NDIS RFT Answered

Feb 21

On 5 February 2024, the NDIA issued an RFT for the provision of Needs Assessment Tool(s) to support NDIS Reform Initiatives.

Of course, this RFT happens just a few weeks after the NACC announced that the conduct of officials at the NDIA has been referred to law enforcement agencies and regulators after a series of probes revealed secret meetings with ministers and gifts from suppliers vying for government tech contracts.

So I guess we can all be comforted by the statement in this RFT: 'The Agency’s policy is to engage in the highest standards of ethical behaviour and fair dealing throughout the RFT process. The Agency requires the same standards from those with whom it deals.' Cool. But let's be clear. Ethics is more than a statement on an RFT, and it is about way more than procurement, which will become evident in the following discussion.

Anyway, it’s no surprise that this RFT is generating a lot of interest, judging by the many private messages I am receiving, and the LinkedIn stats showing just about every vendor checking out my profile.

Would I like to help, they ask? Heck, yeh. But not in the way that vendors - or the NDIA - might expect.

This is my way of helping: putting together my 'response 'to the NDIS RFT...

...and publishing it so that everyone in the market and various oversight bodies such as the JSCNDIS, Senate Estimates, ANAO and NACC can contemplate what this RFT is actually about, and the risks. Curiously, the ANAO audit report of the performance of the NDIA Board - which was due this month - has been pushed back a whole two months to April, after Estimates and probably after the Election. Sweet.

My RFT 'response' is drawn from my various submissions, and covered by Parliamentary Privilege.

Essentially, the NDIA wants to build a novel algorithm, that doesn’t currently exist anywhere in the world, and integrate it with its systems, that are known to be defective, in an environment where data migration from SAP to PACE has not occurred, with zero capability and experience in ever having done this before.

This will be a human and financial disaster. I hope everyone concerned is factoring Malfeasance and Class Action litigation into their considerations.

In no particular order, here are my responses and associated commentary.

WHAT IS REQUIRED

Assessment tool(s)

A needs assessment tool(s) for assessing the support needs of adult (16+ years) NDIS Participants. This component may consist of a single assessment tool or a combination of tools.

Systems Integration + NDIA Build

The RFT also calls for the provision of, or support for, a system integration solution. Tenderers must meet this component of the Requirement by providing a response to Option 1 – NDIA Build, but may also include in its Tender one or more of the other options set out below in its Tender.

That’s pretty scary, an NDIA build + systems integration.

MY COMMENTARY

The RFT is vague and the most asinine read. Let’s look at the un-weighted criteria. When read together, risk screams out from this RFT.

Un-Weighted: Costs

With pricing unweighted, how on earth will VFM be determined. But we have seen this VFM circus before. Remember, the supposed 'price certainty' in the PACE procurement as the definition of VFM. And of course, let’s not forget the 'unintended' disclosure of pricing information to a competitor bidder.

It seems that in dealing with Participants, pricing/costs are a big factor for the NDIS decision makers but not so for its own procurements in the hundreds of millions of dollars.

More on my estimate of costs below.

Un-Weighted: Benefit to the Australian Economy

For a program that occupies such a large chunk out of National revenues, it is extraordinary that 'Benefit to the Australian Economy' is unweighted. Responses from tenders on their perceived value to the Australian economy would be insightful indeed - hopefully this insight sees light of day via FOI or other oversight mechanisms, such as Senate Estimates.

Also Un-Weighted. Compliance and Risk, and Financial Viability and Corporate Capacity

'The extent to which the Tenderer has the financial and corporate capacity to fulfil the resultant Contract obligations'. Why is this un-weighted, particularly for a tool upon which the whole future of the NDIS sits? This is extremely concerning.

The Tool. Does such a tool exist anywhere?

No. This is novel. Such a tool doesn’t exist anywhere in the world.

'The outputs of the tool(s) support appropriate budgeting decisions in respect to NDIS Supports'.

This will effectively be the Budget Calculation Instrument - which will be the apex of NDIS algorithms - a machine that predicts support needs based on defective data (more on defective data shortly). Are we to persist with the folly of predictive algorithms, sounds familiar, RoboDebt?

As stated in my submission to the NDIS Bill Inquiry:

Let’s consider the Budget Calculation Instrument. It is argued that the 'Plan Flexibility' derives from not having to construct a plan budget from individual line items, because the Budget Calculation Instrument will do this. Because this instrument is defined by a legislative mechanism, this calculation will not be a reviewable decision. Does this look like Plan Flexibility?

The Budget Calculation Instrument is a new concept and the tools to give it effect do not exist anywhere in the world. This is exceptionally hazardous policy leading to defective legislation. Expecting an agency in crisis, without the organisational capability to implement, to take on something so dangerous that it is held un-reviewable and undefined in legislation - is seriously hazardous.

With such widespread sector concern over the defectiveness of the 'new' PACE system (replacing the defective SAP CRM system) and disruption to payments - the government would have the NDIA create, without the capability to do so, a yet to be defined Budget Calculation Instrument, that does not exist anywhere else in the world.

But such algorithmic ‘calculation’ concepts appear to entrance the Australian government.

Will this Budget Calculation Instrument be a recast of the Services Australia $200 million Entitlement Calculation Engine (ECE) that was dumped by this government, by Bill Shorten as Government Services Minister? Or a recast of the dumped Home Affairs $250 million 'Permissions Capability Platform', which was the subject of inquiry by the Joint Committee of Public Accounts and Audit?

And why would the NDIA expect a different outcome, when Services Australia with all its resources failed catastrophically.

Or why would the NDIA expect a different outcome to that of the Department of Home Affairs ‘Permissions Capability Platform’, which was dumped undelivered after years of work, at a cost of $250 million.

All three are essentially the same concept: a theoretical automated calculation involving complex variables to determine such things as entitlements, eligibility and budget. And two of these ‘calculation platforms’ have consumed almost half a billion dollars in public funds, only to be terminated undelivered after years of mismanagement.

Here’s my evidence on this to the Senate Inquiry into the NDIS Bill.

Systems Integration + NDIA Build

It is extraordinarily reckless that this RFT even contemplates an NDIA build, given that the scathing JSCNDIS report just a year ago regarding the NDIA’s defective capability and culture. But here we are, with the very same Agency positioning itself to build software based on a paper-based tool.

The RFT reads:

‘Provide the necessary rights for the Agency to build the software based on the paper-based assessment tool(s) and host it in the NDIA’s ICT environment. For example, if a tool is available as a text-based document (e.g. Word document or PDF), the Tenderer must allow the Agency to develop software to allow assessors to administer the tool using NDIA devices. The Tenderer should provide details of what assistance can be provided for this option.’

These two requirements - systems integration + NDIA build - assume a level of technology design and build capability maturity. My submission on the NDIS Bill Inquiry painted a detailed picture of an organisation with a catastrophic capability deficit, defective porous systems, defective data, and resulting harm.

To reiterate. The risk of ignoring the lack of capability is extraordinary.

With such widespread sector concern over the defectiveness of the 'new' PACE system (replacing the defective SAP CRM system) with the sector yet again experiencing widespread disruption to payments.
With the JSCNDIS finding that the NDIA had operated unlawfully for a decade by its invention of the administrative convenience of ‘primary disability’.
The government would have the NDIA create, without the capability to do so, a yet to be defined Budget Calculation Instrument, that does not exist anywhere else in the world…
...and with no apparent regard for the half a billion dollars that has already been wasted by Services Australia and Department of Home Affairs on abandoned attempts at the same concept.

And yet, this RFT explicitly requires an NDIA build. Let’s see what that look like.

To give a sense of what the defective NDIS systems look like, an excursion through the systems outages over the past couple of years paints a troubling picture.

NDIS Systems Outages

There is something seriously broken with NDIS systems. And the symptoms of serious malfunction are hiding in plain sight. Note, all this data and commentary has been provided to the ANAO for the audit on the performance of the NDIA Board. It is disturbing that this ANAO audit report which was due to be released this month (February), has been pushed back to April - conveniently after Estimates and probably the Election.

These PLANNED outages over the past year or so DO NOT INCLUDE payment disruptions (and there are many), other systems unplanned disruptions and unavailability, other service unavailability, data breaches, and just plain lack of accessibility.

The duration of the outages is extraordinary, sometimes going for days.

The reason why this commentary is relevant to this RFT, is that the outages will likely worsen significantly as the NDIA attempts to undertake an in-house build and systems integration into these defective unstable systems.

Defective Data

The unlawful fiction of ‘primary disability’ (see commentary below on Legal Warnings) is hard coded into the NDIS systems, together with a separate categorisation of ‘secondary disability’. Too bad that the concept of 'primary disability' is unlawful - it's hard coded - and the whole system has been built on top of it.

And exactly how is systems integration supposed to be achieved with such wildly defective and missing data. It is known that data migration was not done in the move from SAP to the PACE system, as explained in detail in my NDIS Bill submission. My prediction of the impact of this extraordinary decision was this - my crystal ball 🔮 track record is totally bankable: 👇 👇👇

'Data defects and inaccurate Participant records cannot simply be fixed by moving to PACE, especially when this change looks like data migration and quality control is not happening...I predict that participants will be caught in data migration hell. Get ready for a massive loss of data during this period.'

Here’s what data loss actually looks like. With dates 9999. Unbelievable. There are reports of Participants suddenly disappearing from the system.

Participant disabilities disappearing - disabilities that have already been accepted - suddenly vanishing from the system. The person magically cured by these defective systems. And then having to go through lengthly and costly hunger game review processes to re-instate the funding that was algorithmically and unlawfully stopped. The cessation of funding causes harm.

Why should this matter to tenderers?

Well, if data migration has not occurred as the evidence indicates, then there’s gonna be one hell of a data remediation project. Tenderers might need to build their bids on this assumption, potentially excluding data remediation from the scope of the RFT and bid until there is transparency over the status of data migration and data integrity.

This brings us back to the mystery of price as an unweighted criteria. The real question is, what is the total cost. We know that the NDIA never really answered questions about the fully loaded cost and ‘business case’ for PACE. The Agency's responses to Questions on Notice about this were inconsistent: see my submission for the tear down of the PACE business case.

But wait, there’s more…the business model of change requests.

Change Requests

It should be expected that tenderers will factor a massive premium for risk plus an assumption of future revenues flowing from change requests.

There is a lot of unknowns with this RFT that directly translate to risk. Scope will be near impossible to lock down, particularly given the short response timeframes and lack of architectural documentation (so far).

With the requirement that the assessment algorithm will ‘integrate with NDIA ICT systems’, change requests will likely also arise from those systems it will be integrating with (ie PACE).

The Parliament and oversight bodies should be very concerned with how VFM is determined given that an explosion of costs is about to happen across the NDIS technology ecosystem - including providers.

So that brings us to my estimate of costs.

My Estimate of Costs

The value of the tender has not been included in the tender documents.

But in December 2024, the Government announced $280 million in funding for the NDIA to develop a more consistent approach to understanding participants’ needs for NDIS supports.

Considering the contract period is for 5 years, an initial contract value is likely in the order of $300million-$400million, plus ongoing support and change requests. There is a question as to whether the contract value is uncapped, as has been the case with PACE.

Bottom line, this thing will likely consume half a billion dollars all up, before being dumped as happened with similar adventures at Services Australia and Home Affairs.

That is half a billion not going to Participants. I hope this contracting process is closely scrutinised.

Deaths and Suicide

The algorithms and manual processes - aka RoboNDIS - that are currently being used by the NDIA to implement the ‘NDIS Reforms’ and kick people off the scheme are causing significant harm and killing people in record numbers. Why does this matter to this RFT? Well, because this novel new algorithm which is the subject of this RFT, is intended to accelerate the automation of the removal of people from the scheme and the reduction in plan budgets through prediction.

As many others have done, I myself have written submissions about the risk of suicide that this poses for people with psychosocial disability, the very people being targeted by these 'reforms'. We were not listened to by the govt, instead we have called extremists and keyboard warriors.

But the numbers do not lie.

The terrible outcomes of these ‘reforms’ are already being swiftly realised, with questions now being asked about the increased rate of suicides. This will worsen with the automation sought via this RFT.

Horrifically, FOI documents reveal that changes to the NDIS are already killing record numbers of NDIS Participants. This can be seen by comparing THEN & NOW.

Deaths THEN (2018): About 100-150 people died each month.

Death NOW: About 400-500 ppl per month.

Latest quarter saw 1,326 deaths. Death rate has increased significantly. Working age adults (45-64) are dying in higher numbers.

And there is a gender pattern: more women are dying than men. Latest quarter: 760 women died compared to 545 men. More people are dying than ever before. More people are being removed than ever before. Children are being disproportionately affected.

It beggars belief, that in the face of these numbers on deaths and suicide, and the complete objection by health professionals, legal and disability advocates to this algorithmic assessment process through the NDIS Bill inquiry, now defective law - that this RFT hints that harm might actually result. The RFT states that this tool must be ‘Able to be administered in a trauma-informed way’. How can this thing be ‘trauma informed’ and yet there is no mention of co-design in the RFT and the broader disability community not involved in co-designing the new Support Needs Assessments.

The arrogance is reckless.

Timelines

Then to the matter of timelines, rushed out before Caretaker and the looming Election.

And this for something that does not exist.

Let’s take a close look.

In January, the NACC announced that the conduct of officials at the NDIA has been referred to law enforcement agencies and regulators after a series of probes revealed secret meetings with ministers and gifts from suppliers vying for government tech contracts.

A few weeks later, the RFT is released.

We see that tenderers questions are shut off in March, just after a month. This is important, because it does not appear that any architectural documents - even high level - have been provided. So it could be expected that tenderers might have lots of questions. Or at least they should.

The closing time for tenders is 21 March.

The evaluation is complete a bit more than 3 weeks later. All I can say is, wow. This is extraordinarily swift for something so complex; for a tool that does not exist; and for an Agency whose procurement practices are currently being examined by law enforcement.

There is a month between the preferred tenderer being notified, and the expected start date. This is the period where contract negotiations happen - imagine sorting through scope, assumptions and various contractual and commercial details in such a compressed timeframe. No pressure.

April-May will likely see an election. Will the preferred tenderer keep their A team on hold, while the dust settles from the Election? And at what cost?

The timeline assumes work starting in April 2025, and a mere 5 or 6 moths later (ie by ‘late’ 2025), the contractor will ‘provide all material to allow the NDIA to start administering assessments for NDIS participants (e.g. the updated tool(s), training and accreditation approaches and a system integration solution as relevant)’. That is, software build and systems integration completed in this short period of 6 months or so. Expect that there will be no time for co-design in this compressed timeframe.

My view on the timelines, is that it is naively and dangerously ambitious, but expected for an organisation with such serious and widely documented capability and culture deficits.

Legal Warnings

It is relevant background to this RFT, that over the years, there have been judicial, Parliamentary, administrative and academic scholar warnings and findings regarding the legality of NDIS operations. Politicians, bureaucrats and other interested parties cannot say they were not aware of the potential and actual harm that is happening. Remaining silent and getting on with the job is not an option.

From the RoboDebt Royal Commission

The RoboDebt Royal Commission hearings examined legal advice regarding misfeasance in public office, that if established, could be a basis for a class action. Some of the elements of misfeasance in public office covered in that legal advice, include:

The defendant must be the holder of a public office.
The defendant must have been recklessly indifferent to whether the act was beyond power and recklessly indifferent to the likelihood of harm being caused to the plaintiff. And
The defendant must have acted with reckless indifference to whether the act was beyond power and there must have been, objectively, a foreseeable risk of harm to the plaintiff.

The foreseeable risk of harm is one of the foundations of the RoboNDIS Class Action Royal Commission campaign action.

From the JSCNDIS

In November 2023, the JSCNDIS found that the NDIA has been operating unlawfully for a decade by constructing the actuarial fiction of ‘primary disability’.

‘…the distinction between primary and secondary disability has no basis in governing legislation…which operates as a form of discrimination.’

And yet, as we saw in the discussion above on defective data, this distinction is hard coded into NDIS systems. With this unlawfulness persisting, questions need to be asked about the consequential downstream lawfulness of the assessment algorithm tool, which is expected to integrate with this unlawful construct of a system.

From the Australian Public Service Commissioner

In September 2024, the Australian Public Service Commissioner released the report into the RoboDebt Code of Conduct Inquiry. The Commissioner warned:

‘Public servants have a duty to consider whether a decision is ethically sound. The question cannot be confined to whether a decision is legally and technically possible but also whether it is, in fact, the right thing to do, no matter how hard that may be. It is not open to a public servant to manufacture contrivances, to selectively choose evidence to justify a line of action, or to simply turn away.’

In the face of suicides and deaths, that's a pretty clear directive from the Public Service Commissioner.

From Esteemed Scholars: Professor Carney and Dr Georgia van Toorn

The article, “Decoding the algorithmic operations of Australia’s National Disability Insurance Scheme” written by Professor Carney with fellow scholar Dr Georgia van Toorn, raises potential legal ramifications of the connections being observed between RoboDebt and the NDIA’s use of algorithms.

Professor Terry Carney, whose opinions and articles held weight at the RoboDebt Royal Commission, has also raised legal and ethical questions about the use of algorithms by the NDIA.

The commentary in the Carney/van Toorn article, presents a clarion warning as to the shifting legal minefield in the era of algorithmic decision making, which cannot be ignored. The Carney/van Toorn article refers to these as subterranean systems, ‘because their workings are neither publicly known nor amenable to legal rectification…’

CONCLUSION

This RFT is reckless arrogance in the extreme, that assumes a level of bureaucratic expertise and experience that does not exist.

That there was no co-design with the very people who know about support needs assessments, with hundreds of disability organisations issuing a joint statement of concern over the safety and well-being of people with disability.

https://everyaustraliancounts.com.au/campaign-statement-take-the-time-for-codesign-protect-ndis-participants-safety-and-wellbeing/

That reckless arrogance extends to the agency’s technology competence / incompetence in the face of legal warnings over many years and current law enforcement investigations.

There are some things as a society that we should not do, or agree to, or stay silent about.

Examples over history include landmines and chemical weapons. And the story of IBM and the Holocaust is the most evil episode of the power of technology and automation over populations. How many people and organisations stayed silent?

Added to this list of things to be forbidden must be the use of algorithms by government to determine access to services.

That is, there must be a ban on the algorithmic automation of the sorting and assessment of vulnerable populations and the algorithmic prediction of supports.

If you have read to this point in my ‘RFT response’, thank you. But now, you can’t stay silent and simply treat this RFT as any other procedural procurement. Action is needed.

I urge people and organisations assembling tender responses to reconsider, and abstain from responding. Demonstrate your moral objection to what the government is doing, to the harm that is already happening, and that will now scale in the most horrific way as a direct result of the work that you will be doing or helping to do, and profiting from.