children’s special events… may not be disinterested or uninvested… it may be that these events are simply too loud and bright, and there is an expectation of superficial social interaction with strangers, all of which can be very overwhelming. ⁵²

A sensory and social barrier produces an absence, and the absence is read as a father who could not be bothered. That reading is then available to anybody who later has cause to assess the family.

The authors’ own policy recommendation is the sentence this argument has been circling since Part 2. Services, they write, need to support autistic fathers regardless of diagnostic status, reducing reliance on self-advocacy and enabling accommodations from the outset. ⁵² They arrived at it from nineteen interviews about fatherhood. It is the same conclusion Part 2 reached from the shape of the Declaration, and I take the convergence as some evidence that neither of us has invented it.

One caution, which the authors give themselves: nineteen Australian fathers, predominantly white and highly educated, with single fathers and culturally diverse fathers underrepresented. ⁵³ It is the first study of its kind, not the last word on anything. That it is the first is the part relevant here.

There is a version of this at the start of adult life too. Article 16 covers marriage and founding a family, and the route to both runs through a set of unwritten social rules that autistic adults report having learned from nowhere. Stokes and colleagues found autistic participants describing romantic knowledge acquired from no channel at all - not parents, not siblings, not peers, not observation, not media, not sex education. ⁵⁴ Every route by which the rest of the population absorbs the rules had failed to deliver, and the rules were never written down, because for everyone else they did not need to be.

Part 9:
Restraint, arrest, and a Royal Commission

Article 5 says no one shall be subjected to torture or to cruel, inhuman or degrading treatment. Articles 9, 10 and 11 cover arbitrary detention, fair trial, and the presumption of innocence. ¹

Baron-Cohen reached for history here. I do not need to. Australia publishes the current numbers quarterly.

Under the rules governing the NDIS, a restrictive practice is any practice that restricts the rights or freedom of movement of a person with disability. There are five regulated kinds: seclusion, chemical restraint, mechanical restraint, physical restraint and environmental restraint. Chemical restraint means medication given for the primary purpose of influencing behaviour, as distinct from treating a diagnosed condition. Providers must use them only as a last resort, with authorisation, and must report every use. ⁵⁵

In the quarter from January to March 2025, 15,552 NDIS participants were subject to authorised regulated restrictive practices - 2.2% of everyone in the scheme at that time. The count has risen every quarter since reporting began in this form, from 14,577 in late 2023. ⁵⁵

Those are the authorised ones. In the same quarter, providers notified the regulator of unauthorised restrictive practices affecting 2,471 participants by chemical restraint, 2,688 by environmental restraint, 1,054 by physical restraint and 229 by seclusion. ⁵⁵ Unauthorised means the practice was used without the authorisation the law requires - and these figures are what providers reported about themselves.

I have deliberately not said how many of those people are autistic, because the published breakdown by practice type is not given by primary disability and I will not estimate it. What can be said is that autism is the largest primary disability group in the scheme, at around a third of participants. ⁴⁴

Article 5 is not a historical article. It is a quarterly statistic with a rising trend.

Now the criminal justice articles, where the mechanism from Part 2 turns lethal in a different way.

Blackmore and colleagues examined contact with the criminal justice system among adults assessed at a specialist autism service and found rates of contact well above what the population would predict. ⁵⁶ Across the wider literature, the most common interactions are not offences at all: welfare checks, traffic incidents, being reported for behaviour that turned out to be autistic behaviour, and - frequently - autistic people attempting to report a crime committed against them.

That last one deserves its own sentence. A substantial share of autistic contact with police is autistic people trying to use Article 8, the right to a remedy, and being processed as something else.

The vulnerability data explains why the encounter goes wrong. Trundle and colleagues, in a systematic review and meta-analysis, found victimisation of autistic people running far above general population rates. ⁵⁷ Griffiths and colleagues, developing the Vulnerability Experiences Quotient with autistic adults, documented the same pattern across employment, relationships, finances and the law. ⁵⁸ Pearson, Rees and Forster described how interpersonal victimisation is often not recognised as victimisation by the person experiencing it, because the relationship was read as a friendship. ⁵⁹

Put that beside Articles 10 and 11. A fair hearing assumes a person who can present their account coherently under adversarial pressure, whose demeanour will be read accurately, and who will not agree with a proposition simply because agreeing ends the interaction. Every one of those assumptions is a social-cognition assumption, and every one of them is contested by the diagnosis.

Which brings me to the thing Baron-Cohen asked for.

In 2017 he closed his address with three calls. The first was for an investigation into the violation of human rights of autistic people. ²

Australia held one. The Royal Commission into Violence, Abuse, Neglect and Exploitation of People with Disability ran for four and a half years, took evidence from thousands of people, and delivered a final report in September 2023: twelve volumes, 222 recommendations. ⁶⁰ It is, as far as I know, the largest inquiry of its kind conducted anywhere.

In July 2024 the Commonwealth released its response. Of the 222 recommendations, 84 fell solely to the Commonwealth, 85 were joint with the states and territories, 50 were entirely for the states and territories, and 3 were shared with non-government organisations. The Commonwealth accepted or accepted in principle 130. A further 36 were left for later consideration, and the remainder noted. ⁶¹

I want to be fair about this. It is not nothing. The response carried real money - $227.6 million for a new specialist disability employment program, $39.7 million for disability advocacy, $6.9 million to review the Disability Discrimination Act, and $1.2 million to develop targets to reduce and eliminate restrictive practices. ⁶¹

But look at that last item next to the numbers above, and notice what is being conceded. In 2024 the Australian government committed funds to develop targets to reduce practices that its own regulator was recording against fifteen thousand people a quarter, with the count going up. Not to reduce them. To work out what the targets should be.

And note the wording that carries most of the response: accepted in principle. That is the phrase governments use for a recommendation they do not intend to reject and do not intend to be held to.

So the investigation Baron-Cohen called for happened, in my country, at enormous cost, with extraordinary testimony. That is genuinely more than most nations have done. What followed was a partial acceptance, in principle, of most of it.

His second call was for increased surveillance of the needs of autistic people, so that each year we could see the violations reduce. Australia does that too, quarterly. The restrictive practice count has gone up in almost every quarter it has been measured. Surveillance was never the missing ingredient.

Part 10:
The articles nobody has counted

Article 21 gives everyone the right to take part in the government of their country, and to periodic and genuine elections. Article 27 gives everyone the right freely to participate in the cultural life of the community. ¹

These are short sections, and they are short for a reason I want to make part of the argument.

Australia has compulsory voting, and the default way of discharging it is to attend a polling place on the day: a queue of unknown length, a crowded hall - usually a school - fluorescent light, noise, several scripted exchanges with strangers, and a ballot paper that in the Senate can require numbering dozens of boxes correctly for the vote to count.

I could stop there and call it an Article 21 problem, and I nearly did. It would be wrong, because that is not the only way to vote and has not been for a long time. Australians can vote early, in person, over a period of weeks, and they can vote by post; at recent federal elections a very large share of the electorate has done one or the other. The environment I have just described is avoidable.

So the interesting question is not whether the accommodation exists. It does. The question is how you get it, and the answer is the pattern all of this is about.

You apply. In advance. On stated grounds.

Here are the grounds. To register as a general postal voter you must be unable to attend a polling place because you are: enrolled more than 20 km from one; a patient in a hospital or nursing home and unable to travel; unable to travel due to being infirm at home; caring for a seriously ill or infirm person; serving a prison sentence of less than three years; a silent elector; prevented by religious beliefs; unable to sign your name due to physical incapacity; an overseas elector; serving abroad with the defence force or the federal police; an elector with disability and unable to travel to a polling place from the place you live; or caring for a person with disability. ^{62 63}

Read that list again with an autistic voter in mind, and notice the verb. Travel. Unable to travel. Infirm. Physical incapacity. Every disability-related ground on it is written in the vocabulary of getting there.

But an autistic adult is generally able to travel. Walking to the school hall is not the problem. The problem starts on arrival, and it is made of noise, light, crowding, queueing without a known end point and unrehearsed interaction with strangers - none of which appears anywhere in the criteria. The accommodation is real, and free, and its entry test was drafted by people picturing a wheelchair or a hospital bed.

So the autistic voter has three options. Attend, and pay the cost. Do not attend, and be fined. Or classify yourself under a ground written for somebody else and hope the description is close enough - which requires knowing the option exists, knowing the deadline, deciding that you count as an elector with disability, and being willing to put that on a form. Every one of those steps is the counter from Part 2.

This is also one of the few places in these pages where looking abroad produces something other than a matching set of failures.

The United Kingdom abolished the grounds test in 2001. Any registered elector in Great Britain can have a postal vote for the asking - no reason, no category to fit into, no need to decide whether you count as disabled. ⁶⁴ Canada is the same: any elector may vote by mail without giving an excuse. ⁶⁵ The United States depends entirely on where you live. Around twenty-eight states let any voter request a mail ballot and another eight run all-mail elections, while roughly fourteen still require an excuse - and where those excuse lists survive they read almost exactly like Australia’s, turning on being incapacitated or confined by illness or physical disability. ⁶⁶

So the barrier is not inherent to postal voting, or to compulsory voting, or to disability. It is a drafting decision, and two comparable countries have already unmade it. Where the grounds test is gone, the autistic voter’s problem shrinks to knowing the option exists and meeting the deadline. That is still the self-service problem from Part 2 - but it is a very much smaller version of it, and it costs nothing to remove.

I would not hold up any of the three as a model, and Northern Ireland, which kept its grounds test, is a reminder that this is not a settled matter even within one country. ⁶⁷ The point is narrower: on this one article, the thing standing between an autistic Australian and an accessible vote is a list of categories, other countries have simply deleted theirs, and the sky did not fall.

What nobody can tell you is how often each of those things happens, because it has not been measured. I am not aware of any Australian research on how autistic adults experience voting, or on what they do about it. The one study I could find anywhere is American: Tassone and colleagues surveyed 276 autistic and 361 non-autistic adults and found autistic adults reporting significant barriers to voting and distinctive experiences of political participation. ⁶⁸ It is a pilot, online, self-selected, and conducted in a country with voluntary voting and a different electoral system.

So the honest position is narrower than the one I began with, and it is still worth stating. Australia has built the accommodation, written its entry criteria in a language that does not describe us, and never asked the people concerned whether it works.

Article 27 is in a similar state, and I am wary here for a different reason. Autistic adults do participate in culture, intensely; Grove and colleagues found special interests associated with higher subjective wellbeing, and Botha, Dibb and Frost described autistic community connectedness as a genuine culture with its own belonging and politics, not a support group. ^{69 70} There is no deficit to report and I am not going to invent one.

What is worth saying is narrower. Cultural participation in the public sense costs money and social access - a ticket, a companion, transport, an environment you can tolerate - and Parts 4 and 5 removed most of those. Article 27 is not being blocked. It is being priced out, downstream.

Although at the moment it is also being priced out rather more directly. Under the reforms described in Part 5, from October 2026 NDIS budget allocations for social, civic and community participation supports are reduced by 50%, and capacity building daily activity allocations by 10%, with critical daily-living supports left untouched. ⁴⁸ The logic is clear enough - these are the supports least like personal care and most like discretionary spending. But the line the reform draws is a line between staying alive and having a life, and Article 27 is on the far side of it.

I have flagged both of these as thin deliberately. An argument of this kind is under pressure to claim a violation everywhere it looks, and the correct answer for two of these articles is nobody has measured it. That absence is a finding. Baron-Cohen’s third call was that we should continuously ask autistic people what their lives are like. ² Nine years on, on the article that governs how a citizen participates in their own government, in a country that compels it, we have not asked.

Part 11:
Article 3

Article 3 is nine words. Everyone has the right to life, liberty and security of person. ¹

Hirvikoski and colleagues, using Swedish national registers, found premature mortality markedly increased in autistic adults across a range of causes. ⁷¹ O’Nions and colleagues, in the UK matched cohort described in Part 1, estimated a reduction in life expectancy of around six years for autistic men and women without intellectual disability, and larger reductions where intellectual disability was present. ¹¹ The pattern holds across the national registries that have looked - Sweden, Finland, the United States, Australia, Canada - with mortality at roughly two to three times general population rates.

Suicide is a leading contributor, and Part 1 gave the version of that found in coroners’ files. Here is the version found at population scale.

Kõlves and colleagues followed the entire Danish population aged ten and over - 6,559,266 people - across twenty-two years of national registers, of whom 35,020 had an autism diagnosis. Adjusted for the usual confounders, the autistic group had 3.19 times the rate of suicide attempt and 3.75 times the rate of suicide. ⁷² Suicide attempts ran at 266.8 per 100,000 among autistic people against 63.4 among everyone else. Within the autistic group, women attempted at over four times the rate of men.

Say that in plain words, because the decimals hide it. Autistic people died by suicide at nearly four times the rate of everyone else. The confidence interval runs from under three to just under five, so it may be worse than that.

And look at who “everyone else” is. It is not a healthy control group. It is the entire rest of the country - every other person carrying every other risk factor there is, all averaged in. Four times sits on top of all of them.

That is not a clinic sample, a self-selected online survey, or a specialist service. It is a whole country, counted, over two decades. Brown and colleagues, reviewing the field through to 2024, found the direction consistent wherever anyone has looked. ⁷³

That is the sort of gap that gets a name and a taskforce when it turns up anywhere else in public health. Here it has been in the literature for a decade.

The mechanism work points at something specific. Cassidy and colleagues found camouflaging associated with suicidal thoughts and behaviours. ⁷⁴ Pelton and colleagues tested the interpersonal theory of suicide in autistic and non-autistic samples and found two constructs elevated in the autistic group: thwarted belonging, and perceived burdensomeness. ⁷⁵

Look at those two constructs and then look back at everything before this. Thwarted belonging is Part 8 and Part 10. Perceived burdensomeness is Part 4 and Part 7 - being unable to work at your level, and being required to describe yourself as a deficit in order to be helped.

The mechanisms named in the suicide literature are the same mechanisms this argument has been tracking through the other articles. They are not a separate topic that arrives at the end. They are what the earlier parts produce.

“But that is a choice”

There is an objection to this part, and it is better answered than avoided.

Article 3 reads like a protection against what others do to you. Nobody killed these people; they acted themselves. So, the objection goes, however sad it is, it falls outside the article. It is a choice, and rights are not engaged by choices.

Three answers.

The first is legal. The right to life has never been read as purely negative. States are understood to carry positive obligations to protect life, not merely to refrain from taking it. And where disability is concerned, the Convention makes explicit what is elsewhere implied: Article 10 requires States Parties not only to reaffirm the inherent right to life but to take all necessary measures to ensure its effective enjoyment by persons with disabilities on an equal basis with others. ⁷⁶ Effective enjoyment is not a promise to leave you alone. It is a duty to do something, and it is owed to this group by name.

The second is empirical. If the elevated rate were a property of autism itself, the predictors would be internal. They are not. The constructs that come up are thwarted belonging and perceived burdensomeness - both descriptions of a person’s position relative to other people. ⁷⁵ Camouflaging, which is a response to how others react. ⁷⁴ Minority stress, which Botha and Frost define as operating through social position rather than through anything intrinsic. ²⁸ Unmet support needs. Undiagnosed status. Every one of those is a variable somebody else controls, and not one of them is a symptom.

The third is about the word.

Granted. Call it a choice. The question a rights framework then asks is what the alternatives were, and who removed them.

Everything above has been about the removal of alternatives: the assessment you cannot afford, the job you cannot get, the adjustment you cannot ask for, the complaint you cannot make, the room you cannot be in, the support you can obtain only by describing yourself as worthless. Taken one at a time, each is an obstacle. Taken together, what has been built is a life with the exits closed one by one, mostly by people and institutions who were not being cruel and were following their own rules.

A choice made after the alternatives have been removed is still a choice. It is also, in every other context in which we use the word, what we mean by coercion.

And we know we do not accept the objection generally. When a bullied child dies we do not say the matter falls outside anybody’s duty because the child acted; we look at the bullying. When a patient dies on a psychiatric ward we do not close the file on the grounds that the death was self-inflicted - the whole apparatus of coronial inquiry exists precisely because self-infliction does not end the question of what everyone else did. The reasoning is applied selectively, and it gets applied to us.

One thing I want to be careful about, because getting it wrong here would be worse than not writing the section.

None of the above says these deaths were rational, or inevitable, or that the reasoning that led to them was sound. It says the reverse. The variables are external, which means they are modifiable, and that is the whole argument. A risk that tracks diagnosis, employment, support and belonging is a risk that moves when diagnosis, employment, support and belonging move. Were any of it intrinsic to autism there would be nothing to allocate and no duty to argue about - you cannot owe somebody a different nervous system. Because it is not, there is.

Argue with the rest of it if you like. Autistic adults are still dying about six years younger than everybody else. ¹¹

Part 12:
They do not fail one at a time

Everything so far has been presented as separate articles, because that is how the Declaration is written and how Baron-Cohen’s address was organised. It is the wrong shape.

Follow one person through.

They cannot get assessed, because assessment is private and costs three thousand dollars they do not have - Part 1. Without the assessment there is no adjustment anywhere, and the first place that bites is the interview, where the judgement is made in seconds and there is no document to account for it - Part 2. From there the chain forks, and both forks arrive at the same place. Either the work never comes at all - which is where most working-age autistic people already are - or it comes and then costs more to hold than it costs anybody else, because holding it means performing all day with no adjustment and no explanation. So they reduce their hours, or they are managed out, or they leave before they are - Part 4. Either way the income is lower than the qualifications say it should be, and for a great many there is no income from work at all - Article 23. With little or none of it, housing gets marginal - Article 25. Marginal housing means no private space, so Article 12 goes without anyone interfering with anything. No money means no going out, so Article 27 goes. No going out means the social circle thins, and thinning it removes the last people who might have helped them navigate any of the above.

Now they apply for support. To get it they need the assessment they could not afford, and then they must present themselves, at length and in writing, as an impaired person - Part 7, and Article 1 with it. If they cannot do that well they do not qualify - Part 2.

By this point thwarted belonging and perceived burdensomeness are not psychological abstractions. They are accurate descriptions of where this person now stands, arrived at through eight administrative steps, and they are the two constructs that predict Article 3.

That is why counting violations misses it. These are not ten failures. It is one failure, ten times, each one lowering the ceiling for the next.

Nobody does the illegal thing

Before going further I want to correct an impression these pages may have left, because getting it wrong is how the whole subject gets misread.

Nobody is dismissed for being autistic. That is unlawful, everybody knows it is unlawful, and it would be provable if it happened. It does not happen. What happens instead is assembled entirely out of lawful parts.

Consider how a person is actually removed from a workplace.

First a capability finding. Not a disability finding - a performance one. Difficulty working with others. Communication concerns. A problem with team fit. Every one of those is observable, documentable and defensible, and every one of them is a diagnostic criterion rewritten in the vocabulary of performance management.

Then a reassignment, sideways, into a role with very little attached to it. Not a demotion; a better fit, given the concerns. It would be described as an accommodation if anybody asked, and by the letter of it, it is one.

Then, when the person does not resign - which was the function of the previous step - the role is made redundant. Redundancy is lawful. The position genuinely no longer exists. It was brought into being in order to not exist.

Three moves. None illegal. None of them about disability. And at the end the person is out of work with nothing to complain of, because what happened to them has been taken apart into components and not one component is a violation of anything.

Or take a benefit application. The agency assessing it needs to ask some follow-up questions and telephones to ask them. The number shows as unknown. The applicant does not answer calls from unknown numbers - and the reason they do not is, precisely, the thing they are applying about. Doherty and colleagues found difficulty making appointments by telephone reported by 62% of autistic respondents, the second-highest barrier in their survey. ²¹

There is no way to return a call from a number that was not disclosed. The claim is refused for failure to respond.

Read that file afterwards and there is no disability decision anywhere in it. There is a non-response. The agency followed its process; the process required an unscheduled telephone conversation with a stranger from an unidentified number; and the impairment being assessed is one that makes that exact act hard. None of this is recorded, because from inside the process nothing happened at all except that somebody did not pick up.

I did not invent either of those. Both happened to me.

This is what makes the pattern so difficult to name, and so easy to dismiss. Each individual act is lawful, ordinary and defensible. The discrimination is in the sequence - and no forum in this country reviews sequences. They review decisions, one at a time, each on its own terms, and each one holds up.

Now notice what the record shows at the end of it, because the answer is almost nothing, and it gets to almost nothing by two different routes.

If they never claim, nothing is refused, so nothing is written down. Every institution in the chain applied its own rules correctly to what was in front of it, and would say so, and would be right.

And if they do claim - if they do the hard thing, and lodge the complaint, and sit in the hearing - then there is a meeting, and an officer, and a letter, and somebody does say no. But look at what is being refused. Not the right. The standing to assert it. The finding is that the attribute was not established, which is not a ruling that the discrimination did not happen; it is a ruling that the question will not be reached. ^{3 4}

Two exits, opposite in every respect, arriving at the same silence. One produces no record because nobody was asked. The other produces a record of the wrong question being answered. Ten articles go missing, and in neither case does any document anywhere say that a right was denied to an autistic adult.

That is what I mean when I say there is nobody you could name. Not that everyone behaved well - some of it is refusal, and refusal with a signature on it - but that the refusals are filed under eligibility, and eligibility is not a rights decision, and so nothing accumulates.

That is also why the compounding runs the other way, and it is the part that should give governments pause. Every article in the chain has a policy owner, and each can truthfully say the problem originated somewhere else. Health says it is an employment issue. Employment says it is a health issue. Housing says it is an income issue. Nobody is lying, and nothing moves.

Part 13:
But…

Everything above is arguable. Three parts of it are weaker than the rest, and if the argument gets taken apart, it will be taken apart there.

The first is that this argument is unfalsifiable if I let it be.

There are thirty articles. I have used seventeen. I could have kept going - Article 13 and freedom of movement, Article 17 and property, Article 26 and lifelong learning - and each one would have yielded a paragraph, because a sufficiently determined writer can find a rights violation anywhere.

That is exactly the failure mode to avoid. A claim that every article is violated is not a strong claim; it is an unfalsifiable one, and unfalsifiable claims are ignored by the people who allocate money, correctly. So I want to be explicit about the ranking. Articles 23 and 25 are supported by national statistics and replicated peer-reviewed cohorts, and I would defend them anywhere. Articles 3 and 5 are strongly supported. Articles 6, 12, 16 and 22 are argued from smaller studies, or from statute plus reasoning about how the systems work, and the reasoning is doing more of the load than I would like. Articles 21 and 27 are, on current evidence, unmeasured, and I said so rather than filling the gap.

The second is that much of this is not about autism.

Take the cascade in Part 12 and ask which links are autism-specific. Unemployment leading to housing insecurity leading to social isolation is not an autism mechanism. It is what poverty does to anyone. Bureaucratic systems that demand humiliating disclosure are not aimed at us; that is how means-tested support works for every claimant. Restrictive practices affect disabled people across every diagnostic category, most of them not autistic.

An honest version separates three things - a different cut from the one in Part 3, across the same material. Some of this is general - it would happen to any poor person. Some is disability-general - it happens to disabled people as a class. And some is autism-specific.

The autism-specific part is narrower than the length of all this implies, and it is Part 2. It is that the access mechanism for every one of these rights is a social interaction, and social interaction is the impairment by definition. A wheelchair user faces a building problem, and a building problem has a building solution that can be legislated, funded and inspected. Our problem is that the door is a conversation. That is the claim that is actually ours, and everything else here is what it does when it meets systems that were going to be hard anyway.

The third is legal, and it is the most serious.

The Universal Declaration is not law. It is a declaration - a statement of aspiration adopted by the General Assembly in 1948, with no enforcement, no court and no remedy attached. ¹ Nobody can be taken anywhere for breaching it.

The instrument that does bind is the Convention on the Rights of Persons with Disabilities, adopted in 2006. ⁷⁶ It has an enforcement architecture, a monitoring committee, periodic review, and - through its Optional Protocol - a complaints mechanism. It also contains, in Article 12, an explicit statement that disabled people have legal capacity on an equal basis with others, and in Article 19 a right to live independently and be included in the community, neither of which the 1948 document says in those terms.

Australia ratified it in 2008, the United Kingdom in 2009, Canada in 2010. The United States signed it in 2009 and has never ratified it: the Senate vote in December 2012 came five short of the two-thirds required, 61 to 38. ⁷⁷

So a reader could fairly say I picked the weaker document. I did, knowingly. The 1948 text is the one people have heard of and the one whose sentences still land - and the Convention sharpens the point rather than blunting it. Three of the four countries here have been legally bound by a disability rights treaty for more than fifteen years. Everything in Parts 3 through 10 was measured after that, and none of it shows up as fixed.

The fourth country is a different problem, and it cuts the other way. American autistic adults appear throughout the research this argument leans on, and the United States is not party to the instrument that would give any of it legal traction. For them the Universal Declaration is not the weaker document I chose in preference to a stronger one. It is most of what there is.

The Declaration’s problem is that it has no teeth. The Convention’s problem is that it has teeth and the numbers went the wrong way regardless. Neither of those is an argument that the rights are wrong.

One more, which is mine rather than an objection I expect from others. Almost every study cited here recruited diagnosed autistic adults - people who got through the gate described in Part 1. Whatever the true figures are for autistic adults as a population, the ones I have quoted are drawn from the group best placed to have claimed something, and are therefore likely to be the optimistic version.

In closing:
Everyone

My diagnosis was bought. Around three thousand dollars, and eight months of waiting, in my late fifties, after a thirty-year career I had built without knowing what I was.

I want to be exact about that, because the previous part accused everyone else’s evidence of survivorship and it would be poor form to exempt myself. I am in these pages as one of the people who got through the gate, and I got through it the way the others did - not by qualifying for anything, not because a system decided I was owed an assessment, but because when the money was needed it turned out to be findable. That is luck. It is the same luck that put the participants into every study cited above, and it is invisible from the inside, which is exactly what makes it worth naming.

Until that report existed, nothing written here applied to me in any way a system would recognise. The report is the reason I can write this at all. It is not a reason I earned.

That is what I keep coming back to. Not the unemployment rate, though it is the strongest number here. Not the six years, though that is the one that should stop a room. What stays with me is the ordering: that the first thing you need is a document, and that the document is sold rather than given.

The drafters used the word everyone over and over. It was not carelessness; they had just watched what happens when a state decides some people are a different kind of thing, and they wrote the most absolute word available, thirty times, so it could not be read any other way.

The word held. Nobody in Australia, Britain, Canada or the United States argues that autistic adults are outside it. That fight was won before I was born, and it stayed won.

What was never built was the delivery.

Every right in that document reaches a person through a mechanism, and every mechanism turns out to be a conversation with someone who has the authority to decide about you: an interviewer, a doctor, an assessor, a caseworker, a magistrate. The Declaration guarantees what you are owed. It says nothing at all about how you are supposed to ask, because the people who wrote it could not imagine that asking was the hard part.

For us, asking is the whole of it.

For most of this argument I have been careful not to say that autistic adults are being denied their human rights. Denied implies a denier, and Part 12 is right that there is not one - no meeting, no officer, no moment at which somebody said no.

I have stopped believing that distinction earns what it is being asked to carry.

Put the outcome on one line. Two in five of us in work. Four times the rate of suicide. Six years off the end of our lives. Fifteen thousand people a quarter under authorised restraint, and rising. A diagnosis you can only buy, in a country where almost nothing attaches to you without one. Had a government set out to produce that deliberately, this is what it would look like, and it would look like nothing else.

The difference between a system that refuses you something and one that reliably fails to give it to you is real, and it matters enormously to the people running the system. It makes no difference whatever to the person at the end of it. And it stops working as an excuse at the point where the effect has been counted, published, put to a royal commission and handed back - because continuing unchanged after that is a decision, even if nobody ever signs it.

So, yes. It looks like denial because it works like denial. The only thing it is missing is somebody prepared to put their name to it, and I no longer think that is the difference it is claimed to be.

And there is one more thing about the 1948 document, which it has taken me all of the above to see properly.

Read it again and notice which way it faces. It is not addressed to us. It is addressed to states. Every article is a duty laid on the powerful concerning their treatment of the powerless - what governments shall not do, what they shall secure, what they shall provide. The people in it are not applicants. They are the beneficiaries of an obligation somebody else carries. In thirty articles, nobody is asked to come forward.

That was the design, not an oversight in it. The drafters had just watched what states do to people who cannot make them stop, and they built an instrument that runs downhill - from government to population - precisely because the people it was written for had no way of reaching up.

What I have described is that instrument running backwards.

The duty-bearer waits. The rights-holder initiates. You must notice that you are entitled to something, establish that you are the kind of person it applies to, gather evidence, submit it, explain yourself to whoever decides, and go again when the answer is no. The obligation has been quietly converted into an application, and the weight of it moved from the party with the power to the party without.

And that reversal does not land evenly. A right you have to claim goes to whoever is best at claiming. Human rights exist for people who cannot compel anything from anybody - that is the entire reason for writing them down - so a self-service version delivers them first, and most reliably, to the people who would have managed without them.

The rights are there. Universal, unconditional, written down and agreed.

Getting them is self-service. You have to go and ask - in the right words, of the right person, on a day of their choosing - and asking is the thing this disability is defined by.

And if you do ask, if you find the words and the person and the day, you will be asked first to establish that you are the sort of person entitled to be asking. Which takes a document. Which has to be bought.

So they are not on a shelf, and I should stop saying they are. There is somebody standing in front of them, and they will not ask you what you need. They will ask you what you can prove.

Everyone, the drafters wrote. Thirty times, in the most absolute word they had.

Everyone who can prove it.

⁂

---