This series was originally posted to Storify and has been reproduced here after Storify was shut down. Unfortunately a lot of the inline images I originally tweeted have been lost, but you can still get the gist of which paragraphs of the report I’m talking about.
In this edition of Too Long; Justin Read, Justin reads the Coalition Senators’ dissenting report.
Prepare For Grumpiness
It’s fair to say I was expecting to be more annoyed by the dissenting report than I ended up being. There just wasn’t a lot of there there.
Right, let’s look at this Dissenting Report then, shall we? #tljr
“Coalition Senators recognise that ensuring the integrity of the welfare system is a key focus for the Australian Government.” #tljr
… why? #tljr
Why is this a key focus, but other stuff isn’t? #tljr
The Ombudsman Report
This other report occupied way more of this dissenting report than it should have. It’s like the Senators ignored the whole inquiry process. That’s pretty insulting to all the people who bothered to show up or made submissions.
Anyhoo. The Senators bring up the Commonwealth Ombudsman’s April 2017 report. It’s here: http://www.ombudsman.gov.au/publications/investigation-reports … #tljr
The Senators seem to have skipped over why the Ombudsman wrote the report. I’ll help them out: #tljr
Since the initial rollout of the OCI, the Commonwealth Ombudsman’s office has received many complaints from people who have incurred debts under the OCI. This report examines our concerns with the implementation of the OCI, using complaints we investigated as case study examples.
The Senators note that the Ombudsman’s report noted positive changes. #tljr
Apparently getting something really wrong because you couldn’t be arsed doing the job properly is totally fine if you tweak it later. #tljr
The Senators are quoting very selectively from the Ombudsman’s report. #tljr
The quotes from the Ombudsman’s report are very selective. It’s super-obvious, too.
The report concludes that the ‘February 2017 changes which include improvements to the help functions, explanations and overall usability of the OCI go some way to addressing our concerns about usability of the system.’
That’s not praise, Senators. That’s “fixed some stuff after they were yelled at a lot by multiple people for weeks.” #tljr
There’s a lot to unpack in this bit, because it’s conflating several issues. #tljr
Before I do, though, I’ve scanned ahead and the Coalition Senators appear to be fixated on the Ombudsman’s report. #tljr
See? I told you they had a weird fascination with the Ombudsman’s report.
This is a dissenting report on the inquiry, remember? It seems the Coalition Senators have gotten confused. #tljr
They also seem to think the Ombudsman’s report was praise. It wasn’t. It was pretty damning of DHS and the OCI. #tljr
It wasn’t just me who thought so: http://www.smh.com.au/federal-politics/political-news/not-reasonable-or-fair-ombudsman-slams-centrelinks-robodebt-20170409-gvhf5u.html …
I’m not sure the Senator’s reading comprehension skills are all that flash. #tljr
Anyhow, back to the various bits in this statement. #tljr
“The Ombudsman was ‘satisfied the data matching process itself is unchanged.” Right, because that’s the bit ATO does. #tljr
Although the change to not use TFNs (and get around the Data-Matching Act) appears to have escaped notice. #tljr
Then we have the stupid “error rate” distraction again. But this bit I really want to hone in on: #tljr
This notion of the 20% discrepancy rate not being a problem because it was like this before. That’s really flawed. #tljr
The Maths Of Scaling
Loads of people seem to have trouble understanding the difference between a rate and an absolute value.
I’ll try to explain it.
As I point out towards the end here this ignores the increase in volume.
Remember that in the manual system, it was dealing with 20,000 checks a year. Robodebt automated things, and did 20,000 a week. #tljr
That’s a 52 times increase. If the ‘discrepancy rate’ stays at 20%, you end up with 52 times the number of issues to sort out. #tljr
As you scale things up, you need error rates to go down for the absolute number of errors to stay the same. #tljr
[discrepancies/opportunities to engage/whatever. It’s a mismatch that needs human intervention to sort out.] #tljr
DHS tries to get around this by forcing payment recipients to do all the work of fixing stuff up. Otherwise DHS would have to do it. #tljr
But it can’t be completely outsourced. People need help, and ring up. Because the absolute number of cases went up, the load went up. #tljr
Unless DHS added resources to cope with it, their utilisation rates will have to increase. #tljr
We know that DHS didn’t add staff, so the existing staff will end up with more work. Whoops. #tljr
The phone system was already not coping with the load (28 million busy signals) and if you add load to an already overloaded system? #tljr
“But who could have predicted this would happen?” Uh, anyone who knows stuff about operations management. #tljr
Anyhow, let’s move on. #tljr
More Selective Quoting
Some more selective quoting, but note that OCI doesn’t work if the input data isn’t accurate. #tljr
It is important to note that should the information available to DHS be incomplete, the debt amount may be affected. The Ombudsman noted that ‘it is important for the system design for customers to respond to information requests from DHS so decisions are made on all available information.’
Here’s the full part of the Ombudsman’s report. Note what the Senators left out. #tljr
We are also satisfied that if the customer can collect their employment income information and enter it properly into the system, or provide it to DHS to enter, the OCI can accurately calculate the debt. After examination of the business rules underpinning the system, we are satisfied the debts raised by the OCI are accurate, based on the information which is available to DHS at the time the decision is made. However, if the information available to DHS is incomplete, the debt amount may be affected. It is important for the system design for customers to respond to information requests from DHS so decisions are made on all available information. We have therefore concentrated on the accessibility, usability, and transparency of the system, including quality of service delivery and procedural fairness in this report.
The maths works, but only if the input data is accurate. Which is the problem everyone has been pointing out: when the data is shit. #tljr
The Senators then skip from p1 to p7 for their next cherry-picked quote. #tljr
This approach was endorsed by the Ombudsman which reported ‘In our view, it is entirely reasonable and appropriate for DHS to ask customers to explain discrepancies following its data matching activities as a means of safeguarding welfare payment integrity.’
It’s picked out from a weird place. The stuff around it is not positive. #tljr
DHS makes a decision about whether a debt exists based on the information it has available within the OCI system. This relies on a person being both willing and able to accurately verify their earnings for the review period. If the information available to DHS is incomplete, the amount of the debt may be affected. The case studies in the next section illustrate this. We asked DHS whether it had done modelling on how many debts were likely to be over-calculated as opposed to under-calculated. DHS advised no such modelling was done. In our view the absence of modelling means DHS cannot say how many debts may be under-calculated or over-calculated and by what margin.
The risk of over-recovering debts from social security recipients and the potential impact this may have on this relatively vulnerable group of people, warrants further consideration by DHS. We suggest DHS test a sizeable sample of debts raised by the OCI. The samples should include people who did not respond to the initial letter, as well as people who went online and people who contacted DHS via other channels. We also suggest DHS re-evaluate where the risk for debts calculated on incomplete information should properly lie and investigate whether there are ways to mitigate this risk.
The Ombudsman agrees with me: DHS hasn’t designed the system well and needs to redo the incomplete info correction part. #tljr
The Senators again selectively quote from the Ombudsman’s report: #tljr
This is just stupid. If I were the Ombudsman, I’d be pretty pissed off about this hack job. #tljr
The Coalition Senators “reject the view … that DHS has reversed the burden of proof onto recipients.” Well, you’re wrong. #tljr
I submit to you that the Coalition Senators are best placed to inform us what the inside of their colon looks like. #tljr
This is just stupid. If DHS can’t maintain accurate records, that’s not my problem. I only have to tell you once. #tljr
Coalition Senators recommend that DHS continue to invest in its data and analytical capabilities be further improved. This ought to include an integrated case selection methodology that draws information together from data sources such as annual tax returns, financial income, company tax, foreign pension, family day care and trust income.
Developing a ‘whole of recipient’ review capability to inform an enhanced case selection would improve the recipient experience and interaction with the welfare system, further enhance the integrity of the system and more efficiently use Australian Government resources.
Also note the trend here: DHS should be able to access all data held on you, wherever that might be. #tljr
Hell, they can’t get it right with the info they have now. Who knows what kind of fuckups they’d achieve with *more* data. #tljr
Rather than fix this mess, the Senators appear to want to build some magical big data platform that will … I dunno, *handwaves* #tljr
Wasn’t there something about not providing practical suggestions? Senator, I don’t know that you’d understand me. #tljr
The Senators turn their attention towards communications. #tljr
The Senators re-state various facts about the need for good comms, and that changes were made [after DHS was forced to]. #tljr
What’s with the overuse of the word ‘reject’? It’s basically just saying “Nuh uh!” like a 6 year-old. #tljr
This statement isn’t followed by any logical argument based on facts. They just ‘reject’ it. What tools. #tljr
I’m adding to the “abject mockery now” rule. Use of the word ‘reject’ without any followup gets the point and laugh treatment. #tljr
The Senators use their dissenting report to agree with the actual report. For some reason. #tljr
With regard to the OCI design, testing, and governance, the Senators also agree. In their dissenting report. Because reasons. #tljr
Do Senators not have access to a dictionary? #tljr
It’s okay to not know what a word means and to look it up. It’s a great way to extend your vocabulary. #tljr
The Senators continue in their dissent to… agree with the main report regarding debt collectors. #tljr
Maybe the Senators were feeling a bit left out and wanted some attention? #tljr
The Senators regurgitate DHS’ statement some more (honestly, who’s in charge here, you or the department?) that Privacy Is Good. #tljr
The Senators’ dissent basically boils down to this, which isn’t much. #tljr
Coalition Senators reject the central conclusion of the Chair’s report that the OCI process lacked procedural fairness. Coalition Senators, as did the Government and Ombudsman, acknowledge that communications early in the OCI rollout lacked clarity and gave rise to potential confusion on the part of recipients. However, at no stage did this constitute a lack of procedural fairness as review avenues remained open to recipients, and still do to this day – any person with a debt arising from OCI can request a review and provide new information at any time.
Note that you can’t provide new information when the portal doesn’t work. Or the phonelines are busy. #tljr
Or your old employer no longer exists. Or you’re in a physical Centrelink office. So, not any time. #tljr
The Senators really aren’t very good with what words mean, are they? #tljr
Then there’s this gem: #tljr
Coalition Senators further note the input from some third parties, such as #notmydebt, which were aiming solely at scoring political points and inflaming the situation rather than offering practical assistance in resolving the issues raised.
Let’s have a chat about this, shall we? #tljr
First of all, the issue of “offering practical assistance”. We have. Heaps. DHS ignored us for months. #tljr
If you still want “practical assistance”, I run a consulting company, so you can bloody well pay me. #tljr
I’m not cheap, either, because I actually know what I’m doing. Unlike the clowns who built your system. #tljr
Now then, “inflaming the situation”. I’m pretty sure DHS did that by sending thousands of people debt letters at Xmas time. #tljr
Also the “not consulting anyone at all” bit and fucking up the system design so amazingly. #tljr
Who knows what we’ll inflame next? Maybe the budget? An election or two? I dunno. We’re super-powerful, apparently. #tljr
Then this “aiming solely at scoring political points”. That’s obviously nonsense. #tljr
It’s all been right here on the Internet for anyone to see for themselves! How dumb are you? We can all see you! #tljr
And while political point scoring wasn’t the main goal, let’s take a little look at that, shall we? #tljr
The hashtag has been active for *months*. We got an Ombudsman’s report. A Senate inquiry. OAIC is investigating DHS. #tljr
We scored *loads* of political points, you scored nil, and we did it as a *side-effect*. #tljr
Not bad for something cobbled together by volunteers in their free time largely online and with zero budget. #tljr
And that’s where I think we can end this edition of Too Long; Justin Read. Huzzah. #tljr