Data for (Self/Business/Democratic) Governance – Alternative Data Governance // Economies
31 May 2021

Prof Julian Thomas, Director, ADM+S
Dr Jake Goldenfein, Associate Investigator, University of Melbourne node
Prof Karen Yeung, Partner Investigator, University of Birmingham
Peter-Lucas Jones, Deputy Chairman, Maori Television Service
Watch the recording
Duration: 1:15:06


Prof Julian Thomas:

Welcome everybody, to this seminar on data for governance. And this is the fourth and final seminar in our series on alternative data governance and alternative data economies. My name’s Julian Thomas. I’m the Director of the ARC Centre of Excellence for automated decision making and society, known as ADM+S. We’re very pleased to be hosting these seminars together with our colleagues in the centre for artificial intelligence and digital ethics at the Melbourne law school, at Melbourne university. And the humanizing machine intelligence project at the Australian National University. We’d like to begin by acknowledging the people of the Woi Wurrung and Boon Wurrang language groups of the Eastern Kulin nations, on whose unceded lands we’re working here in Melbourne. And we’d like to respectfully acknowledge their elders past and present, and the centre also acknowledges the traditional owners of the lands and waters across Australia.

Now this seminar series, as you’ll know if you’ve been following it or watching it on Youtube, we’ve been exploring alternative strategies for governing data and the digital economy, and what we’ve been trying to do is to stimulate some rethinking of questions around data governance by looking a bit more closely at what data is and how it works in the digital economy. How the laws of organizations interact with the conditions of the digital economy. How platforms can be understood not merely as actors in a market but as actors who are also making markets. And we’ve been thinking about how data can be repurposed as a tool for collective and democratic action, and for alternative modes of governance and control. And that last point is very much the theme I think, of today’s talk. Today’s talk is actually around data for governance and we’re thinking of governance here in terms of public administration. And also in terms of governance of communities and peoples and their languages. So, I will start, if I can, by introducing my colleague in the ADM+S centre, Dr Jake Goldenfield who’s been convening these seminars from the law school at the university of Melbourne. Jake is an investigator in our centre and a brilliant cross-disciplinary theorist of data governance, whose thinking is continuing to shape our work in this field. You may have seen on that slide at the beginning that Mark Andrejevic was listed as a panellist, as a respondent or discussant in today’s program. We think Mark might be able to join us at some point but he’s had a clash with another event which has restricted his capacity to participate at any event. I’ll hand over to you Jake now, to introduce our speakers and share the discussion. Thanks.

Dr Jake Goldenfein:

Thanks, Julian. Thanks everyone. So it’s extremely wonderful for everyone who’s able to join us now and for all of those who I anticipate will be able to watch this, the recording of this, on Youtube when it’s posted. We’ve had really good following, sort of a asynchronous being – like one of the new university buzzwords – asynchronous following of our program so far. Once again, thanks to Julian and to the ADM+S centre, and to Kate at Melbourne law school and HMI at the ANU. If you’ve been following these seminar series you’re probably tired of seeing me and hearing my voice, and we were hoping that you might get a reprieve from that but not this time unfortunately. But in the spirit of minimizing my participation, i’ll share the chairing duties to whatever extent I can, with Julian. But I want to take the opportunity now to introduce our speakers for today’s seminar. Professor Karen Yeung is the university of Birmingham’s first interdisciplinary chair, taking up the post of interdisciplinary professorial fellow in law ethics and informatics at the university of Birmingham, in the school of law and the school of computer science. And she is an honorary professor at Melbourne law school. Professor Yeung is actively involved in several technology policy and related initiatives in the UK, and worldwide, including those concerned with the governance of AI which is one of her key research interests. In particular, professor Yeung is involved in the UN global judicial integrity network, and is a former member of both the EU’s high-level expert group on artificial intelligence and the council of Europe’s expert committee on human rights dimensions of automated data processing and different forms of artificial intelligence.

Her former policy roles include chair of the Nuffield council on bioethics working party on genome editing and human reproduction, and membership in the world economic forum, global future council on biotechnology. Professor Yeung is currently a principal investigator on several projects from various funding sources including participating in our ARC centre of excellence for automated decision making in society. Previously she was a welcome trust funded scholar seeking to investigate the legal, ethical, technical and governance challenges associated with using blockchain in healthcare contexts. Her recent academic publications include the book algorithmic regulation, co-edited with Martin Lodge, published in 2019 with oxford university press. And the oxford handbook of law regulation and technology, co-edited with Rodger Brownsword and Eloise Scott Ford, in 2017. And she’s an administer, admitted to practice as a barrister and a solicitor.

Now, our second speaker today is Peter Lucas Jones. Peter Lucas is deputy chair of Māori television and the CEO of Tahiku Media, which is an award-winning broadcasting organization and innovation hub based in Kaitaia. He has led the Te Gu media group to create a suite of natural language processing tools which enable the creation of new digital products and services that leverage Māori speech recognition, speech to text systems, audio transcription systems, text-to-speech, and other language tools. He has a keen interest in data and ethics and led Tahiku media to develop the largest labelled and tagged Māori language corpus specifically created for Māori language. And now this system represents the first time that automatic digital transcriptions have ever been available to Māori, opening up a range of applications from the transcription of archival audio recordings to the development of speech interfaces for computers. He also informed the development of the Kaitiakitanga data license that Tahiku media use in their strategic technology partnerships. As a leader in broadcasting, Peter Lucas understands issues of digital exclusion and solutions associated with digital inclusion. Peter Lucas is also chair of the national EOE radio network, the society whose objectives is the advancement of Māori radio communication development and operations, and the Māori radio network consists of 20 radio stations that serve the Māori population.

Peter Lucas is also a board member of Matatini which is a centre of research excellence hosted by the university of Auckland. Now, with those long introductions I will pass it over to our first speaker professor Karen Yeung, to talk about her research project on new public analytics. Thank you so much.

Prof Karen Yeung:

Thank you, Jake. It’s a pleasure to be here and firstly, and I don’t think I’ve had a chance to congratulate Julian on the establishment of the centre. And it’s really, really wonderful to be part of your great network, and a real privilege to be involved in all the exciting work that you’re doing. And now I gather that I’m just going to speak for a little bit of time and then we’re going to have a kind of chat discussion? So, I haven’t prepared much in the way of comments but I just wanted to introduce you to what I call my neglected pet project, which is on the new public analytics. And I’m going to tell you a little bit about this idea and where it comes from. And I’m delighted to say that even though I started talking about this in 2018, other people have very kindly credited me with the term the new public analytics. I haven’t actually managed to write this up yet because I’ve been so distracted with other projects, and the reason why this is my neglected project is because it’s not funded. And so all the funded stuff gets priority over everything else, as well as policy work. So, I am sorry for those of you who keep saying, when are you going to write that up? It’s coming, I promise.

Okay, so let me say just a little bit of background where I’ve also spoken about this, and if you don’t mind moving the slides on so I can just give a little pointer to where I’ve been with this before. So, I initially started sharing these ideas in Hong Kong and in winter back in the summer of 2018. And I started developing a set of ideas that were really trying to understand a particular set of digital transformations which were taking place that I was observing, not just in Britain, although I started my interests in Britain, but also in other places around the world. And so, I was really looking for some kind of analytical framework to help draw together some of these ideas as well as to situate these developments within kind of historic transformations in the public sector, as well as providing a platform for thinking about what the implications might be. So, this led me to start developing this idea of the new public analytics. And then since that time I’ve thought about it further and originally I was planning to give this as a keynote at the Montesquieu lecture in Tilburg last year, but then Covid happened, and so that would have forced me to write up this paper but that’s being postponed. But I will, I promise, write it up. So, if we move on to the next slide when eventually I do actually get to give that lecture in fact, I subsequently developed it into a series of three, so it’s a trilogy. But again, the other two haven’t been written up. But the second lecture was for the Swedish network, so I just want to, by way of introduction, tell you a little bit about this concept as I’m trying to develop it. And then from there we can have questions and I can explain a little bit about the motivation. So, in terms of what is this new public analytics?

Well, in order to explain what it is, I guess it’s helpful to step back in time a little bit and refer to what I call the predecessor to the new public analytics. And for those of you who are scholars of public administration or public management or public policy, or even public law, it’s more than likely you’ll be familiar with the term the new public management. So, the new public management was a term coined back in the 1980s to denote a new or renewed emphasis on the importance of management and production engineering in public service delivery, and it was primarily a program of institutional reform across the public sector, associated with the turn to managerialism and often linked to doctrines of economic rationalism and neoliberal ideology – particularly faith in the power of markets and competition in driving efficiency and quality improvements in public service delivery, although scholars differ in relation to the key traits they identify with it. But its central doctrinal content was a mix of ideas drawn from corporate management and from institutional economics or public choice. And so, one of the critical features of the institutional restructuring associated with the new public management was an organizational separation based on a crucial claim distinction between policy matters on the one hand, and operational matters on the other, reflecting a central doctrinal tenant of public sector managerialism which rested on the idea of giving heads of public organizations more discretionary decision space over operational matters, in exchange for direct accountability over their actions. And so wouldn’t express the ostensible objectives of new public management as to get away from the multiple and conflicting goals of public policy, and instead set clear transparent missions and guidelines, leaving managers free to manage within clear parameters.

So, one of the motivating forces behind new public management was this attempt directly to mimic private sector managerial logic. So, as consumers and corporations behaviour in the private sector changed, there were direct demands for government information and transactional practices to change in parallel ways. So, we see this movement of organizational reform sweeping across the public sector in the UK and US, but also it started to infiltrate many other countries. And so it became understood as a phenomenon that was happening all around the world and probably have reached its height by the mid-1990s. So, Carol Harlow, she’s a leading public law scholar at the time in 1997, she refers to the overwhelming infiltration of new public management into public administration describing it at that time as a permanent feature of the administrative landscape. So, that’s where we’re by in the mid-1990s. But by around 2005 or by 10 years later, Christopher Hood who was the scholar who originally coined the term, remarked that in his view arguably new public management no longer held sway over the public sector reform agenda. And we’ve moved beyond that even though the term still has use. So, I guess one of the questions I was interested in is well, what happened after that time? What was going on in the public sector, and in particular how can we understand a particular set of digital transformations that were going on in that time. And I can say something about the evolution from new public management to new public analytics but I won’t do that now. I can do that in the questions if you’re interested.

So, I want – now that having set the scene if you like, and talked about the predecessor to the new public analytics which was the new public management – I now want to talk about what I’m referring to as the new public analytics. So, if we fast forward to today, we can see that public administration appears to be – at least in my view – embarking upon what’s shaping up to be another significant wave of transformation marked by rapid and widespread embrace of aggregate decision-making systems, AVM systems, which rely on digital automation, albeit with varying degrees of sophistication. And these tools and systems are increasingly being deployed and experimented with across many domains. And for a wide range of purposes. I’m not going to offer you a taxonomy, but if you want a wonderful summary and description of some of the applications, do have a look at the algorithm watch automating society 2020 report where you see a great description on a wide range of ADM systems in many different contexts. And you can see that they range from very simple rule-based digital automation systems that are really not much more than a digital vending machine, through to those which utilize machine learning in real-time surveillance systems, including those that purport to identify suspicious behaviour in public places. I deliberately use the term purport and many of these systems are intended protectively to identify vulnerable risky individuals, whether that’s children, students, or potential criminals, through to using chat bots to offer citizens automated advice and automated content analysis systems to digest citizen submissions to public consultation.

So then, just quickly some core features of this idea of the new public analytics. I highlight seven features and I’m going to quickly whip through them, very happy to talk more about it in the chat. So, I argue that we can identify some distinctive technologies, logics, practices, and features that combine to form an emerging paradigm, or ideal type if you like, in public administration service delivery, which are characterized by firstly algorithmic regulation as their underpinning mode of governance or form of social ordering, that ultimately relies on the mathematical logic of computational algorithms as their basis for ordering and coordination. Secondly, they draw heavily – although not exclusively – on a particular form of knowledge that is produced by methods theoretically grounded in data science and statistics. So, considerable emphasis placed on applying software algorithms to large data sets to find hidden insight often aimed at predicting future behaviours, and therefore capable of producing that holy grail actionable insight. This knowledge is then used to support, inform, and increasingly to automate decisions. And so you can see that within this form of transformation, there’s critical reliance on data where basically the more data is assumed to be better. So, identification and processes of data collection are critical and it also means that we need to have an extensive and embedded data infrastructure in order to collect this data. Although, in current practice, a lot of the data that’s being utilized is old-fashioned administrative data that is collected manually and episodically, not in this kind of smart way that’s touted. Fourthly, that this knowledge is used to inform decisions in various ways, increasingly to automate decision making so we can take the human out of the loop and deliver services directly without the need for human intermediation. Although the extent to which that happens is highly variable, but nonetheless these systems are often part of a much more complex set of socio-technical systems often resulting in decisions that have very significant impacts on people, objects, and in the environment. So, many practices to score rank, to trigger some kind of action. Whether that’s to prioritize, distribute, manipulate, admit, or exclude automation, varies from very simple automation through to smart automation.

Six is the aim of these systems is to optimize outcomes, and in that sense there’s a direct borrowing from the language of commerce. The aim being to mimic their successful deployment, particularly in the use of predictive analytics. But unlike the private sector, the overarching substantive goal of public service delivery is not quite so clear. And finally it claims to be politically agnostic. And by that, I mean neither affiliated with left or right political agendas, presented as devoid of grand political aspirations, although we’ll see it’s a deeply political project despite invoking and relying – and I’m going to suggest a series of five ideological claims. Notably, the ideologies of automation, dadaism which includes algorithmic ordering, smartness, responsibilisation, and the seamless user experience. So, on that note, I will shut up and I will hand back to you, Jake. Thank you.

Dr Jake Goldenfein:

Okay, thank you so much for a whirlwind tour through this idea. It sounds like there’s really so much in there that I think we’ll have an opportunity to unpack, but just from the outset I think it’s an incredible intellectual innovation to make so clear, this connection between some of those particular ideologies of public administration as they’re re-articulated through different technology sets of technological logics and practices. Now, I suppose shifting register somewhat from the idea of how a state might use data to govern citizenry, our next presenter Peter Lucas Jones, will talk about how particular groups within a state or with different ideas of statehood, might themselves govern data. Rather than be the subject of governance by data. So, Peter Lucas, if you’re ready to talk about your projects, please go ahead.

Peter-Lucas Jones:

Very honoured to be amongst you this evening, and of course I support the expression of acknowledgement and respect to the traditional owners of the land where you guys are based in Melbourne. I just want to share my screen, so just do that quickly, and I’m going to just do my presentation. Can you guys see that? Now my name is Peter-Lucas Jones, and the umpire that I am going to be speaking about this evening is about indigenous data governance and it’s about recognizing sovereignty. And it’s not just about data, it’s about people, it’s about place, and it’s about our identity. Māori data sovereignty is the specific angle that I’m going to take, and I’m talking about our haka papa. And now haka papa is our connection to our land, to our language, to everything that we hold dear or that is special to us. I’m going to offer a Māori radio perspective and I’m going to be talking about trust and accountability. And trust and accountability is everything when you’re a guardian of data. I’m going to be talking about from a Māori perspective, and that is about Māori self-determination, indigenous self-determination. And when we talk about that we’ve got to express some understanding about data repatriation. Why? Because the data that is held by universities, that is held by people and organizations that don’t belong to that data, need to think about how the people of that data can be better connected with that data. And so, I’m talking about digital opportunities for indigenous people. Why do we exist? And I’m talking about Te Hiku Media, the organization that I am the CEO of. We exist to reverse language loss and language decline as a result of systemic racism, and we do that through promoting Māori language as a preferred medium for the transmission of all things Māori.

This is a picture of our people walking. Walking, why? Because that is how we had to present the petition to have our language recognized as an official language of this country Aotearoa, New Zealand. No matter where do we come from, how did we get here Treaty of Waitungi is an agreement signed between different chiefs of different tribes from Aotearoa, New Zealand, and the British crown. Now, my role as CEO of Te Hiku Media in our organization, belongs to the five tribes of tihikwa tikka, which is the very far northern tip, the top of the island there of the north island. And the tribal groups from that area Ngati Porou.

We’re 30 years old now. December last year, our organization turned 30 years old. We were born out of the movement that was the Māori language claim and of course our claim for our rights in terms of radio frequencies and now we’re moving into participating in claiming rights for our people in so much as 5g is concerned, and things like that. So, it’s all about issues of ownership. It’s all about security, and it’s all about governance of that data. I’m specifically talking about Māori language data, but just to express some point of view on this matter, I think it’s important to recognize that 52 percent of the jail population – the men in jail are Māori – we only make up 16 percent of New Zealand’s population. 63 percent of the jail population as Māori woman in the women’s jails. 63 percent. Like I said, we only make up 16 of the New Zealand population. So, when we think about AI, when we think about bias, when we think about machine learning, we’ve got to consider the governance of that data because much of that data is biased. And if we’re feeding models or training models on data that is biased, we’ve got to take a step back because what does that tell you about incarcerating Māori people for example? That’s something for us to think about, but I’m just drawing our attention to that because this data governance conversation is so wide and we’ve got to be able to think about things from many perspectives. Our vision is there.

One important thing for me to mention is we have a proverb amongst our people and it is, our language is the life essence of our people.

This is a photo of my own grandmother and her niece and her cousin, and what we have here is copious amounts of audio data where we have elders from our tribal groups that are offering stories in our language. And of course we’ve been collecting that now for 30 years. Data is the new oil. Data is the new goal. We must protect our opportunity to grow not only ours but our moko Puna, the members of our people, our grandchildren. In the future, one of the things we’re looking at is the intergenerational transmission of sound. Many of our people have had this data beaten out of them through what we call colonial processes of white assimilation. So, we often regard these things from a cultural position, and starting with, we must look back to look forward. Because the sound of our people has very much been influenced by colonization and the fact that the dominant language is English, and when people say languages naturally evolve, that is true, but when a language is beaten out of generations of your people, that’s not an evolution, that’s assimilation. So, decolonizing the sound of our language through data governance, data management, and of course the way that we apply our values and our principles to artificial intelligence, is really important for us. It’s about speaking native sound into our future, and language is a significant repository for our knowledge. Indigenous knowledge has value, all of a sudden it’s popular, it’s fashionable to have an affiliation, to have some connection with this type of data. And we must protect that for our own people. And the reason I say that is because equity is something that needs to be considered. And so our approach is affirmative action for Māori and other indigenous people, in so much as we offer access to our API, our tools, those sorts of things. With that in mind, where did we start with this? Well we were a radio broadcaster that had been broadcasting for many years by recording the stories of our people. We had recorded the stories about every river, every tree, every mountain, every beach, every type of plant that exists in our environment and our landscape. We then decided to make a conscious effort to transmit this data to those of our people that are alienated from us. So, the best way for us to do that was through digital connectivity. We were mindful that many of the people in our tribal groups did not live in our tribal areas. They had either moved away in search of work, or their families had relocated for some sort of reason that was related to livelihood and well-being. But we wanted to find a way to connect them back with who they are. This is one of our marae and we were one of the first stations or first Ewing tribal broadcasting groups to start broadcasting our tangihanga or our Māori funeral rights, which go for many days. This is the tangihanga.

And so, I worked with Mao and Nainoa Thompson, and people of that calibre, to reimagine star navigation, which is the way that our people had moved from Hawaii or in the pacific through to Aotearoa.

So, we also broadcast Kapa haka, or traditional dance, which is a major attraction point for people. It’s a way for us to connect people that are living outside of our tribal areas to the activities, topical issues, current affairs of our time. One of the important projects that we did was looking at how could we create a digital matrix of idiomatic expressions, colloquialisms, preferred words, things like that, from our archives. And so, what we did was started tagging and labelling, transcribing large quantities of Māori language, acoustic data that takes such a long time, especially when you’re doing it by hand. What we found was we had mapped not only our language but natural examples of these idiomatic expressions. Idiomatic expressions, colloquialisms, are the cornerstone of native speakers. What we found was that we were unable to do what we needed to do. There just was not enough of us. So, we decided we had to teach computers how to speak our language. So, why do we do it? It’s about revitalization. Why does revitalization exist? Well, like I said before, we have generations of our people that were systematically exposed to the assimilation policies that they beat the language out of our parents and our elders. I was blessed to grow up with the language, but many people of course are not. So, we wanted to create tech opportunities for our own people and we knew that as Māori people, we knew we needed this technology transcribing our native speaker radio archives as a window into our past. But it’s also a window into our future, how we used open source tools. And we captivated the passion of our community. And what did that happen, what eventuated from that was some automatic speech recognition. So, we built our own web app and then we had a competition and applying cultural intelligence and data governance is so important if what you want to do is engage with people. So, just to draw a picture, there was a couple of other projects that were happening for like six months that gathered 500 hours of English language, labelled data. Kike I mentioned before, we are only 16 percent of the population, and we gathered 316 hours of labelled data in 10 days. And that was pretty amazing for us, like I’m going to tell you the truth. You’ve got to manage the quality of the data, you’ve got to find a winner, you’ve got you know people. Ringing up Peter-Lucas, my auntie’s cousin’s niece did not, she got a thumbs down and not a thumbs up in terms of the quality, you know what I mean. And so you’re managing comms with your community. It’s major, because even when they’re wrong, sometimes they’ve got a right to be wrong and you’ve got to work with your community to work out what’s the way forward, who owns this data.

I always say that is the wrong question. That is the western question. That is the question from the point of view of making money out of the people that have been marginalized for generations. And so what we talk about is indigenous god guardianship, and that when you grow 30 years of trust with a community to do the right thing, you have a different level of accountability. And that accountability is back to our eye, that’s back to the people of the land, the people that we are. So, what are the principles and the values? One of the principles is we must always maintain sovereignty. Why? Because now earlier I talked about the whakapapa and that’s about creating opportunities. It’s not just about the history. It’s about creating opportunities in the future and we’re mindful of the land grabs, and we’re creating digital landscape for our own people who have been marginalized for such a long time. And when we think about how land was privatized, we think about the colonization of our islands, we cannot run away from the fact that data is the new oil. I just want to take us back – if we give tech companies access to our data, could they sell language as a service back to our people, the very people that beat it out of our ancestors. Could they? Well, if there’s money in it, they will. So, I just want to draw our attention to a couple of examples.

Now, I’m not sure maybe some of you are familiar with the aloha situation, where native Hawaiians were being sent cease and desist letters for using Hawaiian words. Now, that is by white American people. Their own businesses that then capture indigenous words and then copyright them. Who does the language belong to, not who owns it. Who does it belong to? We look at this thing from Jill Lingo. More people are learning Irish on Duolingo than there are native Arab speakers. For a marginalized community, a remnant community of native speakers of a language, that’s a real slap in the face. Particularly when generations of your people have had to carry the burden, carry the future of that language by maintaining it as a method of communication. Lionbridge is an artificial intelligence information group that contacted not only us as Māori radio, but native Hawaiians, other Māori groups. Why? Because they were wanting to create ASR for our language. I want to talk about the ancestry.com DNA thing as well.

Everyone wants to go and get their DNA, you know. As indigenous people, we have to be mindful of targeted medicine. We have to be mindful of what actually is contained in our DNA. We belong to communities, we belong to a collective. Very rarely, things considered from an individual’s perspective. As individuals, do we have a right culturally when I talk about those values and those principles, to then give away what’s contained in our DNA? How might we license our data? Well, we have our own approach and we call it a kaitiakitanga license. We operate by principles of Māori data sovereignty and there are many different principles from different perspectives. We’re a Hokianga based group, so our principles are very Hokia-based. Now when we talk about profits, how are the profits that are made out of the data that belongs to people channelled back into those communities? That’s something that we always think about and that sits at the heart of our kaitiakitanga license. What we’re doing at the moment is we’re developing a multi-language platform and it’s about ASR for our languages. We belong to a language group that includes Tahitian, native Hawaiian, Samoan, Tongan, Te Rio, Māori, of course Cook Island Māori, and the Marquesas Islands and easter island, and other groups as well. So, I had some demos to show you guys. I don’t know how much time we’ve got, but I just wanted to give you a quick example.

So this is our license. Essentially it means, it says talk to us, we are not supporters of open source because open source is for the privileged. Open source is great when you’ve had a wonderful education and the opportunity to learn how to make the best of a situation where data is consumed. Our position is one of affirmative action for Māori and indigenous people. Just quickly – oh sorry, so that’s pretty much it. You know, I could go on and on about this, but I know that I’ve only got 20 minutes and I’ve probably talked over that. We created – can you see that? So like here, we’ve created the first Māori language speech. I’m just going to auto transcribe that. So, this is this morning’s news. So, we do news bulletins in our language every day. This is the first time that this has been developed for an indigenous group but we’re mindful that in creating this, we also create an opportunity for those that want to mine our language, because our language is the home of our indigenous knowledge.

So, that’s an example of what we’ve created and I just want to thank you guys for giving me the opportunity to talk. I mean, repatriation for data of data for communities is really important. I’m talking from the perspective of my community but I encourage all people in all communities to take a new approach to data covenants.

Dr Jake Goldenfein:

Thankyou Peter-Lucas. That was really brilliant and fascinating, and thank you so much. I want to open up a discussion, initially with our panellists and our attendees, please feel free to enter questions into the Q and A. And while you’re formulating those questions in your mind, I just wanted to say how fascinating that series of projects and that history that you’ve just described are, and also to talk about how clearly it falls into some of the things we’ve been discussing in this seminar series. We’ve been talking for you know, we’ve been oriented around this idea that the existing system of laws that structures data flows and data relations between primarily, platforms and individuals, but also states and individuals, offer to individuals this idea of control that is sometimes interpreted as one of ownership, like property, and sometimes this sort of idea of control over identity. But how precisely it’s not the kind of control that matters, it’s not the kind of control over data engenders, it’s not that it’s not the kind of control that gives a capacity to self-determine or to govern through data. And what you’ve just been talking about is a different set of relations to data, that is precisely the kind of control that is meaningful in this context. So, thank you so much for sharing your experience and your knowledge with us.

So I’ll first ask our panellists if there are any questions. I also note that Karen has limited time and might have to depart a few minutes early, just to add on to this special flavor of this particular event, but please identify – I can see, Jeanie, is that a question?

Participant 1:

Thanks so much, Jake. It was more a comment than a question but I guess I have a question that goes to both of our speakers, and I thank you so much for your presentations which were really important and elegant, quite frankly. So, thank you.

My question is, and I suspect it doesn’t have an answer but it’s really your observations because, we’ve heard from Peter about this sort of concern about data colonialism that you know, this snaffling up of indigenous knowledge for making product, for making profit and selling that information back to communities, and Karen touched on this issue of using data to embed I think, assumptions about the way people should behave in society and then use that to further discriminate and re-entrench inequalities. So, we know that these risks are here, of data being used under a supposedly objective framework, but to in fact continue colonialist or oppressive practices. But my question is this – the other argument is that it’s really important for people who are underrepresented in mainstream societies, or for indigenous peoples and knowledge, or for oppressed people or disenfranchised people to in some way be able to use the insights from predictive analytics and data to be able to advance their own cause. And how do we do that in a situation where to be included in a data set, opens the possibility for oppression, but not to be included means that decisions are being made that leave you out. I don’t know how we begin to think about that. I’d love to hear your thoughts.

Prof Karen Yeung:

I think probably Peter should start on that one and then I’ll follow up. Peter, that’s alright?

Peter-Lucas Jones:

Ok, and thank you. I think that’s a really important question because it gets to the heart of this very quickly. And so the project that I lead is the first ever time that a non-university group was ever awarded this type of investment for a data science project. And I think what we need to do is be able to think beyond the classroom. We need to be able to think beyond the university and think about how we encourage and invite investment in community-led data science projects, because in that we invite a different style of leadership. And then we invite a different perspective and we invite a different future. And I think we’ve got to really think about what is our preferred future? Is it what we’re dealing with now, or is there something better? And so I would suggest that like, when it’s about indigenous data one thing I noticed about Australia is every time we’d go to a conference, I was like why are the indigenous people not invited to speak about their own kaupapa, their own topics, their own issues? There was always a non-indigenous person that was leading the discussion. That’s got to stop. That has to stop, and I think when universities encourage leadership from the community, we invite a different result.

Dr Jake Goldenfein:

Karen, did you want to follow up?

Prof Karen Yeung:

Yeah sorry, I just lost sound at the end of Peter-Lucas’s comments. So, it’s fantastic. Amazing initiative, Peter. Really wonderful to see that. So, I just want to speak to Jeannie’s question and fundamentally, the thing that’s so troubling about everything that’s going on is that data is presented as disruptive and transformative, but it’s just reinforcing on steroids existing structures of power, actually. And it’s accelerating the power of commerce over civil society. And so, the fact that institutional Loki’s of skills and data becomes really crucial, and what we’re seeing is you know, if you like that the university industrial complex where money is meted out, I mean the money that goes into the engineering and computer science faculties to support public/private partnerships to build services, military or otherwise, that will generate lots of money. And it’s all corporatism by another name. They try and – you know that the movie is how do we capitalize on our academic expertise in the computational fields in order to promote a very particular agenda, which is not really about as Peter says, it’s not really about the people at all. It’s mostly about money and power. But this is not new, right . This is actually the heart of the capitalist project. And so, it becomes so important then I think to do two things, as Peter-Lucas says, we need to be driven by a vision of what we want our future to be, not simply by the momentum of capitalism. And this is one of the things I think, that’s really missing in the debates. It’s governed by tropes about transformation and disruption, but is this the kind of disruption that we want? Nobody asks is this actually positive disruption? And then of course if communities don’t have the skills and the resources and the energy which is so unusual to see Peter’s heading like this, then those institutions and capacities are not harnessed from the bottom up and we still get top down control, that simply reinscribes the power. So, I think there’s a real need to move out from this kind of, try and find these alternative pockets, if you like, where we can motivate – be motivated by different concerns. And I love this, you know, for me the whole debate about data governance is still so problematic because of the unique properties of data. Because they are data, is instantly copyable and transcribable and transferable at almost no cost instantaneously. So, the troubles once you’ve let it out there as clearly as Peter-Lucas is very conscious of, it can be captured and exploited by somebody else, and we’re still – I don’t know how we grapple with that problem. But I do think it’s one of the biggest challenges of the age which is inherent in the characteristics of data. So, we use this privacy paradigm about you keep control, but I call this the naked photo problem. Once my naked photo’s out there, it’s up there, there’s no getting it back, even though I might want to get it back. And that is really one of the fundamental problems of data governance. That once I share it and I share it on terms that I trust you to share, to respect what it is, that I want you to do with the data, but actually I can’t control that. And so you know, I really wish – Peter looks well in the guardianship model – but I’m worried for you that there’ll still be data leakage and data escape, and you’ll be vulnerable to exploitation because that’s the way unfortunately the dynamic of daily use and exploitation is working. And I don’t have the answers to that, but it does trouble me. Sorry, Jeannie, to be so pessimistic, but that’s where I stand on that one.

Peter-Lucas Jones:

I think that’s a really important point because we know that we have always been taken advantage of in the past and history has an unusual way of repeating itself. And so in terms of this particular project that I’m leading, I think the fact that we achieved the 10 error rate in Māori language ASR was less important than the fact that we showed that this sort of data science project can be led by the people that live in the community, and that’s I think, the paradigm shift that is the mind shift that we’re hoping to share, in terms of a message. That it doesn’t have to be led by the universities, and I mean, like once you have universities leading this sort of stuff then you’re dealing with all the ethics, all the things that are alien from our culture. And we’re mindful around the whole data storage stuff, like we’ve got our data principles and our values and that’s where we’re heading with this because the whole legal framework around it is moving and changing, but our values and our principles remain consistent. And we’re worried too, you know Karen, we’re worried too, and I think the important thing is that this is not an issue, it’s a movement. And as an issue, we want people to be aware of this movement and people to become part of this movement, and that’s I mean, I hope Jeannie, I hope that somewhat answers your question.

Dr Jake Goldenfein:

Wonderful, thanks so much. I find the idea of the necessity and the importance of a community-led data science project so compelling, and it’s something that I think a lot of people are thinking about in different ways. And it’s certainly something that’s reached kind of the academic mainstream in a particular form, in the discussion around intermediaries like trusts and fiduciaries. But often you know, one feels a bit of dismay that this is just sort of a repackaging of an old story. And actually, it’s Karen, one of your colleagues Sylvie Delacroix’s paper about bottom-up data trusts I think, was really interesting. And it raises these questions around how a community-led data science project sort of defines the purpose that it orients itself around. Now like, in your example the purpose is really clear and really important, but in sort of other contexts like the consumer internet where you have disparate groups of people who are trying to coordinate around a meaningful purpose, it becomes perhaps a more difficult story to tell. And I wonder if you have any sort of experience of ideas, of different notions of purpose, or different communities that form to to pursue these projects?

Peter-Lucas Jones:

Is that for me or Karen?

Dr Jake Goldenfein:

That’s for whoever would like to answer.

Peter-Lucas Jones:

Okay cool, well I’m going to talk about water resource concepts. So like, GNS sort of modelling of water aquifers and stuff like that, you start to think about well, who wants a star. Why is the modelling around aquifers so important? Is it because water is a human right or is it because water is a type of gold? And I think that’s why I get back to community-led data projects, because it’s the data and the governorship of that data, and I believe that indigenous people have a place in leading the governance of that data. Why? Because we talk about country, we talk about being the people of the land and the land is connected to the water. The land doesn’t stop, it goes and goes, and then the water starts. You know what I mean? But under the water is still the land. And so my point is that when it comes to mapping or any sorts of data, like particularly in those areas, still the governance of that data is so important. Because who’s to say that that data won’t be used in some way to penalize the people in the future? And that’s my point of view. I’m not saying that that’s the only point of view but I think we need to protect our right to grow, and in protecting our right to grow we must take into account that the digital environment is the last frontier of colonization. They already took our language, they took our land, and then they take your culture and then they appropriate it, and then they want to be you and sell you back to you again. So, my point is that it’s so important for us to be part of the governance conversation because that’s I think, what really has an impact on the way that people might use the data.

Prof Karen Yeung:

So, I just want to jump in there and highlight some of the difficulties we have in terms of data that’s collected about groups and populations, and of course the classic example is the race to exploit MHS data, right.

So the British government thinks wow, this fabulous NHS and the medical records of the entire population, how valuable is that. Let’s monetize that, literally off the back of the British public, without their consent and without being able to do almost anything about it now. And the problem is of course, once it’s anonymized it’s no longer personal data. So, it’s not covered by contemporary data protection or it can be exploited in that way. And of course the rhetoric about that is, but imagine all that fantastic insight that we could glean from that data in terms of medical research. And that’s absolutely true. We do want medical research to benefit populations. But the trouble is that there are multiple applications of that data in ways exactly as Peter-Lucas talks about, can be used to exploit the very people whose data it is. We’ve seen this in relation to national biobanks for example, that were initially motivated by a desire to benefit particular populations, to understand their genetic heritage better and downstream decades later it gets exploited for profit. So, this is a real, it’s a really intractable problem that hasn’t as yet been solved. And i don’t know how you do the data governance there but this is again, it’s associated with the properties of data. It has insight inside it and people want to capture it and exploit it, and how do we impose limits on that that enable those whose data it refers to, to actually retain sovereignty over that data. I don’t know the answer to that, but I see it as a really serious problem. I think it’s a privacy issue as well, you know. As individuals, it’s a privacy issue. But then there’s also the collective and we’ve got to have a – well, I think we’ve got to look at this from so many different points of views and the more people that are participating in the conversation will enhance our understanding of what the problem is now, and what the problem could be in the future, if we don’t take a new approach to the way that data is governed.

Dr Jake Goldenfein:

Thank you so much Karen. A question from Damian.

Participant 2:

Thanks, Jake. And thanks to Peter-Lucas and Karen. It was really interesting and I did appreciate I suppose, I appreciate it might be the wrong word but the reference to the Irish language one was, I definitely enjoyed that. Well maybe enjoyed is the wrong word, as well. But yeah, I suppose my question is for Karen a little bit. Because I mean, I think it kind of fed throughout your talk but I was wondering, I mean a lot of these discussions around automating the public sector, they’re always framed in terms of efficiency. They’re always thinking oh well, you know, let’s spend less resources and get more impact and this false dichotomy within it. Like you know, the fact that I suppose, efficiency gains aren’t really achievable in that sense. You end up throwing money down – well basically. So, I’m wondering to what extent like this kind of false goal of efficiency feeds into your thinking, because I suppose in your seven points, I can’t remember what it was but you had like different philosophies underpinning data’s and things like that. And I’m just kind of curious as to where you see efficiency play because in my mind, I think it features kind of everywhere.

Prof Karen Yeung:

Yes, that’s a really good question, and I’m still struggling with this, which is to what extent is this a neoliberal project. Because the new public management was very much a neoliberal project and it’s definitely motivated by efficiency drivers, or should I say, the more for less challenge. So, there’s definitely a move towards reducing labour cost, yeah. And in the longer term, so even though there might be big set up costs to automate, in the longer term you’ve reduced your salary cost and so it’s hard to know. It’s probably too early to say whether the efficiency mandate has been delivered because we haven’t had enough time to see whether they’ve recouped on the investment, if you like. I suppose one of the things I’m still trying to work out is to what extent. So, new public management was very much a kind of developed western democratic economy movement, if you like, and so I don’t think we saw that taken up so much in other kinds of political economies or political communities. But you definitely see the take up of analytics in authoritarian states, for example, which are not driven by neoliberal idol ideologies in the same way. And so one of the questions I’m actually struggling with is to what extent do I want this concept to do work, that can cross different kinds of political communities. Or whether I want to simply focus on western capitalist social conditions. And I’m not sure about the answer to that. I suspect that focusing on the kind of capitalist agenda will give the concept more analytical purchase, but I’m not really sure. So, I’m really actually interested in your feedback on that, because I haven’t pinned that political dimension to the world yet. I think you can see particular instantiations of the way it’s played out in western contexts as a neoliberal project but perhaps you could argue that it plays out in authoritarian states according to a different set of logics and trajectories, and maybe it would be interesting to contrast those. But I’m not sure, I’m genuinely open-minded as to which would have more analytical purchase. And so, I’d really welcome comments on that. I don’t know whether that answers your question.

Dr Jake Goldenfein:

Just following up on that, that’s a really interesting thing that Damian, you raise, and Karen responding to. I suppose, I read inherent in the sort of the project of new public management. Not only a sort of reconfiguration of the mode of governing a citizenry but also a reconfiguration of the mode of governing the administration itself. And so, there is built-in this sort of logic of what the role of a state is and what the role of an administration is. I wonder if perhaps you see those logics extended through what you’re theorizing as new public analytics?

Prof Karen Yeung:

Yeah. So, that’s a really interesting question and one of the things that developed alongside the new public management was increasing centralization in terms of trying to ensure control over different arms of government. Because if you’re giving more discretionary power to the discrete little pockets of public service, to delivery, then you wanted to make sure if you’re at the centre, that you exert tight control over them anyway. So, in some ways this is what Scott referred to as a mirror image development. This development of bureaucratic regulation which they called it at the time of the new public management scenarios. And I think actually that logic feeds nicely into a more datafied environment where you have your central control room and your dashboards. They give you a window on everything, at least this is the rhetoric that central controllers have. This amazing control room they can step in and have a bird’s eye view of god’s eye view, they call it, of everything, and intervene whenever they feel. At least that’s the rhetoric that’s put forward to it. So, this centralizes control more tightly because you get real-time data. At least that’s the intention. Certainly what happens with surveillance cameras and if you roll that out at a more fine-grained level into every single domain in which you’re trying to digitize, then that gives you quite a lot of centralized potential to at least see what’s going on and attempt to intervene, even if you can’t do it successfully. So, I think you can certainly see the appeal for the central controller. Wow, that would be quite nice to have.

Dr Jake Goldenfein:

Yeah, fascinating. And I suppose it plays into the idea of the design of data infrastructure as well, and can control over data flows. And it links very much into what Peter-Lucas was talking about, about the necessity of communities effectively having control over their own data infrastructures and the ways in which it engages with broader data infrastructures like those of the state. Peter-Lucas, were you going to jump in with a comment?

Peter-Lucas Jones:

Yeah, just want to say thank you for having me. I’ve got another zoom meeting. It’s funny, we live in Zoom world now, and yeah I just want to say thank you Jeannie. I really enjoyed your question. I just want to say it’s been great. Damian, your question, your responses Karen, thank you for having me and inviting me Jake. And I hope to be part or contribute or anything in the future, if you guys think it would be of value to the work that you’re doing. Thank you so much.

Dr Jake Goldenfein:

Thank you so much, Peter-Lucas. I appreciate that. Kia Ora. And I note that Karen also has to depart soon. So, maybe this is actually the time that we’ll wrap things up. Special tThursday evening format. I’ll just hand things over to Julian to end everything up, but thank you both so much. Really wonderful discussion, really meaningful, and I enjoyed it really so much. Thank you.

Prof Julian Thomas:

Thank you very much Peter-Lucas and Karen. And thank you Jake. I mean, thanks everyone for being involved. I think it’s been a really extraordinarily interesting evening. Two very different accounts of how we have to expand our understanding of what’s at stake in alternative modes of data governance and also what’s possible, what alternatives are possible.

I wanted to especially also thank Jake who’s designed and coordinated this whole series and of very lively provocative stimulating conversations, papers, discussions. They’ve all taken slightly different forms. Sometimes they’ve gone in unexpected directions, they’ve opened up all sorts of areas for further work. They’ve involved a remarkable set of speakers including those we’ve heard tonight, from Australia, and across the world. So, very big thank you to all of those. But I don’t think anyone, I can’t imagine anyone apart from Jake bringing this particular combination of people together. So, a very big thank you to you Jake and to all our panellists and respondents and discussants, and participants who’ve asked such good questions which have really enlivened these conversations.

I just by way of finishing, also to thank our partners at university of Melbourne centre and the ANU, and to our colleagues in the ADM+S ARC Centre. Loren Dela Cruz and Kathy Nicholls who’ve really made this possible in terms of making the technology work for us in interesting and challenging circumstances sometimes. Lastly, just a reminder to you all that there are recordings of these talks that are available on our ADM+S centre YouTube channel and this one will be up there in due course. So, really encourage people to jump back and have a bit of a look if you’ve only seen one or two of these, and please keep in touch with us for news of upcoming events. We’d really like to keep you involved. So, thanks to everybody and goodbye for now. Thank you.