EP16: Data, Governance, and Public Service: Ian Oppermann

WEBVTT
00:00:10.308 --> 00:00:18.876
Welcome to Making Data Better, a podcast about data quality and the impact it has on how we protect, manage and use the digital data critical to our lives.
00:00:18.876 --> 00:00:27.012
I'm George Higley, partner at Blackstuff Consulting, and I'm delighted to join my partner, steve Wilson.
00:00:27.012 --> 00:00:28.394
Great to see you again.
00:00:28.394 --> 00:00:30.405
Let's get right into it.
00:00:30.405 --> 00:00:34.088
We've got a terrific guest today, ian Opperman.
00:00:34.088 --> 00:00:43.688
Among his many accomplishments and we'll ask Ian to you have to brag about them yourself he's on Wealth to Check a couple of years he's advisor to a legal tech startup.
00:00:43.688 --> 00:00:53.991
He was struck by your role as a professor at the University of Technology in Sydney, most recently as chief data scientist for the state of New South Wales.
00:00:53.991 --> 00:00:58.299
Nsw has been a leader in data analytics and open data.
00:00:58.299 --> 00:01:09.034
So, ian, we're really delighted to have you here and look forward to hearing your thinking and a bit of the story of New South Wales and the Digital Driver's License and a lot of other topics, all right.
00:01:09.034 --> 00:01:11.305
So, Ian to you, love to hear about what you're doing today.
00:01:11.640 --> 00:01:14.629
George, great to be here and great to be talking about data quality.
00:01:14.629 --> 00:01:26.054
As I mentioned during the introduction, I was the Chief Data Scientist for New South Wales government for eight and a half years, an experimental role which was really about helping government think about data differently and see data differently.
00:01:26.054 --> 00:01:28.027
Now left and I've joined a startup.
00:01:28.027 --> 00:01:40.611
Of course, the thing you do after being in government is join a startup and that startup is all about helping governments do government differently and, of course, using data, ai and digital tools to do things differently.
00:01:40.611 --> 00:01:54.552
Joined with a former minister, the former CEO of what's called Service NSW and the former minister's chief of staff, and we're out there doing good things with data, digital and AI to help other governments around the world.
00:01:54.956 --> 00:02:06.230
Fantastic, and I myself had the privilege of consulting to Service NSW on and off, especially through COVID, and seeing firsthand you know the agility of that organisation.
00:02:06.230 --> 00:02:17.188
It recently won an award actually for service innovation, which was just such a great result and a real testimony to the quality of the people there.
00:02:17.188 --> 00:02:32.189
The innovation included things like the Data Analytics Centre which Vian I think you had a title role of some sort there, but maybe you could tell our listeners about the importance of data and how New South Wales really sort of walked the talk in open data and open government.
00:02:33.259 --> 00:02:53.491
So I joined New South Wales government in 2015 as the first ever chief data scientist and was given not only the gun and the badge, but also a piece of legislation that said you had to share data with what's called the Data Analytics Centre, and the centre was all about tackling wicked policy challenges or problems that are complex, subtle and ultimately have people's behaviour at heart.
00:02:53.491 --> 00:03:14.280
We adopted the philosophy that if we're going to tackle a wicked problem, we need to deliberately re-complicate that problem and try to address as much of it as possible, and that, I have to tell you, was really an unusual philosophy to take in the government, which typically is about keeping it simple.
00:03:14.280 --> 00:03:21.597
That was helping government think differently about how to use data to understand problems differently, use data sets a myriad of different data sets to see the problem differently.
00:03:21.597 --> 00:03:22.906
Every data set is incomplete.
00:03:22.906 --> 00:03:26.050
Every data set is less than 100% quality.
00:03:26.050 --> 00:03:39.391
Every data set doesn't give you full coverage, but it's a point of life, and the whole philosophy was let's see how many different ways we can understand that problem and then use analytics and AI to really make sense of it and do things differently.
00:03:40.520 --> 00:03:44.352
The flow I was taking to Service NSW was now we better understand the problem.
00:03:44.352 --> 00:03:49.530
Let's deliver differently, and my colleague, david Rees, was really the champion of Service NSW.
00:03:49.530 --> 00:04:01.912
He took the philosophy that government can, government can do this, government can be agile, government can experiment, government can try, test and learn and build out services in an agile way.
00:04:01.912 --> 00:04:06.211
We had all been working together for some years when COVID hit.
00:04:06.211 --> 00:04:16.430
And when COVID hit, we were ready to respond and that was the really big difference between before and after that beginning of that experimentation in 2015.
00:04:16.430 --> 00:04:27.990
When COVID hit, we could respond with data-driven tools and data-driven elements of identity and data-driven ways of delivering services, because we'd spent years experimenting up to that point.
00:04:30.380 --> 00:04:37.271
And of course there's a really good social contract, I think in Australia at least, for the use of data, the reasonable use of data.
00:04:37.271 --> 00:04:41.985
You know our COVID response in New South Wales is very strong through New South Wales Health.
00:04:41.985 --> 00:04:52.430
The public more or less embraced the idea that contact tracing was going to be a thing and you know we had such a rash of private sector contact tracing cloud solutions.
00:04:52.430 --> 00:04:58.930
It was a mess and I think it was Victor Dominello himself who said we should make this easy and uniform for citizens.
00:05:01.141 --> 00:05:02.725
But I also want to like tag.
00:05:02.824 --> 00:05:06.627
The point you just made about complicating things, I think is genius.
00:05:06.627 --> 00:05:16.853
I'm a mathematician and scientist by training and I think about generalizing and in engineering we try to simplify, look for simplifying assumptions.
00:05:16.853 --> 00:05:20.029
In science, sometimes we're looking for that grand general theory.
00:05:20.029 --> 00:05:33.452
Usually generalization makes things more complicated, but in our area of digital identity, we think that generalising the digital identity problem to a data problem surprisingly makes things simpler.
00:05:33.452 --> 00:05:39.127
That convergence I'm going to come back to that later, the convergence of governance of data and ID.
00:05:39.127 --> 00:05:50.266
But you know, for now can we go big and talk about data sharing, because I think that that's one of the defining themes of the digital economy.
00:05:50.266 --> 00:05:53.752
We're doing open banking in Australia.
00:05:53.752 --> 00:05:54.701
We've got open data.
00:05:54.701 --> 00:06:02.502
How do you think about data sharing generally and what are the societal parameters that you think about?
00:06:03.644 --> 00:06:08.228
So, steve, you've addressed the entire Pandora's box with your set of comments there.
00:06:08.228 --> 00:06:10.127
Let me try and unpack the sub-elements of it.
00:06:10.127 --> 00:06:19.208
The first was around social licence, and it's an interesting position to be in government and be talking about social licence.
00:06:19.208 --> 00:06:22.009
Social licence is typically associated with mining companies and such things.
00:06:22.009 --> 00:06:24.449
You don't get a choice about engaging with government.
00:06:24.449 --> 00:06:41.060
When I started in 2015, there was this real concern that if government was doing data analytics, if government was doing analysis of data at a sort of at scale, that there would be this concern from the public and the perception of public perception really inhibited a whole lot of activities.
00:06:41.060 --> 00:06:48.591
It wasn't the public saying no, we don't want you to do this, it was inside government people's thoughts about what the public might say.
00:06:48.591 --> 00:06:49.942
That was really a limiting factor.
00:06:49.942 --> 00:06:51.807
So we went out there and tested it.
00:06:51.807 --> 00:07:00.547
We got the data and the data was, quite simply, we don't want to answer the same question because we want to deal with one government, we don't want to deal with 15 different parts of government.
00:07:00.547 --> 00:07:08.129
So we actually spent some time using the data to understand how people felt about government using the data Absolute genius.
00:07:08.129 --> 00:07:11.721
We also had the leadership you mentioned, victor W Lowe.
00:07:11.721 --> 00:07:14.730
We had a minister who said this is ridiculous.
00:07:14.730 --> 00:07:17.045
Government cannot be this complicated to work with.
00:07:17.045 --> 00:07:17.988
It has to be simple.
00:07:17.988 --> 00:07:22.324
It has to be like dealing with service companies, because government largely is a service.
00:07:22.324 --> 00:07:34.045
So we had the most high-level leadership constantly battling the ridiculousness of engaging with government, which gave us all a license to do it better and do it differently, to reimagine how government can actually work.
00:07:34.045 --> 00:07:44.581
When COVID hit, the ministers said we are going to do things differently, we are going to do things simpler, we are going to make it easier to engage Into.
00:07:44.581 --> 00:08:06.560
The line here, of course, is we want people to be able to live their lives within the constraints of COVID and what people are likely to do so simple things like QR code check-ins and QR code check-outs, simple things like delivery of vouchers for people to try and stimulate the economy was all done because we all felt that we were enabled and allowed to do things differently.
00:08:06.560 --> 00:08:11.333
And, of course, service New South Wales was the vehicle that allowed that delivery.
00:08:11.333 --> 00:08:17.101
Data Analytics Center was the place that allowed us to understand where best to place those resources.
00:08:17.983 --> 00:08:20.389
But all of that requires data sharing.
00:08:20.389 --> 00:08:22.620
So, steve, I'm catching up with your series of questions.
00:08:22.620 --> 00:08:28.771
All of that relies on data sharing and data sharing, especially within Google, is extraordinarily hard.
00:08:28.771 --> 00:08:32.283
There are three broad sets of reasons why people aren't sharing data.
00:08:32.283 --> 00:08:40.888
They're unwilling, they have concerns about all sorts of things, they're unable, they don't have the frameworks or they believe they're not allowed or actually not allowed to do data sharing.
00:08:41.048 --> 00:08:44.061
And two of those changed really dramatically during COVID.
00:08:44.061 --> 00:08:45.225
People were willing to share data.
00:08:45.225 --> 00:08:48.951
They really understood by that stage the importance of doing it.
00:08:48.951 --> 00:08:55.830
They also had legislation that said explicitly you will and you must share data in order to make things better.
00:08:55.830 --> 00:09:04.184
And so, if we got down to the frameworks of the unable, how do you share data with appropriate governance, especially data about people?
00:09:04.184 --> 00:09:09.969
So privacy, especially data which is sensitive, so sensitive subjects and particular health is a sensitive subject.
00:09:09.969 --> 00:09:17.188
Building out those data sharing frameworks that we had been experimenting with for years suddenly became important to put into action.
00:09:17.188 --> 00:09:24.330
From an advancement of the maturity and data sharing and use within government, covid was amazing.
00:09:24.330 --> 00:09:27.100
It was, of course, the turn of the maturity and voucher use within government, covid was amazing, but it was, of course, a terrible pandemic.
00:09:27.100 --> 00:09:38.951
But from the mechanism to move us forward, it was amazing from the perspective of everyone's maturity and focus improved for the volume of voucher.
00:09:40.361 --> 00:09:44.850
Yeah, absolutely Necessity, classically being the mother of invention in that case.
00:09:46.461 --> 00:09:50.711
So we had been inventing for years, but it was all kind of interesting experimentation.
00:09:50.711 --> 00:09:52.884
The difference was we had to deploy it.
00:09:52.884 --> 00:09:55.840
We had to actually stop experimenting and actually do so.
00:09:55.840 --> 00:09:59.191
In this case, necessity was the mother of implementation.
00:10:00.380 --> 00:10:00.883
Nice one.
00:10:00.883 --> 00:10:01.726
I like that twist.
00:10:01.726 --> 00:10:05.870
Look, there's other things going on, a lot going on.
00:10:05.870 --> 00:10:13.048
We've got a Data Availability and Transparency Act at the federal level in Australia.
00:10:13.048 --> 00:10:17.469
We've got national data archives.
00:10:17.469 --> 00:10:28.296
We've got things like the fear principles that are supposed to provide a governance for data sharing that are supposed to provide a governance for data sharing.
00:10:28.296 --> 00:10:34.212
We had that sort of metaphor about trying to describe data as an asset or as a resource.
00:10:34.212 --> 00:10:40.812
But how do you on the economic importance and maximize the economic outcome of data as a resource?
00:10:42.441 --> 00:10:44.368
You see you're touching on some very, very big issues.
00:10:44.368 --> 00:10:47.350
Let me try and approach this from a couple of different perspectives.
00:10:48.000 --> 00:10:51.530
You talked about the Data Evaluating Transparency Act.
00:10:51.530 --> 00:10:58.125
Legislation describes things at the principle level and that's great.
00:10:58.125 --> 00:11:05.990
That says this is how we're going to do this, but it's a bit like describing the principles of changing a tire or a bit like the principles of how you're going to wire up your house.
00:11:05.990 --> 00:11:19.322
It's useful in the sense of saying this is what we should be doing, but it's not practical in the sense that you consider a group of people and consider them talking about changing a tyre on their car at a principles level.
00:11:19.322 --> 00:11:21.350
But someone needs to pick up the tools.
00:11:21.350 --> 00:11:27.551
Someone actually needs to get in there and do the doing, and a lot of the work so far has all done.
00:11:27.551 --> 00:11:40.027
Whether it's the Intergovernmental Agreement on Banacharing versus the Banacharing-Rose Act doesn't help unless someone is willing to pick up the tools and a lot of the good intentions.
00:11:40.027 --> 00:11:46.390
During COVID we had the Intergovernmental Agreement on Banacharing that says you are allowed, we are willing and you're allowed.
00:11:46.390 --> 00:11:49.750
So out of the unwilling, unable, allowed, it spoke to the two ends of it.
00:11:49.750 --> 00:12:03.182
It didn't help with the frameworks and starting to consider data as actually something which is not just an Excel spreadsheet, not just something that you do a little bit of analysis on If you want to use it for important purposes.
00:12:03.182 --> 00:12:19.476
You actually need people who know and understand what data is, who know and understand what data governance is, and who know and understand how to actually use the data for appropriate purposes and then put guidance, restrictions or prohibitions around the data products you create.
00:12:19.476 --> 00:12:25.303
That thinking still doesn't largely exist in the world, so I don't agree.
00:12:25.323 --> 00:12:25.947
There are purposes.
00:12:25.947 --> 00:12:32.190
People use data for a very, very narrow, specific purpose or they use data in a very narrow, specific context.
00:12:32.190 --> 00:12:33.759
But the general use of data.
00:12:33.759 --> 00:12:43.267
We're only just beginning to understand the power, the complexity, the material nature of data and starting to build frameworks around it.
00:12:43.267 --> 00:12:46.929
Most people think of data as an Excel spreadsheet.
00:12:46.929 --> 00:12:48.868
That's the mental model most people have.
00:12:48.868 --> 00:12:54.532
The practical reality is if that data is important or to be used for an important purpose.
00:12:54.532 --> 00:13:10.071
It's really important to understand where the data came from, how it got to, what the chain of custody was, what the authorizing frameworks are, the elements of data quality, and then assess whether that data is fit for the purpose you want to use it for.
00:13:10.071 --> 00:13:18.955
And then when you create something from that data whether it's a summary or a subset, or an alert, an alarm, a decision or an action or even just an insight.
00:13:18.955 --> 00:13:33.307
It's really important to understand how to contextualize appropriate use of that data product and put guidance, restrictions or prohibitions on how you use that insight so that it can appropriately be applied for important purposes.
00:13:33.307 --> 00:13:35.398
If you're using data for fun, it doesn't matter.
00:13:35.398 --> 00:13:42.001
If you're doing correlations between by consumption and political outcomes, it doesn't matter because that's a fun pilot trick.
00:13:42.001 --> 00:13:44.168
It matters when it matters.
00:13:44.168 --> 00:13:51.567
And if it really matters, it's important to understand the data before you use it, as you use it and after you create those data products.
00:13:52.059 --> 00:13:57.548
That understanding is opening just a door to a lot of people, I mean for these analogies of the past.
00:13:57.548 --> 00:14:01.849
You know diaries of the new world and that's such a terrible, terrible analogy.
00:14:01.849 --> 00:14:03.726
It doesn't fit in so many different ways.
00:14:03.726 --> 00:14:09.279
I guess the assumption is that it's plentiful and you can create value from it.
00:14:09.279 --> 00:14:11.126
But that's completely wrong.
00:14:11.639 --> 00:14:13.447
Data is much more like the new electricity.
00:14:13.447 --> 00:14:17.748
You have to really evoke yourself if you apply data inappropriately.
00:14:17.748 --> 00:14:20.495
And we all, of course, know that.
00:14:20.495 --> 00:14:23.730
You know we all have that basic understanding of electricity from a very early age.
00:14:23.730 --> 00:14:26.769
We know not to stick a fork into a toaster.
00:14:26.769 --> 00:14:28.868
We know not to stick our tongue into a light socket.
00:14:28.868 --> 00:14:30.826
But each and every one of us can make a toaster.
00:14:30.826 --> 00:14:38.629
Each and every one of us can change a light bulb, but if we go to rewire a house, you know you need to get an expert in electrician.
00:14:38.629 --> 00:14:44.068
If you're going to build a power station, you know you need a group of really experts to do that.
00:14:44.770 --> 00:14:46.934
But with data, we all think that we're the home handyman.
00:14:46.934 --> 00:14:50.765
We just pick up the power tools and drill holes in walls and do silly things with data.
00:14:50.765 --> 00:15:09.611
We need to get to the point where we understand that data is extraordinarily powerful or can be extraordinarily powerful, and when it's extraordinarily powerful, we need to build the right frameworks around it to ensure that we do it in a way which is appropriate and we put restrictions, guidance around the data products we create.
00:15:09.611 --> 00:15:28.873
The alternative is what happens now, at least within government People won't share data because they're concerned about what you're going to do with it and not only you, but the next person and the next, next person will do with the data or the data products or they're concerned that someone can stand on one leg and squeak and re-identify an individual.
00:15:28.873 --> 00:15:37.908
So we either don't share or we boil the goodness out of it and release open data, which is not that useful and really really open data sets are just not that useful.
00:15:38.879 --> 00:15:39.884
You asked a very big question.
00:15:39.884 --> 00:15:41.125
You made a very big statement.
00:15:41.125 --> 00:15:49.361
Data sharing can be done in systematic ways those people sitting around looking at the tire and using principles to try and get that tire changed.
00:15:49.361 --> 00:15:57.251
Someone needs to pick up the framework, someone needs to pick up the tools and do the doing around data sharing frameworks that actually make sense.
00:15:57.251 --> 00:16:05.224
But consider, before you use the data, the fitness for the purpose as you use the data, and what to do with those data products once you create them.
00:16:07.388 --> 00:16:10.274
So you said a number of things that are just fascinating.
00:16:10.274 --> 00:16:15.436
You're touching on research, and I think it's a word that is underestimated.
00:16:15.436 --> 00:16:28.097
I think that the word has been bastardized by the slogan do your own research, as if the person in the street can analyze data and understand it and reach robust decisions.
00:16:28.097 --> 00:16:39.292
I think that we've cheapened the respects institutionally for professional research, but there's a fork in the road here and we could go both ways, and you've been involved with both of these.
00:16:39.292 --> 00:16:48.792
One of them is the economics of data, or the beautiful book Infonomics that you pointed me towards, ian, a couple of years ago, so we'll put this in the show notes.
00:16:48.884 --> 00:17:07.667
But Infonomics, I think, was a at work about 2018 and I think it was focused on almost like a taboo topic, which is how do you, how do you realise value, how do you monetise data, now that there's a lot of dark surveillance, capitalism and a lot of dark agendas behind that idea.
00:17:07.667 --> 00:17:14.229
But I think that Laney was trying to make that idea maybe, but I think that Laney was trying to make that idea maybe respectable or governable.
00:17:14.229 --> 00:17:24.924
So that's one thing, and we could talk about the economic outcomes whether or not using economics to get social outcomes is a good idea, and the other one is data quality.
00:17:24.924 --> 00:17:36.474
You drew attention recently to the new ISO 8000 international standard on data quality, and I guess that's going from principles to tools, isn't it?
00:17:36.474 --> 00:17:44.834
So can you walk us through whether or not data quality as a standard is something that's going to lead to some good outcomes as well?
00:17:47.007 --> 00:17:49.675
Once again, you've struggled me until I had to do a box.
00:17:49.675 --> 00:17:53.464
Let me see if I can step through some of those, but that lady's got infinite knowledge.
00:17:53.905 --> 00:17:55.152
It's a treasure chest, isn't it?
00:17:55.152 --> 00:17:58.214
It's not all problems and poisonous snakes.
00:17:59.266 --> 00:18:04.734
It's a chest of some sort with some stuff inside it, and what you make of it is whether it's treasure or Andorra's box.
00:18:04.734 --> 00:18:09.071
So that lady tried to build a framework that said data is valuable.
00:18:09.071 --> 00:18:12.659
We all think about that analogy data is the new oil.
00:18:12.659 --> 00:18:18.394
As much as I hate it, the intention there is to try and say it's valuable, but no one's really sure how it's valuable.
00:18:18.394 --> 00:18:24.627
So what we've done tried to do was put a framework in place that said you can use data for operational purposes.
00:18:24.627 --> 00:18:26.492
You can use data for strategic purposes.
00:18:26.492 --> 00:18:30.606
You can use data to create insights and all of that's sort of the internal uses of data.
00:18:30.606 --> 00:18:34.048
And there are more than simple operational purposes.
00:18:34.048 --> 00:18:37.010
You could and should be using your data for strategic purposes.
00:18:37.151 --> 00:18:38.832
Most organizations do not.
00:18:38.832 --> 00:18:41.594
Most organizations don't realize the value of their own data.
00:18:41.594 --> 00:18:45.817
They also center around some external value aspects to data.
00:18:45.817 --> 00:18:47.759
What would someone claim for it?
00:18:47.759 --> 00:18:50.460
Or what markets would you create?
00:18:50.460 --> 00:18:54.002
Or what new value would you create if you mixed your data with someone else's data?
00:18:54.002 --> 00:18:57.515
And then, of course, there are elements of putting up potential loss as well.
00:18:57.515 --> 00:18:59.585
What would it cost to reproduce the data?
00:18:59.585 --> 00:19:05.131
What would be the impact of you not having exclusive access to that data and I thought that was really, really good.
00:19:05.131 --> 00:19:10.528
And then what I'd been trying to do at the time was work out a government version of that.
00:19:10.528 --> 00:19:12.769
So he talked very business-focused.
00:19:12.769 --> 00:19:15.271
I thought what would government do with data?
00:19:15.271 --> 00:19:22.198
And I'd also been trying to work out how we can build frameworks, literally an accounting framework for data.
00:19:22.198 --> 00:19:24.538
So it was a very nice meeting of minds around that idea.
00:19:24.538 --> 00:19:30.728
We are, to this day, still without an accounting standard which values data and tell.
00:19:30.748 --> 00:19:32.734
Facebook that their data is not valuable.
00:19:32.734 --> 00:19:35.413
It is the majority of the value of the company.
00:19:35.413 --> 00:19:42.734
Tell LinkedIn their data is not valuable and it's the majority of the value of the company, but we don't have a way of measuring it from a finance perspective.
00:19:42.734 --> 00:19:45.374
So that problem is still being tackled.
00:19:45.374 --> 00:19:47.912
But I think we're getting better at saying it's in here somewhere.
00:19:47.912 --> 00:19:49.016
The problem's in here somewhere.
00:19:49.016 --> 00:19:52.365
We don't quite know what it is, but it's in here somewhere.
00:19:52.365 --> 00:19:53.413
The problem's in here somewhere.
00:19:53.413 --> 00:19:54.220
We don't quite know what it is, but it's in here somewhere.
00:19:54.144 --> 00:20:10.528
Data quality is a really interesting element and as much as I hate the oil analogy for data, I often try to help people conceptualize by saying that data comes in different qualities and you can still use poor quality data, but what you can do in terms of reliance on the outcome is quite different.
00:20:10.528 --> 00:20:23.097
So it's like putting in Australia we have 91 grade, 95 grade, 98 grade petrol and we have diesel and you need to know whether your car needs to have or can run on 91, 95, 98, or whether it runs on diesel.
00:20:23.097 --> 00:20:29.212
You need to make sure you absolutely don't mix them up, but you can still run your car, your high-performance car.
00:20:29.212 --> 00:20:30.174
It can still run on 91.
00:20:30.174 --> 00:20:31.910
It just doesn't work so well.
00:20:31.910 --> 00:20:35.295
Or you can run your clapped-out car on 98.
00:20:35.295 --> 00:20:37.692
It might be good things for the engine.
00:20:37.692 --> 00:20:43.270
The point is you need to understand if it's fit for the purpose you want to use it for.
00:20:43.270 --> 00:20:55.096
We trust when we go to a petrol station, we trust that all the work around, where it came from, how it got there, that chain of custody, that chain of problems we trust all that away because we trust the brand of petrol station.
00:20:55.096 --> 00:20:57.599
But we still have to make the assessment for ourselves.
00:20:57.599 --> 00:21:02.375
Is it fit for the purpose I want to use it for?
00:21:02.375 --> 00:21:04.105
Would I inadvertently put diesel into my petrol engine?
00:21:04.105 --> 00:21:08.770
Would I inadvertently put 91 into my high-performance engine, which requires 98?
00:21:08.770 --> 00:21:10.688
So we still have to make that decision.
00:21:10.907 --> 00:21:28.153
Data policy is, if you think about all the effort, all the engineering, all of the process, control, all the governance that goes into getting that petrol into the bowser with data, we just say I've got an Excel spreadsheet, let's just pop it in and see whether we get some insights out.
00:21:28.153 --> 00:21:38.238
We need to put the same effort into the data in order to get the same reliance on the data products that we create.
00:21:38.238 --> 00:21:41.334
And again, those data products can be really, really broad.
00:21:41.334 --> 00:21:46.653
It could be a plot, it could be an insight, it could be an alert, an alarm, a decision for action.
00:21:46.653 --> 00:21:55.371
It could be any one of those things, but in order to rely on that, it's really important that we've done that work around the gardens.
00:21:55.411 --> 00:22:03.105
There are several things that I looked at in the past as chief data scientist, where I ran what's called the AI review committee, and AI is a use of data.
00:22:03.105 --> 00:22:12.157
We looked at really important uses of data like sepsis prediction, emergency department patient emergency department is about to go through a sepsis episode.
00:22:12.157 --> 00:22:17.474
An algorithm says alert and someone rushes over and treats the patient.
00:22:17.474 --> 00:22:20.916
That is literally a life and death use of data.
00:22:20.916 --> 00:22:22.545
The alert is not a decision.
00:22:22.545 --> 00:22:26.135
That arc of alert alert decision action.
00:22:26.135 --> 00:22:37.575
It's not doing something, but it raises the attention of a human being who responds in a heightened state of awareness to say this patient is about to go through a sepsis episode.
00:22:37.575 --> 00:22:44.294
Getting that wrong either false positive or false negative can be really visibly impactful to that person.
00:22:44.294 --> 00:22:46.993
It could be literally life or death.
00:22:46.993 --> 00:22:58.827
In order to trust the data, in order to trust the data product, we really need trust in that governance process that got us to that point, because if someone's life depends on that alert.
00:22:58.827 --> 00:23:07.153
Even though it's a human alert, that alert leads to a heightened state of awareness around responding to a life-critical condition.
00:23:07.153 --> 00:23:09.698
We really need all that governance to be in place.
00:23:10.384 --> 00:23:18.849
So data quality is really important and the standard you enter the 8,000-series standards comes in many, many parts.
00:23:18.849 --> 00:23:27.050
It actually says to us that data quality is not the issue of whether you've got day-to-day, month-to-month year-to-year or month-to-month day-to-day year-to-year.
00:23:27.050 --> 00:23:30.359
It's not about simple things like that simple format of things.
00:23:30.359 --> 00:23:33.189
It's about the entire governance process of data.
00:23:33.189 --> 00:23:35.971
How does data flow into your organization?
00:23:35.971 --> 00:23:41.731
What are the controls and the chain of custody, the chain of authorizing frameworks?
00:23:41.731 --> 00:23:43.551
How did the data get collected?
00:23:44.047 --> 00:23:50.327
It speaks to if I hesitate to say the life cycle of data, because life cycles in data are either trivial or infinite.
00:23:50.347 --> 00:23:52.605
So it tries to speak to life cycle of data, because life cycles in data are either trivial or infinite.
00:23:52.605 --> 00:23:59.791
So it tries to speak to the element of life cycle and it says these are all the things you need to consider.
00:23:59.791 --> 00:24:04.626
Let me revisit that analogy of the petrol station.
00:24:04.626 --> 00:24:12.914
It allows us to re-imagine that from a perspective of data and say there's a lot more to this than just bank format.
00:24:12.914 --> 00:24:20.316
There's a lot more to it than whether you've got the year right to one decimal point or the number right to one decimal point.
00:24:20.316 --> 00:24:24.965
There's a lot more to it and it's really important to know and understand that.
00:24:24.965 --> 00:24:54.700
If your use of data is important, that you really have to at least be aware of these standards preferably have someone you've organised and you understand these standards or know where to go to get that extra help in order to enact these standards, to have appropriate confidence and therefore trust in the data before you use it, the data as you use it and the data after you, the data products you create, what you carry around those and the data.
00:24:54.720 --> 00:24:57.587
After the data products, you create what you can around those.
00:24:57.587 --> 00:25:14.653
Well, ian, steve and I, as we talk about data sharing, we're often using the term data supply chain, because you've got issuers of data and they often get the data to the party that needs it to make a decision, which should be multiple entities.
00:25:14.653 --> 00:25:26.085
In between touching that data, you talk about data provenance and recording the evidence of that supply chain journey.
00:25:26.085 --> 00:25:39.030
One of the major producers of critical data is government itself, creating Medicare numbers, driver's license numbers and many of those my birth records and the like.
00:25:39.030 --> 00:25:53.644
Many of those numbers are used very commonly in commercial contexts, whether that was my design or not, but they still are very important for use for onboarding purposes, for example, creating an account relationship.
00:25:54.006 --> 00:26:00.019
So, stephen, I've been thinking about government, as, of course, you're issuers of these numbers.
00:26:00.019 --> 00:26:15.250
Thus far, they've really been issued either on paper or in plain text, and plain text, of course, is the target of every breach and the whole planet is affected.
00:26:15.250 --> 00:26:35.618
The number of producers or it's all led to go any closer to identifying me as a safe driver or as a driver?
00:26:35.618 --> 00:26:37.669
Or why is this driving me to South Wales?
00:26:37.669 --> 00:26:44.874
Why don't we turn that into a verifiable credential that has security.
00:26:44.874 --> 00:26:50.834
It's presented out of a wallet, out of a security device, so that hardware and software will be together.
00:26:50.834 --> 00:27:09.550
That's the way we think it would be very helpful for government to expand its remit, if you will, in terms of that data issuer Really providing a more useful to society version of the data that it's already producing.
00:27:09.550 --> 00:27:11.651
How hard is that?
00:27:12.484 --> 00:27:14.090
How hard is it to get.
00:27:14.806 --> 00:27:15.027
Again.
00:27:15.027 --> 00:27:15.469
You're laughing.
00:27:15.469 --> 00:27:16.633
That tells me something already.
00:27:16.633 --> 00:27:22.156
How hard is it to get government to expand its thinking about?
00:27:22.156 --> 00:27:25.375
If we are issuers of data?
00:27:25.375 --> 00:27:32.038
How can we get that into society in a more useful way?
00:27:35.008 --> 00:27:38.277
So if I were still with government, you would just ask the career-limiting question.
00:27:38.277 --> 00:27:40.472
So thank you for asking that.
00:27:41.164 --> 00:27:42.571
Well, that's also delighted.
00:27:42.571 --> 00:27:44.211
You aren't in government any longer.
00:27:44.845 --> 00:27:48.977
Yeah, it's not that long ago, so I'm still a little shy with questions like that.
00:27:48.977 --> 00:27:51.207
It's really hard for a couple of reasons.
00:27:51.207 --> 00:27:51.607
Let me wind back.
00:27:51.607 --> 00:27:52.808
Let me try, and perhaps that question must have a stake.
00:27:52.808 --> 00:27:53.410
I'll be going, so I'm still.
00:27:53.410 --> 00:27:54.371
I'm sure I've written questions like that.
00:27:54.371 --> 00:27:55.332
It's really hard for a couple of reasons.
00:27:55.332 --> 00:27:55.673
Let me wind back.
00:27:55.673 --> 00:27:57.375
Let me try and approach that question much like I have with Steve.
00:27:57.375 --> 00:27:58.855
You've asked a really important and big question.
00:27:58.855 --> 00:28:02.381
The first part of that is, of course, that's what we need to do.
00:28:02.381 --> 00:28:11.125
We are stuck in the 20th century thinking about creating honeypot kind of sense, where we bring all the data together and make ourselves a target for cyber attack.
00:28:11.125 --> 00:28:16.857
Once upon a time you had to be interesting for someone to put the effort into trying to attack you from a cyber perspective.
00:28:17.184 --> 00:28:18.952
Nowadays we're all just collateral damage.
00:28:18.952 --> 00:28:21.353
You don't have to be at all special.
00:28:21.353 --> 00:28:24.974
Every data set is going to be attacked at some point.
00:28:24.974 --> 00:28:28.614
It's just the nature of the world we live in.
00:28:28.614 --> 00:28:32.555
So let's not create those data sets in the first place.
00:28:32.555 --> 00:28:38.116
The problem is most people in government can't see a different way forward.
00:28:38.116 --> 00:28:41.074
Most people wouldn't think that that's just the way you do stuff.
00:28:41.074 --> 00:28:42.345
You just build data sets together.
00:28:42.345 --> 00:28:46.934
So there is an important mindset change that must happen.
00:28:46.934 --> 00:28:49.038
That isn't yet happening.
00:28:49.038 --> 00:28:50.826
So that's the real problem.
00:28:50.826 --> 00:28:55.250
Virtualization of data is a way of doing this.
00:28:55.250 --> 00:29:11.362
Being able to give people you were hinting at it, let me make it explicit giving people their own explicit control of their most important data assets, such as their identity, is really an important way of thinking about how we should go forward.
00:29:16.525 --> 00:29:18.134
Most of the time, I don't need to ask you the question what's your date of birth?
00:29:18.134 --> 00:29:20.265
I'm asking you the question in front of government are you old enough to enter this license premises?
00:29:20.265 --> 00:29:22.430
Is A greater than B?
00:29:22.430 --> 00:29:25.318
And if the answer is yes, then off you go.
00:29:25.318 --> 00:29:27.266
I don't need to know your date of birth.
00:29:27.266 --> 00:29:28.528
I don't need to know where you were born.
00:29:28.528 --> 00:29:32.055
I don't need to know all the sort of stuff that people typically will take from you.
00:29:32.695 --> 00:29:35.970
If you try to enter a license for purposes, I don't need to know whether you have a responsible.
00:29:35.970 --> 00:29:40.913
I don't need to know your date of birth, whether you've got a responsible service of alcohol license.
00:29:40.913 --> 00:29:43.346
All I need to know is is it valid?
00:29:43.346 --> 00:29:44.007
Do you have it?
00:29:44.007 --> 00:29:44.548
Is it valid?
00:29:44.548 --> 00:29:45.348
Is it within date?
00:29:45.348 --> 00:29:48.413
And the answer to that question then allows that next step to happen.
00:29:48.413 --> 00:30:01.715
That thinking is a step beyond where most people are, because we're so used to having physical identity that we hand over, which has lots more information that is actually required to answer the question are you old enough to go on this license premises?
00:30:01.715 --> 00:30:04.531
But that's where the work needs to go.
00:30:04.531 --> 00:30:05.134
Sorry, george.
00:30:05.865 --> 00:30:08.974
So I totally get your answer.
00:30:08.974 --> 00:30:15.458
What is it that people like us need to do to change that mindset?
00:30:15.458 --> 00:30:20.317
I mean, honestly, that's kind of the mission of why we might even be doing this podcast.
00:30:20.317 --> 00:30:22.390
We want this discussion to be out there.
00:30:24.115 --> 00:30:38.449
Okay, many, many times I've sat around a table with government colleagues and listened to their plans for how they're going to do things in future and thought to myself you're kidding, you're absolutely kidding.
00:30:38.449 --> 00:30:46.481
Really, this is what we plan to do and then try to help people understand there is a way of doing things differently.
00:30:46.481 --> 00:30:59.755
The hardest thing in the world to do is change someone's mind If they're set on a path of action they've got their planning, they've got their processes, they've got their resources, they've got their budget and you try to say to someone don't do it that way, do it differently.
00:30:59.755 --> 00:31:02.020
It's really hard to do.
00:31:02.020 --> 00:31:07.577
You're an annoying barking dog on the side of the road where they're on a mission to do something.
00:31:07.577 --> 00:31:10.432
It genuinely takes a crisis.
00:31:10.432 --> 00:31:14.713
It genuinely takes a crisis to help people think that they can do things differently.
00:31:14.733 --> 00:31:18.955
Covid was an amazing agent for change for the way we think about data.
00:31:18.955 --> 00:31:22.994
Data breaches, unfortunately, are an amazing agent for change.
00:31:22.994 --> 00:31:28.051
When it impacts a group, they suddenly listen, they're suddenly ready to do things differently.
00:31:28.051 --> 00:31:32.521
But watching everybody else do things or getting good advice, that's.
00:31:32.521 --> 00:31:35.938
That's interesting, george, but I'm busy, busy doing grown-up stuff over here.
00:31:35.938 --> 00:31:37.142
I don't really need your advice.
00:31:37.142 --> 00:31:44.308
So, unfortunately, in some cases, you have to live through a couple of crises, unfortunately.
00:31:44.308 --> 00:31:46.178
You have to help people when when they fall in the hole.
00:31:46.178 --> 00:31:47.243
You have to help people when they fall in the hole.
00:31:47.243 --> 00:31:48.170
You have to help people out.
00:31:48.630 --> 00:31:58.641
Let's say this is allowed doing things differently and the really, at least in Australia, there's been a tendency for governments to de-skill over the last decades.
00:31:58.641 --> 00:32:09.403
So we have a lot of generalists in government who are not technical, who don't get the next level of detail and who just think well, you know, this is what we do, this is how we do it.
00:32:09.403 --> 00:32:18.257
So there isn't a simple answer, but being there to help this thing, these are elements of how we get going.
00:32:18.257 --> 00:32:28.339
So I hope this podcast, I hope someone listens to this podcast and says, oh, maybe I can do things differently, maybe I can look at this differently, but it's very hard to change the mind of someone.
00:32:28.339 --> 00:32:30.478
It's really hard to change the mind of a public servant.
00:32:30.478 --> 00:32:38.038
It's really, really hard to change the mind of a public servant who's got all their pieces, their program logic laid out and their budget and their resources are ready to go.
00:32:38.038 --> 00:32:39.715
It's really tough to do.
00:32:40.269 --> 00:32:52.034
We've got other catalysts that I guess are programs that governments have extensively committed to, like Digital ID now and open banking, do you find?
00:32:52.034 --> 00:33:17.337
Well, it's a leading question, ian, but we find patterns through these programs that are common and they suggest to us maybe there's an emerging sophistication about how to govern data, what the parameters are that are governable, what can be standardized, what should be decentralized and left to be determined on a local sort of basis.
00:33:17.337 --> 00:33:29.297
One of the anxieties I have about ISO 8000 is that we had to see this process or this movement become too robotic in the way that ISO 9000 did.
00:33:29.297 --> 00:33:44.596
It poisoned the well by making quality management like an algorithm or a spreadsheet and it took a lot of the judgment and intuition and humanity out of quality management.
00:33:44.596 --> 00:33:49.517
And you know 30 years and I think ISO 8000 is going to do a better job.
00:33:49.517 --> 00:34:00.637
But are you getting to the point where you can imagine how this weird, intangible thing is going to be governable in a general way?
00:34:01.990 --> 00:34:02.836
Thank you, steve.
00:34:02.836 --> 00:34:04.877
Once again you've answered the right question.
00:34:04.877 --> 00:34:07.017
On the record, I love standards.
00:34:07.017 --> 00:34:08.717
I'm a huge fan of standards.
00:34:08.717 --> 00:34:14.556
Without standards, we are electrocuting ourselves, we are sticking our tongue in the light, we're doing all that dumb stuff.
00:34:14.556 --> 00:34:24.139
There is a tendency for standards to get a little brilliantly focused, a little heavy, and you touched on an example right there.
00:34:24.139 --> 00:34:34.708
I think there are other standards that get so cumbersome that they become less useful than they could be, but here's one I prepared earlier.
00:34:35.251 --> 00:34:46.202
We've just finished, after multiple years of work at ISO IEC, so at JTC1 standard on data usage 2.5207, terminology and use cases.
00:34:46.202 --> 00:34:59.809
Believe it or not, defining the terminology, getting the definitions right, is something that largely didn't exist or happened in a multitude of different environments, and we fought like cats in a bag about getting those definitions.
00:34:59.809 --> 00:35:05.478
Because the cloud people think this, the document exchange people think this, the AI people think this, the smart cities people think this.
00:35:05.478 --> 00:35:07.778
We said let's get that down.
00:35:07.778 --> 00:35:12.114
And then guidance for usage, and one of those usages, of course, is AI.
00:35:12.114 --> 00:35:20.239
So we connected to the world of the AI folks, we connected to TAO, we connected to all this stuff and we said you know, it needs to be simpler.
00:35:20.239 --> 00:35:24.802
It also needs to acknowledge that there is uniqueness in every circumstance.
00:35:27.369 --> 00:35:32.317
There is some of that judgment that's required, but the formula for these standards is, quite simply what do we need to know about the data before we use it?
00:35:32.317 --> 00:35:36.498
What do we need to understand about the data as we assess it for fitness purpose?
00:35:36.498 --> 00:35:40.155
What are the guidance, restrictions or prohibitions we need to put around the data product?
00:35:40.155 --> 00:36:03.300
We've read the thing that I've been sort of slipping into the conversation to know is exactly what's in that standard, and I think it balances the slipping into the conversation to know is exactly what's in that standard, and it acknowledges the fact that context matters, your environment matters, your sensitivities matter, your particular circumstances about your data matter, but most of it 80% of it, by the way actually is a repeatable process.
00:36:03.300 --> 00:36:13.320
But the 10% is your judgment and your context and your environment and you need to do some level setting about layers of control very high control, high control, moderate control, low control, no control.
00:36:13.610 --> 00:36:16.119
You get to decide, but this is how you decide.
00:36:16.119 --> 00:36:19.219
This is a recipe for you to decide.
00:36:19.219 --> 00:36:22.014
You still make that dish, you still bake that cake.
00:36:22.014 --> 00:36:24.898
It's your cake, but this is how to turn the oven on.
00:36:24.898 --> 00:36:26.335
This is the temperature to turn the oven at.
00:36:26.335 --> 00:36:29.797
This is how long to think about it should go on the oven for.
00:36:29.797 --> 00:36:37.360
But you're still doing that subjective confinement to not make it robotic, because data is really complicated.
00:36:37.360 --> 00:36:45.940
It's so complicated that in fact, there are seven degrees of infinite complexity in data sharing and use.
00:36:45.940 --> 00:36:50.788
Seven degrees of infinite complexity, data sharing and use, which we're going into a lot of trouble.
00:36:50.788 --> 00:36:53.945
But that basically comes down to every data set can be used for an infinite number of purposes.
00:36:53.945 --> 00:36:56.697
They might not be fit for purpose, but you can use it for an infinite number of purposes.
00:36:56.697 --> 00:37:05.039
Context changes the sensitivity of data, so you bring that data set close to another data set or you know something about something in the data.
00:37:05.039 --> 00:37:07.438
The context of use changes it.
00:37:08.510 --> 00:37:13.701
All those ambiguities, all those possibilities, try to create the things you create.
00:37:13.701 --> 00:37:36.717
Whether it's a salary data set or a self-set or an insight, that same issue exists, that context matters, and you can use that insight for a range of different purposes and there's a couple more steps in there, but it means that it's so mercurial, it's so difficult, so slippery, it's like smoke, it's like trying to eat your arms around something which is really dangerous.
00:37:36.717 --> 00:37:47.277
So if you can flatten all those ambiguities or uncertainties into a couple of different types, layers of control, then you can start to do something useful with it.
00:37:47.277 --> 00:37:49.978
That's what these new standards talk about.
00:37:49.978 --> 00:38:00.021
It says this is what we mean, these are the definitions and we mean the same thing consistently and these are the ways of thinking about how you use data.
00:38:00.021 --> 00:38:02.480
And therefore it's 5.7,.
00:38:02.480 --> 00:38:08.717
Terminal Newscast is 5.212, a guidance for usage, and one of those usages just happens to be AI.
00:38:08.838 --> 00:38:09.199
Brilliant.
00:38:09.199 --> 00:38:17.701
Let's get those details from you and put them in the show notes later on and see how much of that material is accessible to our audience.
00:38:17.701 --> 00:38:19.315
It's fantastic guidance.
00:38:19.315 --> 00:38:19.858
Thanks, Ian.
00:38:21.632 --> 00:38:22.918
It was a lander of love.
00:38:22.918 --> 00:38:25.110
A lot of landers and a lot of issues.
00:38:25.110 --> 00:38:33.217
People who try to get their arms around it start to talk about things like what's the difference between data and information.
00:38:33.217 --> 00:38:37.360
That one's easier to dial out, but then information and knowledge and wisdom.
00:38:37.360 --> 00:38:40.275
It was three in the morning.
00:38:40.275 --> 00:38:42.016
We were really close to the end of the standard.
00:38:42.016 --> 00:38:44.615
Someone said we haven't talked about wisdom.
00:38:44.615 --> 00:38:49.117
I said no, the ISO definition of wisdom.
00:38:49.117 --> 00:38:50.172
We'll probably understand it.
00:38:50.172 --> 00:38:52.197
If not, we're ending this conversation.
00:38:52.197 --> 00:38:55.518
Wait, I love it.
00:38:56.351 --> 00:39:05.860
Well, I think it behooves us to exercise some wisdom at this point and to bring this to a conclusion.
00:39:05.860 --> 00:39:07.817
Ian, thank you very much.
00:39:07.817 --> 00:39:11.701
I'm really grateful for your insights and your thoughts.
00:39:11.701 --> 00:39:23.757
I'm going to be thinking about how we shift government minds when they already have all the momentum in the world for a particular vision and we're using them over slightly.
00:39:23.998 --> 00:39:24.739
Thank you very much.
00:39:24.739 --> 00:39:25.922
Thanks so much, ian.
00:39:25.922 --> 00:39:30.856
Here's to data and information and, if we can get, to wisdom.
00:39:30.856 --> 00:39:33.909
Thanks for helping us along the way Later.
00:39:33.909 --> 00:39:42.652
Thank you.