[00:00] Mallory Durran:An individual researcher might have to go through 22 different pathways just to have access to the information they need to do the cohort analysis to even get their clinical trial off the ground. And you can imagine the kinds of delays that that represents.
[00:14] Ravi Gurumurthy:Hi, I'm Ravi Gurumurthy. I'm the CEO of Nesta and welcome to this miniseries of Policy Fix where we'll be focusing on what Andy Burnham should do in his first 100 days. We'll be covering lots of topics from energy to today's issue, which is digital public infrastructure. Um, and with me on this subject is Mallory Durran, who is the Chief Technology Officer here at Nesta, and she's a computational neuroscientist by background. She spent 10 years uh trying to make government better in terms of digital services and AI. And um, it's a great pleasure to have her here today. Mallory, welcome.
[00:46]Um, before we get into the conversation, do you want to say a little bit about how you got into this question of digital sovereignty and—and your background in the issue?
[00:54] Mallory Durran:Yeah. Um, maybe haphazardly, actually. Um, but a—a long-standing set of, you know, challenges or things I cared about. So as a very sort of early-career researcher technician, um, I was doing one of those jobs that, um, you know, loads of people do, kind of menial tasks. And, um, you know, in the early 2000s, there was an app for everything, and—and we were sort of Instagram for fitness. Um, and I was manually filtering at the time inappropriate content and photographs.
[01:23]Um, and you know, with the technical background that I had as a—a lover of maths and statistics, I thought I could pick up machine learning and then I wouldn't have to look at these photos myself. Um, and so created a machine learning algorithm to do uh that sort of sorting for me. And I'm sure uh inquiring minds can imagine the kinds of photos you might inappropriately get on a Instagram for fitness app in the early 2000s.
[01:47]Um, but really that was my first introduction into: what are the rules for participating? Um, how do we create environments that allow people to express themselves, uh, but not force content on—on people that really genuinely nobody wants or needs to see? Um, and sort of that machine learning pathway brought me into a larger world of the infrastructure that then enables that.
[02:10]Um, so it's awesome if you, you know, have the—the best and brightest ideas for how a computational model of progression of Parkinson's disease, which is something that I—I worked on in my—my early career, might work. But if you don't have the—the data infrastructure, the compute, um, and the environment in which to carry that out, your—your ideas just won't come to fruition. You can spend, you know, your entire life working it out with a pen and paper, um, or pencil and—and paper.
[02:38]So, um, when I joined the—the UK government, I had these big dreams about how the computational modeling um that I had been doing in my research life might be applied to things like uh winter in the NHS, understanding patient flows, allowing for operational environments to be more prepared or respond uh or react to influxes. Um, but when I arrived, I quickly realized that, you know, the—the digital infrastructure, um, the—the cloud environments, the available compute, the join-up of data, even the standards of what an individual variable might mean across hospital environments was just not in play.
[03:14]Um, and so, you know, everybody talks about the—the benefit or value that data, technology, now AI could bring to our public services, to systems. But none of that is possible unless you set the standards, um, and the—the infrastructure in a way that actually lends or leads toward that. So my, you know, my interest in digital public infrastructure, and indeed the standards and guidelines and principles through which we participate in digital environments, is a—a long-standing journey that's had a—a few wiggles along the way, but yeah, really comes from first-line experience.
[03:48] Ravi Gurumurthy:Okay. So, necessity is the mother of invention. [laughter] And um, we're going to talk about um digital public infrastructure, digital sovereignty quite a lot. But I want to start, because this is a podcast about what Andy Burnham should do, with Andy Burnham, where um in his first speech where he talked about launching Number 10 North, he—he—he didn't actually mention AI at all, but he did talk a lot about public control, and that's been a—a theme already. And that language is quite similar or related to the question of AI sovereignty or digital sovereignty—putting people and the country in control.
[04:26]Um, but when—when you think about sovereignty, you've got to think about the whole sort of layer from energy, compute models, um critical minerals. How do you—how do you think about sovereignty? What is—what is—what is important to own, what is important to, and what is actually tractable? Um, and I think there's a whole spectrum, isn't it, between having it locked down yourself versus being dependent on people you trust versus being able to accept adversaries owning it. Um, so give us a sense of what's important, also what's tractable.
[05:00] Mallory Durran:Yeah, absolutely. Um, I—I think that the—the sort of fabric of that kind of infrastructure lives in Manchester. I think Number 10 North doesn't have to state that it will be, um, you know, technology-enabled or—or AI or have principles of sovereignty, because, you know, we—we owe a lot of what we now think of as AI, computing power, the principles of computing, um, to, you know, bright minds that started and coalesced um in—in Manchester through history. So, um, I think that's a—an amazing base start.
[05:28]But you're right, when we talk about digital sovereignty, there is a tech stack, and that tech stack builds from the cables and wires that move energy across the country. It is how that energy is generated. It moves through data centers into cloud systems, through to data models or the representation of information um in—in binary, in—in—in data models, um through to the—the patterns and protocols that allow application layers—so, you know, the app that you interact with on your phone to be able to call or access that information. And then there's that access layer: so who gets to access what, for which reasons.
[06:08]And all of those are important things to debate in thinking about where you might play in creating sovereignty within the system. Now, I'm not going to pretend that I am an infrastructure engineer and have the—the sort of last line on whether, you know, Britain's grid is sufficient for the kind of demand that might be placed by data centers. Um, but, you know, just in our day-to-day work, even here at Nesta, we do lots of thinking about and trying to find ways around use of AI because of the way that we govern, um, you know, special category or personal data. And if we want to use the most advanced AI models, there actually isn't enough, you know, available compute or power to have data residency remain in the United Kingdom.
[06:58]So those sort of bottom layer of the tech stack and the physical infrastructure that is supporting um is a real challenge and something that I think we need to bring communities into thinking about um how we get that sort of data center um and—and core infrastructure right. A layer up from there, when we start to talk about the cloud, um I think it's really important what our information passes through and—and who holds and governs that along the way. But the nature of the way the technology market has emerged means that there is not a credible UK-based or even um really a competitor to AWS, to Microsoft Azure; there isn't a credible European alternative to that.
[07:44]Um, and so when we think about where to play, finding the component pieces where we can take back control without having unrealistic expectations of the market fundamentally changing and being able to catch up, both in um scale of the offering, the cost at which that can be provided if you're sort of new versus incumbent—um, I think that the—the real opportunity for the UK really lives the next layer up. So in our—our data models, in setting really good standards for how, you know, our public sector data in particular um is held and governed so that smaller players can start to build application layers um and potentially eventually move to compete with some of the other cloud or service providers. But if we get our data layer and standards right, I think we're, you know, much better able to harness um that capacity for economic growth in the technology sector.
[08:41] Ravi Gurumurthy:So—so you're saying that data layer is really, really important. We should probably think about the other layers in the stack. Some of them are more tractable than others. Um, and there are probably some where you think they're important, but they're not actually easy to do very much about. Is that fair?
[08:57] Mallory Durran:Yeah. So let's use an example. Um, so open banking um was a—a program of work taken by the UK financial sector that looked at setting standards, and that means everything down to: how do we represent particular banking variables, how do we hold dates and names, uh balances, right? Everything from the—the—the standard table to um how that in—what the security layer looks like on top of that infrastructure, so that um an—another application could come and draw that information accurately, as well as what security standards needed to exist for an organization like a new fintech startup to be allowed to access that information.
[09:42]So all of these standards were negotiated with the sector and implemented. It's not true that banks aren't allowed to buy cloud services or indeed data infrastructure from American or European or who—whoever they should choose. But in providing those infrastructure services, that company needs to meet the standards that have been set by the UK banking and financial sector. So they sorted out that data layer, as well as the security and access layer.
[10:17]And now if you can meet those standards, you get to play in the space. So we saw an absolutely massive increase in the fintech sector on the back of the sort of 2018 coming in of—of those standards. Um, so you've got, you know, the Monzo and the Revoluts of the world, and suddenly, you know, as a participant in banking, I can look at all accounts across all banks in a single application, and I get to make a choice as a customer about which user interface suits me. And that's only possible because they got the—the data and the access layers right. And that has fueled, you know, business growth and economic benefit, but also benefits for me as a—a customer of banking.
[10:57]So, um, you know, Lloyds or Barclays or indeed Monzo can choose whatever infrastructure provider they would like, but those people have to come in and can only sell services that match or meet the—the credentialing and the standards that have been set by the sector.
[11:12] Ravi Gurumurthy:Great. And so, and you're saying that like open banking, though, is a bit of an exception, because if you take um whether it's energy or health, most other sectors, they haven't created those common standards that allow um for a variety of different players to come in, data to be shared in a trusted way, citizens to have consent over that.
[11:32] Mallory Durran:Yeah, exactly. And maybe ironically, um, the—the problem isn't a technical one; it's a human one.
[11:39] Ravi Gurumurthy:Mhm.
[11:40] Mallory Durran:It is not complicated to decide on a tech stack, but getting a bunch of different organizations with different incentives, um, with lots of players with creative ideas into a single place to agree and then invest in implementation um is really a human problem rather than a technology one.
[11:58] Ravi Gurumurthy:And in terms of the leadership for this, you can place the emphasis on governments to do this, um, but there are lots of examples of technical standards being agreed by industry, like Bluetooth...
[12:08] Mallory Durran:Yeah.
[12:08] Ravi Gurumurthy:...where, um, you know, it's not a sort of state creation, in fact, it's, you know, across multiple geographies. To what extent do you think that the—the digital public infrastructure that we need to create in this country is about government leading, or can it be led more bottom-up? Is there enough um incentive to collaborate because people don't want to necessarily compete on certain things, or does this require actually a bit of government uh direction?
[12:31] Mallory Durran:I think it definitely requires government direction, but I think you're right that there are a good examples of where standards have been, you know, met or decided upon without there being a government intervention. You just need a—a critical mass. Um, and I think consumers or citizens have a huge voice in that. Um, so, you know, a really recent example is all of the different cables that we have had for our phones and small electronic devices over time. I think you're now seeing a consolidation around USB-C.
[13:05] Ravi Gurumurthy:Even Apple had to do that.
[13:06] Mallory Durran:E—even Apple had to move. Um, and that probably, I don't know, reduced the number of sales of cables they might have had because of other providers, but the people wanted it, and the people have power, and the people have a voice. Um, and so I think there's a real moment here for citizens, groups of citizens, groups of users to be voicing where and how they want to see things. And I think—but I think there is—there is potential for particular sectors to come together to—to sort of start the charge. But I think even in open banking, um, you know, there is potential within the financial sector that's not yet been reached. It's—it's...
[13:46] Ravi Gurumurthy:I mean, a lot of people say that open banking was great, but the full benefits were not actually released. We sort of—it was a sort of half-victory, if you like.
[13:53] Mallory Durran:Yeah. And I mean, a—a victory is a victory. Um, and you know, as someone who uses banking applications, I sure am glad for it, but I think there's—there's wider benefit. And I think, um, you know, especially some of the smaller players in the open banking space—so people who might be more like connectors of information, building those mid-layer applications rather than the big banks themselves—will say that they've had an ebb and flow of the quality of imp—implementation of different sort of waves of standards or adaptation, largely driven by how clear and forceful government's position on the implementation of those standards were, and that legislation played a massive, important role in actually getting things over the line.
[14:38]Um, so I think we can't rely on the goodwill of an individual sector alone; often government does need to intervene. Um, and I think there also has to be someone really making the case for what the benefit could look like in order to get that coalition of the willing going in—in the first place. So really clear articulation of benefit within a particular sector to—to move things forward.
[15:03] Ravi Gurumurthy:So why don't you paint a picture of what that could look like? Let me—take imagine if you did open health or open energy. What—what could open health look like, and—and what—yeah, what's the size of the benefit?
[15:13] Mallory Durran:Yeah. So I think health is a—a no-brainer for the UK. Um, we've got a brilliant life sciences sector. Um, and I mean, as a researcher coming to the UK to, and you know, institutions that are older than the country that I—I came from, um, I was so wowed by, you know, being in these classrooms where you know that all of the great inventions of life sciences and health innovation came out of—like, we've—we've got this, you know, globally um unparalleled strength in the—in life sciences.
[15:47]Um, but for a clinical researcher trying to, um, you know, bring forward a new mode of delivering a—a cancer drug, or, um, you know, where data science has, you know, sparked up, "Oh, we could use this sort of GLP-1 instead of for heart conditions, we're seeing weight loss reduction," like what does it look like to measure across in order to get access to the data at the scale required? There are something—HDR UK, the Health Data Research UK organization, did a mapping of all the potential access points. Um, an individual researcher might have to go through 22 different pathways just to have access to the information they need to do the cohort analysis to even get their clinical trial off the ground. And you can imagine the kinds of delays that that represents.
[16:36]But if we got that layer, the—the data model right, if it was, you know, a consistent approach to bringing health data together for research across individual controllers, trusts, GPs, um, the UK would have an unparalleled quality um and scale of connected, high-quality information which could expedite clinical trials. You know, the UK would become the best place uh for drug companies to be doing every phase of trial, which would give, you know, UK citizens access to new treatment and care earlier.
[17:15]Um, similarly, kind of based on that what we saw in open banking, if you've got a way to govern who can then access what is genuinely sensitive information that should be, you know, protected and not given out willy-nilly, that in—in open banking you had, um, you know, Financial Conduct Authority setting standards for um who can access this information and why. Yes, you have to be a certified financial institution to access, but once you meet those security standards, you're allowed to innovate on top of it.
[17:45]And in the—in the health sector, that could mean a huge growth in innovation for personalized healthcare. Um, you know, a—a real health tech sector that has access to information. Whereas an individual citizen, I could say, "Yeah, you know what? I have a rare disease in my family, and I want to have individualized recommendations for me on the basis of what I should be looking for or caring about." Um, and I could choose, you know, within a—a growing health tech sector what was right for me, um, and—and really have agency over um how and who and where my data is accessed and—and used. So I think there's a huge innovation opportunity that sits within it, as well as, you know, research and um health benefits that could be seen in the public sector.
[18:32] Ravi Gurumurthy:So it feels like to me at the moment, though, you've got um huge potential from linking data together. When you do it badly, um, you either create a—a platform where a single company has uh too much power and lock-in, uh, and that has two problems: one, it often creates a consumer backlash, um, you know, you've—you've got a problem at the moment with some GPs deciding that they don't want to play ball with Palantir, for example—you get that sort of backlash; but you also then get the kind of harvesting of benefit to a single player who can lock out others, rather than building quite a rich ecosystem.
[19:06]Um, I mean, India is in—really, India is the sort of the—the obvious example where they started with identity, moved to payments, um, and now they've got a kind of very rich ecosystem off—off the back of it. One example I always like is the example of transportation where um in India, Uber used to drive—used to charge a daily driver commission, um, and then because they created this digital public infrastructure, different apps can connect with drivers um pretty freely, and they weren't charging an annual—a daily driver commission, and so Uber were outcompeted essentially; they had to drop that down. And what they've done is create a world where um gig workers are keeping more of their money, uh, and consumers are also keeping more of the—the surplus. Um, and it's not a sort of a big-state thing; it's a—it's a very minimal public infrastructure with lots and lots of different, um, you know, competitive apps, private and public, coming off the back of it.
[20:06]Is that—I guess, is—is that—is that vision true for other things? Could we do that in parking? Could—one of my favorite ones—um, or, you know, other things that are very consumer sort of—where the consumer benefit is going to be really clear?
[20:18] Mallory Durran:Yeah, absolutely. I think when you get digital public infrastructure right, you fundamentally reshape the market. Um, and I think, you know, all kudos to the brilliant technical layer built by the, you know, the India Stack team. What allowed for that was setting rules, right? Any application, whether it was Uber or the homegrown ones, had to meet the same set of information governance and sharing rules, which is why, you know, um drivers were able to port, you know, not just their credentials of like license and registration across different applications, but also their um uh rider stars. So if I as a driver want to move, I don't lose the credibility that I have built; I'm able to move that across.
[21:04]And that required India to say, "Yep, you can play in our market, but these things have to be true. It has to be true that you can port that information in different ways." And so what they did there was, yes, a very clever technical layer, but actually it's about being unequivocal about the rules for participation: if you want to play in our market, then you've got to do it our way. And that created a totally different kind of competition that had benefits for a wider group of people, or um a different power balance in the structure. And I—I think that that's true wherever digital public infrastructure is done well.
[21:41] Ravi Gurumurthy:Just a word on language: "infrastructure" is an interesting term, because is this essentially regulation where you're defining a set of standards in a marketplace, or is it actually infrastructure in the way we think of it, like a road, where there is a sort of something that—that is being built? I think it's relevant because um sometimes this can make—this can sound almost super technically challenging, and then we start getting pessimistic about our ability to do anything quickly or—or well.
[22:11] Mallory Durran:Roads are an interesting one, right? Because yes, they're physical infrastructure, but there are also standards. Um, you know, if you're going to build a new road, you're not allowed to decide just to make it, you know, 12 centimeters narrower than the other because that's convenient, right? There—there are acts and governance for what we mean by something being safe. Um, and sure, there are lots of examples of streets that um are higgledy-piggledy in all sorts of ways because they were built before, but we're also dealing with that in digital infrastructure, right? Legacy systems that don't meet the um the potential of what is possible now.
[22:48]So I think in digital infrastructure, it's a similar mix, right? There are patterns, protocols, architectures, blueprints, diagrams that you can set out for what it just basically looks like to do it well, but there still has to be some kind of agreement for: what does that look like, and why, in order for it to, you know, work across systems.
[23:13]So, um, yes, it's legislation. You know, as we said, open banking moved as quickly as it did because there was a really strong push from government, and legislation made it clear what was expected by when and why, and then the implementation of the technology and the actual designs kind of sat separate.
[23:36] Ravi Gurumurthy:Mhm. Mhm.
[23:36] Mallory Durran:[clears throat] From that.
[23:37] Ravi Gurumurthy:And just some—give me some more use cases. The reason I like parking is just that I feel like we can do it quite quickly. It's one of those things that annoys people when they have to sort of go to a different parking app every time. Um, um, I think there's—there's almost a: let's educate um the public about the value of things like public infrastructure that could be done quite quickly. Are there any—what are your favorite sort of smaller use cases, if you like, that could be done quickly?
[24:01] Mallory Durran:Yeah. I mean, parking is an interesting one because there are some parking applications that do have an open and shared set of standards. You know, some—some local authorities have deals with—with sort of multiple providers, and therefore you go to a single interface and you can enter your parking location code in and it—and it works, and some that have not jumped on board. So that's a case where there was a coalition of the willing, and some of that comes together in a really neat way that means you don't have to go outside of a single application, and some of it doesn't. And it's because probably there is no governing body that says, "You must meet this now." And so I think that's a good example of where a coalition of the willing can get so far. The infrastructure exists, um, but yeah, it hasn't sort of gone all—all the way.
[24:46]Um, I've got really high hopes for a project that the Open Data Institute is—is leading um to look at pooling volunteering opportunities and the requirements for those volunteers uh across the country. Um, I don't know if you've ever tried to, you know, find something in your local area, you thought, "You know what? For the next few months I've got a bit of extra time on my hands; I'd really like to, you know, go and read in a—in a community center or a hospital." It's actually really tricky to—to find those. Um, and so the Open Data Institute are looking at how they create uh a sort of open layer for how you would pull in all of those volunteer opportunities.
[25:21]And I think on top of that, there's a—a really interesting opportunity for: what does it look like to also be able to open up um the—the credentials required? So if I'm going to go into a care home, for example, I need to have a DBS check. Um, and so it's not just about pooling the volunteer opportunities, but it's actually about saying, "What—what are the requirements for meeting this, and can I create an automated pathway to pull in: does this person have an up-to-date DBS check?" And I think, you know, we strengthen communities, we prove public—digital public infrastructure, um, and we make it easier for people to take a credential that belongs to them and port it across, you know, public-interest use cases.
[25:59] Ravi Gurumurthy:Credentialing is an interesting use case in its own right, isn't it? Could go much broader than that.
[26:04] Mallory Durran:Yeah. And I think there are some really big proposals for things like a digital ID, for example, um, but some really good use cases that fall below the layer of, "We need to legislate for this standard." So, GOV.UK One Login—I don't know if you have updated your passport, driver's license...
[26:22] Ravi Gurumurthy:Um, you know, I did actually. I was on the sofa thinking, "I've got to do my passports," and my wife was going, "Are you coming to bed?" And I went, "I'll be a while; I've got to do this for the kids. We're going on holiday." 3 minutes later I went up to bed and she went—and she was actually annoyed that I'd given up. And I said, "No, I've actually done it all." And it was—that—it was amazing.
[26:41] Mallory Durran:Yeah, my—my most recent favorite example was I had updated my driver's license and then went to apply for my first British passport, um, and was able to use the same photograph because that was all stored within an account, and that didn't require the implementation of a unique ident—identifier across all systems. It's a brilliant technical layer, bringing things together for citizens.
[27:00] Ravi Gurumurthy:In a sense, that's where I was going to go next, which is: we've obviously had the latest attempt at um at digital ID. I'm actually not even sure where we've got to, if I'm completely honest, on—on digital ID, but to what extent do you need to do digital ID as the layer before you can then do the others, as India did actually, or um is there a different route through this?
[27:21] Mallory Durran:Yeah, I'm sure that there are really brilliant people super in the detail of this that might disagree with me. Um, but I don't think you need to do digital ID first, and I think GOV.UK One Login is a really good example of that. And the reason I think that is that even if you create a unique identifier tomorrow for every citizen, there's a significant amount of fixing within existing systems that needs to be done in order to carry out implementation of that sort of unique ID system.
[27:52]Um, you know, one example of that would be, uh, because I had a student visa and then moved to a working visa, I actually am two people in the NHS, and I once showed up to a hospital um, having, you know, significantly cut my hand, bleeding everywhere, and they're like, "Well, which of these two people are you? Which of these addresses is correct?" And we have NHS numbers. It just so happened that I ended up with a—a dual account. So even within systems that have an identifier code, there is work to be done to improve, consolidate, update infrastructure to allow for what is the promise of that sort of digital ID system. So having an identifier does not automatically mean you have joined-up systems with interoperable technology.
[28:36]And so I think we get into this unnecessary chicken and egg where we, you know, don't update our infrastructure to be able to improve join-up because one day we might have a digital ID. But even if we decided and went hell for leather on digital ID tomorrow, there would be base work to get that right as well. And so I—I think it's a false dichotomy, yeah, and that there is more and better that we could be doing—One Login as an example of that, um, like the—the one GOV.UK stuff, um, that would improve citizens' experiences and interactions with applications before you get to the point of implementation that, if we were to decide to create a unique digital ID, um, that would be easier to implement because we had done that underpinning work.
[29:21]So I think we're unnecessarily tying ourselves in knots over the real questions from citizens about why and how you would do a digital ID in a way that was, you know, fair and reasonable.
[29:32] Ravi Gurumurthy:Yeah. And also one challenge we've got in doing in this country is that with India, um, there was a real, huge value in creating their digital payments layer and—and wallet, but, you know, we've kind of solved some of those problems in a different way, possibly suboptimally, but it's quite hard to sort of um have the same transformation here because it's not a greenfield site in quite the same way. Is that—is that fair? Is that an impediment?
[29:57] Mallory Durran:Totally. Um, yeah, I mean, I—I waxed lyrical there about the brilliant life sciences, you know, the history of this country and, you know, the advantage that that brings us. Um, there is a flip side to that, which is if you've got a thing that sort of works some of the time, that can be harder to build upon, or especially where you have 15 different versions of a thing that have sprung up quite organically, actually how you consolidate to a single thing. So yeah, in some ways I do think building from legacy systems is harder than starting from carte blanche. Um, and it really does come back to vision and leadership and ability to create that coalition. So the—the problem isn't actually a technical one; it's a human one.
[30:43] Ravi Gurumurthy:Okay. So um let's pretend that you're now um Andy Burnham. Sorry, Andy. Um, what would you—what would you actually do? What would be, you know, the three to five things you would prioritize now?
[30:58] Mallory Durran:Yeah. So I mean, we've—we've talked about digital sovereignty, and I think making really clear the government position on that. I think lots of private-sector organizations, and in particular infrastructure companies, are trying to think about how they respond to a call for sovereignty. If I'm Andy Burnham and his team on day one, I am going to come out of the gate and say, "The United Kingdom is moving to a standards-based system for procurement, for data models, and we will, you know, appoint the right technical people to make those decisions sector-over-sector in the way that we have done with open banking. Um, but this is how you will play in the UK digital sphere, mhm, unequivocally across sectors forever."
[31:40] Ravi Gurumurthy:So standards-based procurement system, that just means if you're going to play in X market, these are the kind of common standards that will always apply for a long period of time.
[31:52] Mallory Durran:Exactly. And the—the benefits of that are preventing vendor lock-in. So if there's a transparent architecture from your sort of base data model through storage, through APIs, then A, if a provider isn't doing a good job, then someone else can come in and follow that blueprint and replace them. Um, so it raises the quality of standards.
[32:13]It's also true that, you know, in—in most places in the public sector you don't have people who are experienced engineers or architects with the sort of final procurement decision on the systems or services, sometimes because it's not sensible to have, you know, that kind of technical knowledge in every team everywhere that might need a digital service. Um, but it means that, you know, companies trying to sell things have loads of bells and whistles on, and the person trying to buy doesn't know which ones are going to be a meaningful value-add versus, you know, what extras aren't necessary.
[32:46]And so being able to set those standards both means the quality of the service should be better, you have the ability to replace a service more easily and transparently if something better comes around (either in a value-for-money perspective or in a provision perspective), protects you against that lock-in, and will lead to better value for money.
[33:05] Ravi Gurumurthy:And does that require a standards body um either across multiple sectors or in each sector?
[33:10] Mallory Durran:Definitely. Um, I think it's probably sector-specific, but you probably do want a—a single governing body to make sure that the linkages across things make sense. You know, there are common variables between health and welfare and tax systems. Um, and this is boring stuff like how you even...
[33:34] Ravi Gurumurthy:Yeah.
[33:34] Mallory Durran:...define an address or something.
[33:36] Ravi Gurumurthy:Yeah. I mean, like, have you ever, you know, opened a new spreadsheet in Excel and had the dates column turn into exclamation marks? We're basically trying to prevent that at a really base level, right? So that's the first—that's the big thing.
[33:47] Mallory Durran:That's the big thing.
[33:48] Ravi Gurumurthy:What else?
[33:48] Mallory Durran:Um, so I think I—I alluded to it there, but getting the right people into those decision-making positions um and treating talent as a national emergency. I think the public sector cannot...
[34:02] Ravi Gurumurthy:You spent so much time in—in—in government, in Number 10 and in uh the Incubator for AI. You're saying talent is a national emergency.
[34:11] Mallory Durran:Talent is a national emergency. There are incredible people um that exist, but they have a really hard job. Um, it is not easy to be the lone technical voice at a table full of people who want good things to happen, but have absolutely no technical grounding um or—or knowledge. And I think if you look at, you know, procurement systems as an example, you do have to have people who understand the difference between uh federated and fabric-based data service. Um, and that isn't something that can be upskilled in a civil servant overnight; it's genuine knowledge and training and, you know, a need for exper—experience of implementation of that at scale. So you need your risk and decision-making owners to have real technical skill.
[34:59]And then we've sort of not talked specifically about AI, but having a good cadre of engineers is not the same as having frontier AI talent. And I think a—I mean, I'm an immigrant. The UK is a great place to live and to, you know, care about frontier research. I think that we have a moment and an opportunity coalescing around some of the big bets that the government is making on sovereign AI and some of the brilliant companies that have been, you know, built or seeded here, but we've got to make it easy to get the best and brightest talent into those places.
[35:39] Ravi Gurumurthy:How—how optimistic are you on those big bets on sovereign AI? To what extent are they—I mean, the kind of outside critique would be: "This is tiny and we're late."
[35:49] Mallory Durran:I think that that's true. Um, but I think it would be silly to say, "We're small and we're coming late, and therefore we shouldn't try." I think that we're going to need to catalyze private-sector investment and think about commercial models in a different way; it can't just be state investment. Um, and I think that there's a different risk appetite around consolidation of, you know, picking a player and running with it um that needs to happen so that, um, you know, the sovereign AI players don't fall over because we don't have the investment in data centers or compute, which is something we've already seen happen.
[36:24]So I don't think that the journey is easy and that it will happen tomorrow, but I think that it would be um—I think it would be an underestimation of the ambition and capability of AI engineers and researchers already here, and the ones that might be attracted especially to a nation that says, "We're here for openness, you know, open-standards AI, um, and we're here for sovereignty." And I think that there's a huge appeal there, um, but multiple systems have to come together in order to make it possible to get that talent in, to pay them well, um, and to keep them in the UK.
[36:59] Ravi Gurumurthy:If I was um—if I was Andy Burnham now hearing your pitch a bit, we would go, "Okay, great. That's all really wonky plumbing."
[37:05] Mallory Durran:Yeah.
[37:06] Ravi Gurumurthy:"Uh, what's the—what can I promise the country uh if we do all this? You know, give me a sense of in three years or five years, what might be different?"
[37:16] Mallory Durran:Yeah. Yeah. So I think health is probably the—the area that speaks most to people. Um, and if we take that example of having an amazing health tech sector on the back of this, um, I would have more choice in how I access information and support and care um for my health. I think that would matter. I think that, you know, better, faster, more efficient treatment um within our NHS and operational systems and services matters.
[37:48]But I think it's also about economic growth and opportunity. That health tech sector is about providing, you know, better access to different kinds of care and services for people, but it's also jobs, and it's also, um, you know, value creation...
[38:02] Ravi Gurumurthy:When you talk to them enough and say, "Yes, scale-up capital is a problem, but also access to data, clinical trials, is also a big constraint," and this would solve that.
[38:10] Mallory Durran:Exactly. And I—I think, um, if you—if you give access to the data, the case for investment will be so much stronger. So I think some of the early health AI ventures really did struggle with scale, not only because of capital, but because the path to deploying that capital into the kind of model training or application that was required in order to then go back to investors for more cash just wasn't there.
[38:36] Ravi Gurumurthy:Make it the best place to do the—do the thing, and then the money might flow. Make Britain the best place to have data-driven innovation.
[38:43] Mallory Durran:That's a great place to end. Um, Mallory, Britain is the best place to do data-driven innovation. Thank you very much.
[38:49]Thank you, and good luck to Andy and his team.
[38:53] Ravi Gurumurthy:If you enjoyed this episode, please do like, share, and subscribe wherever you get your podcast. We've got more coming. For those of you who don't know us, Nesta is a research and innovation foundation based in the UK. We design, test, and scale solutions to the big challenges of our time. We're funded by a charity endowment and are politically neutral. For more, do visit www.nesta.org.uk.

[00:00] Mallory Durran:An individual researcher might have to go through 22 different pathways just to have access to the information they need to do the cohort analysis to even get their clinical trial off the ground. And you can imagine the kinds of delays that that represents.
[00:14] Ravi Gurumurthy:Hi, I'm Ravi Gurumurthy. I'm the CEO of Nesta and welcome to this miniseries of Policy Fix where we'll be focusing on what Andy Burnham should do in his first 100 days. We'll be covering lots of topics from energy to today's issue, which is digital public infrastructure. Um, and with me on this subject is Mallory Durran, who is the Chief Technology Officer here at Nesta, and she's a computational neuroscientist by background. She spent 10 years uh trying to make government better in terms of digital services and AI. And um, it's a great pleasure to have her here today. Mallory, welcome.
[00:46]Um, before we get into the conversation, do you want to say a little bit about how you got into this question of digital sovereignty and—and your background in the issue?
[00:54] Mallory Durran:Yeah. Um, maybe haphazardly, actually. Um, but a—a long-standing set of, you know, challenges or things I cared about. So as a very sort of early-career researcher technician, um, I was doing one of those jobs that, um, you know, loads of people do, kind of menial tasks. And, um, you know, in the early 2000s, there was an app for everything, and—and we were sort of Instagram for fitness. Um, and I was manually filtering at the time inappropriate content and photographs.
[01:23]Um, and you know, with the technical background that I had as a—a lover of maths and statistics, I thought I could pick up machine learning and then I wouldn't have to look at these photos myself. Um, and so created a machine learning algorithm to do uh that sort of sorting for me. And I'm sure uh inquiring minds can imagine the kinds of photos you might inappropriately get on a Instagram for fitness app in the early 2000s.
[01:47]Um, but really that was my first introduction into: what are the rules for participating? Um, how do we create environments that allow people to express themselves, uh, but not force content on—on people that really genuinely nobody wants or needs to see? Um, and sort of that machine learning pathway brought me into a larger world of the infrastructure that then enables that.
[02:10]Um, so it's awesome if you, you know, have the—the best and brightest ideas for how a computational model of progression of Parkinson's disease, which is something that I—I worked on in my—my early career, might work. But if you don't have the—the data infrastructure, the compute, um, and the environment in which to carry that out, your—your ideas just won't come to fruition. You can spend, you know, your entire life working it out with a pen and paper, um, or pencil and—and paper.
[02:38]So, um, when I joined the—the UK government, I had these big dreams about how the computational modeling um that I had been doing in my research life might be applied to things like uh winter in the NHS, understanding patient flows, allowing for operational environments to be more prepared or respond uh or react to influxes. Um, but when I arrived, I quickly realized that, you know, the—the digital infrastructure, um, the—the cloud environments, the available compute, the join-up of data, even the standards of what an individual variable might mean across hospital environments was just not in play.
[03:14]Um, and so, you know, everybody talks about the—the benefit or value that data, technology, now AI could bring to our public services, to systems. But none of that is possible unless you set the standards, um, and the—the infrastructure in a way that actually lends or leads toward that. So my, you know, my interest in digital public infrastructure, and indeed the standards and guidelines and principles through which we participate in digital environments, is a—a long-standing journey that's had a—a few wiggles along the way, but yeah, really comes from first-line experience.
[03:48] Ravi Gurumurthy:Okay. So, necessity is the mother of invention. [laughter] And um, we're going to talk about um digital public infrastructure, digital sovereignty quite a lot. But I want to start, because this is a podcast about what Andy Burnham should do, with Andy Burnham, where um in his first speech where he talked about launching Number 10 North, he—he—he didn't actually mention AI at all, but he did talk a lot about public control, and that's been a—a theme already. And that language is quite similar or related to the question of AI sovereignty or digital sovereignty—putting people and the country in control.
[04:26]Um, but when—when you think about sovereignty, you've got to think about the whole sort of layer from energy, compute models, um critical minerals. How do you—how do you think about sovereignty? What is—what is—what is important to own, what is important to, and what is actually tractable? Um, and I think there's a whole spectrum, isn't it, between having it locked down yourself versus being dependent on people you trust versus being able to accept adversaries owning it. Um, so give us a sense of what's important, also what's tractable.
[05:00] Mallory Durran:Yeah, absolutely. Um, I—I think that the—the sort of fabric of that kind of infrastructure lives in Manchester. I think Number 10 North doesn't have to state that it will be, um, you know, technology-enabled or—or AI or have principles of sovereignty, because, you know, we—we owe a lot of what we now think of as AI, computing power, the principles of computing, um, to, you know, bright minds that started and coalesced um in—in Manchester through history. So, um, I think that's a—an amazing base start.
[05:28]But you're right, when we talk about digital sovereignty, there is a tech stack, and that tech stack builds from the cables and wires that move energy across the country. It is how that energy is generated. It moves through data centers into cloud systems, through to data models or the representation of information um in—in binary, in—in—in data models, um through to the—the patterns and protocols that allow application layers—so, you know, the app that you interact with on your phone to be able to call or access that information. And then there's that access layer: so who gets to access what, for which reasons.
[06:08]And all of those are important things to debate in thinking about where you might play in creating sovereignty within the system. Now, I'm not going to pretend that I am an infrastructure engineer and have the—the sort of last line on whether, you know, Britain's grid is sufficient for the kind of demand that might be placed by data centers. Um, but, you know, just in our day-to-day work, even here at Nesta, we do lots of thinking about and trying to find ways around use of AI because of the way that we govern, um, you know, special category or personal data. And if we want to use the most advanced AI models, there actually isn't enough, you know, available compute or power to have data residency remain in the United Kingdom.
[06:58]So those sort of bottom layer of the tech stack and the physical infrastructure that is supporting um is a real challenge and something that I think we need to bring communities into thinking about um how we get that sort of data center um and—and core infrastructure right. A layer up from there, when we start to talk about the cloud, um I think it's really important what our information passes through and—and who holds and governs that along the way. But the nature of the way the technology market has emerged means that there is not a credible UK-based or even um really a competitor to AWS, to Microsoft Azure; there isn't a credible European alternative to that.
[07:44]Um, and so when we think about where to play, finding the component pieces where we can take back control without having unrealistic expectations of the market fundamentally changing and being able to catch up, both in um scale of the offering, the cost at which that can be provided if you're sort of new versus incumbent—um, I think that the—the real opportunity for the UK really lives the next layer up. So in our—our data models, in setting really good standards for how, you know, our public sector data in particular um is held and governed so that smaller players can start to build application layers um and potentially eventually move to compete with some of the other cloud or service providers. But if we get our data layer and standards right, I think we're, you know, much better able to harness um that capacity for economic growth in the technology sector.
[08:41] Ravi Gurumurthy:So—so you're saying that data layer is really, really important. We should probably think about the other layers in the stack. Some of them are more tractable than others. Um, and there are probably some where you think they're important, but they're not actually easy to do very much about. Is that fair?
[08:57] Mallory Durran:Yeah. So let's use an example. Um, so open banking um was a—a program of work taken by the UK financial sector that looked at setting standards, and that means everything down to: how do we represent particular banking variables, how do we hold dates and names, uh balances, right? Everything from the—the—the standard table to um how that in—what the security layer looks like on top of that infrastructure, so that um an—another application could come and draw that information accurately, as well as what security standards needed to exist for an organization like a new fintech startup to be allowed to access that information.
[09:42]So all of these standards were negotiated with the sector and implemented. It's not true that banks aren't allowed to buy cloud services or indeed data infrastructure from American or European or who—whoever they should choose. But in providing those infrastructure services, that company needs to meet the standards that have been set by the UK banking and financial sector. So they sorted out that data layer, as well as the security and access layer.
[10:17]And now if you can meet those standards, you get to play in the space. So we saw an absolutely massive increase in the fintech sector on the back of the sort of 2018 coming in of—of those standards. Um, so you've got, you know, the Monzo and the Revoluts of the world, and suddenly, you know, as a participant in banking, I can look at all accounts across all banks in a single application, and I get to make a choice as a customer about which user interface suits me. And that's only possible because they got the—the data and the access layers right. And that has fueled, you know, business growth and economic benefit, but also benefits for me as a—a customer of banking.
[10:57]So, um, you know, Lloyds or Barclays or indeed Monzo can choose whatever infrastructure provider they would like, but those people have to come in and can only sell services that match or meet the—the credentialing and the standards that have been set by the sector.
[11:12] Ravi Gurumurthy:Great. And so, and you're saying that like open banking, though, is a bit of an exception, because if you take um whether it's energy or health, most other sectors, they haven't created those common standards that allow um for a variety of different players to come in, data to be shared in a trusted way, citizens to have consent over that.
[11:32] Mallory Durran:Yeah, exactly. And maybe ironically, um, the—the problem isn't a technical one; it's a human one.
[11:39] Ravi Gurumurthy:Mhm.
[11:40] Mallory Durran:It is not complicated to decide on a tech stack, but getting a bunch of different organizations with different incentives, um, with lots of players with creative ideas into a single place to agree and then invest in implementation um is really a human problem rather than a technology one.
[11:58] Ravi Gurumurthy:And in terms of the leadership for this, you can place the emphasis on governments to do this, um, but there are lots of examples of technical standards being agreed by industry, like Bluetooth...
[12:08] Mallory Durran:Yeah.
[12:08] Ravi Gurumurthy:...where, um, you know, it's not a sort of state creation, in fact, it's, you know, across multiple geographies. To what extent do you think that the—the digital public infrastructure that we need to create in this country is about government leading, or can it be led more bottom-up? Is there enough um incentive to collaborate because people don't want to necessarily compete on certain things, or does this require actually a bit of government uh direction?
[12:31] Mallory Durran:I think it definitely requires government direction, but I think you're right that there are a good examples of where standards have been, you know, met or decided upon without there being a government intervention. You just need a—a critical mass. Um, and I think consumers or citizens have a huge voice in that. Um, so, you know, a really recent example is all of the different cables that we have had for our phones and small electronic devices over time. I think you're now seeing a consolidation around USB-C.
[13:05] Ravi Gurumurthy:Even Apple had to do that.
[13:06] Mallory Durran:E—even Apple had to move. Um, and that probably, I don't know, reduced the number of sales of cables they might have had because of other providers, but the people wanted it, and the people have power, and the people have a voice. Um, and so I think there's a real moment here for citizens, groups of citizens, groups of users to be voicing where and how they want to see things. And I think—but I think there is—there is potential for particular sectors to come together to—to sort of start the charge. But I think even in open banking, um, you know, there is potential within the financial sector that's not yet been reached. It's—it's...
[13:46] Ravi Gurumurthy:I mean, a lot of people say that open banking was great, but the full benefits were not actually released. We sort of—it was a sort of half-victory, if you like.
[13:53] Mallory Durran:Yeah. And I mean, a—a victory is a victory. Um, and you know, as someone who uses banking applications, I sure am glad for it, but I think there's—there's wider benefit. And I think, um, you know, especially some of the smaller players in the open banking space—so people who might be more like connectors of information, building those mid-layer applications rather than the big banks themselves—will say that they've had an ebb and flow of the quality of imp—implementation of different sort of waves of standards or adaptation, largely driven by how clear and forceful government's position on the implementation of those standards were, and that legislation played a massive, important role in actually getting things over the line.
[14:38]Um, so I think we can't rely on the goodwill of an individual sector alone; often government does need to intervene. Um, and I think there also has to be someone really making the case for what the benefit could look like in order to get that coalition of the willing going in—in the first place. So really clear articulation of benefit within a particular sector to—to move things forward.
[15:03] Ravi Gurumurthy:So why don't you paint a picture of what that could look like? Let me—take imagine if you did open health or open energy. What—what could open health look like, and—and what—yeah, what's the size of the benefit?
[15:13] Mallory Durran:Yeah. So I think health is a—a no-brainer for the UK. Um, we've got a brilliant life sciences sector. Um, and I mean, as a researcher coming to the UK to, and you know, institutions that are older than the country that I—I came from, um, I was so wowed by, you know, being in these classrooms where you know that all of the great inventions of life sciences and health innovation came out of—like, we've—we've got this, you know, globally um unparalleled strength in the—in life sciences.
[15:47]Um, but for a clinical researcher trying to, um, you know, bring forward a new mode of delivering a—a cancer drug, or, um, you know, where data science has, you know, sparked up, "Oh, we could use this sort of GLP-1 instead of for heart conditions, we're seeing weight loss reduction," like what does it look like to measure across in order to get access to the data at the scale required? There are something—HDR UK, the Health Data Research UK organization, did a mapping of all the potential access points. Um, an individual researcher might have to go through 22 different pathways just to have access to the information they need to do the cohort analysis to even get their clinical trial off the ground. And you can imagine the kinds of delays that that represents.
[16:36]But if we got that layer, the—the data model right, if it was, you know, a consistent approach to bringing health data together for research across individual controllers, trusts, GPs, um, the UK would have an unparalleled quality um and scale of connected, high-quality information which could expedite clinical trials. You know, the UK would become the best place uh for drug companies to be doing every phase of trial, which would give, you know, UK citizens access to new treatment and care earlier.
[17:15]Um, similarly, kind of based on that what we saw in open banking, if you've got a way to govern who can then access what is genuinely sensitive information that should be, you know, protected and not given out willy-nilly, that in—in open banking you had, um, you know, Financial Conduct Authority setting standards for um who can access this information and why. Yes, you have to be a certified financial institution to access, but once you meet those security standards, you're allowed to innovate on top of it.
[17:45]And in the—in the health sector, that could mean a huge growth in innovation for personalized healthcare. Um, you know, a—a real health tech sector that has access to information. Whereas an individual citizen, I could say, "Yeah, you know what? I have a rare disease in my family, and I want to have individualized recommendations for me on the basis of what I should be looking for or caring about." Um, and I could choose, you know, within a—a growing health tech sector what was right for me, um, and—and really have agency over um how and who and where my data is accessed and—and used. So I think there's a huge innovation opportunity that sits within it, as well as, you know, research and um health benefits that could be seen in the public sector.
[18:32] Ravi Gurumurthy:So it feels like to me at the moment, though, you've got um huge potential from linking data together. When you do it badly, um, you either create a—a platform where a single company has uh too much power and lock-in, uh, and that has two problems: one, it often creates a consumer backlash, um, you know, you've—you've got a problem at the moment with some GPs deciding that they don't want to play ball with Palantir, for example—you get that sort of backlash; but you also then get the kind of harvesting of benefit to a single player who can lock out others, rather than building quite a rich ecosystem.
[19:06]Um, I mean, India is in—really, India is the sort of the—the obvious example where they started with identity, moved to payments, um, and now they've got a kind of very rich ecosystem off—off the back of it. One example I always like is the example of transportation where um in India, Uber used to drive—used to charge a daily driver commission, um, and then because they created this digital public infrastructure, different apps can connect with drivers um pretty freely, and they weren't charging an annual—a daily driver commission, and so Uber were outcompeted essentially; they had to drop that down. And what they've done is create a world where um gig workers are keeping more of their money, uh, and consumers are also keeping more of the—the surplus. Um, and it's not a sort of a big-state thing; it's a—it's a very minimal public infrastructure with lots and lots of different, um, you know, competitive apps, private and public, coming off the back of it.
[20:06]Is that—I guess, is—is that—is that vision true for other things? Could we do that in parking? Could—one of my favorite ones—um, or, you know, other things that are very consumer sort of—where the consumer benefit is going to be really clear?
[20:18] Mallory Durran:Yeah, absolutely. I think when you get digital public infrastructure right, you fundamentally reshape the market. Um, and I think, you know, all kudos to the brilliant technical layer built by the, you know, the India Stack team. What allowed for that was setting rules, right? Any application, whether it was Uber or the homegrown ones, had to meet the same set of information governance and sharing rules, which is why, you know, um drivers were able to port, you know, not just their credentials of like license and registration across different applications, but also their um uh rider stars. So if I as a driver want to move, I don't lose the credibility that I have built; I'm able to move that across.
[21:04]And that required India to say, "Yep, you can play in our market, but these things have to be true. It has to be true that you can port that information in different ways." And so what they did there was, yes, a very clever technical layer, but actually it's about being unequivocal about the rules for participation: if you want to play in our market, then you've got to do it our way. And that created a totally different kind of competition that had benefits for a wider group of people, or um a different power balance in the structure. And I—I think that that's true wherever digital public infrastructure is done well.
[21:41] Ravi Gurumurthy:Just a word on language: "infrastructure" is an interesting term, because is this essentially regulation where you're defining a set of standards in a marketplace, or is it actually infrastructure in the way we think of it, like a road, where there is a sort of something that—that is being built? I think it's relevant because um sometimes this can make—this can sound almost super technically challenging, and then we start getting pessimistic about our ability to do anything quickly or—or well.
[22:11] Mallory Durran:Roads are an interesting one, right? Because yes, they're physical infrastructure, but there are also standards. Um, you know, if you're going to build a new road, you're not allowed to decide just to make it, you know, 12 centimeters narrower than the other because that's convenient, right? There—there are acts and governance for what we mean by something being safe. Um, and sure, there are lots of examples of streets that um are higgledy-piggledy in all sorts of ways because they were built before, but we're also dealing with that in digital infrastructure, right? Legacy systems that don't meet the um the potential of what is possible now.
[22:48]So I think in digital infrastructure, it's a similar mix, right? There are patterns, protocols, architectures, blueprints, diagrams that you can set out for what it just basically looks like to do it well, but there still has to be some kind of agreement for: what does that look like, and why, in order for it to, you know, work across systems.
[23:13]So, um, yes, it's legislation. You know, as we said, open banking moved as quickly as it did because there was a really strong push from government, and legislation made it clear what was expected by when and why, and then the implementation of the technology and the actual designs kind of sat separate.
[23:36] Ravi Gurumurthy:Mhm. Mhm.
[23:36] Mallory Durran:[clears throat] From that.
[23:37] Ravi Gurumurthy:And just some—give me some more use cases. The reason I like parking is just that I feel like we can do it quite quickly. It's one of those things that annoys people when they have to sort of go to a different parking app every time. Um, um, I think there's—there's almost a: let's educate um the public about the value of things like public infrastructure that could be done quite quickly. Are there any—what are your favorite sort of smaller use cases, if you like, that could be done quickly?
[24:01] Mallory Durran:Yeah. I mean, parking is an interesting one because there are some parking applications that do have an open and shared set of standards. You know, some—some local authorities have deals with—with sort of multiple providers, and therefore you go to a single interface and you can enter your parking location code in and it—and it works, and some that have not jumped on board. So that's a case where there was a coalition of the willing, and some of that comes together in a really neat way that means you don't have to go outside of a single application, and some of it doesn't. And it's because probably there is no governing body that says, "You must meet this now." And so I think that's a good example of where a coalition of the willing can get so far. The infrastructure exists, um, but yeah, it hasn't sort of gone all—all the way.
[24:46]Um, I've got really high hopes for a project that the Open Data Institute is—is leading um to look at pooling volunteering opportunities and the requirements for those volunteers uh across the country. Um, I don't know if you've ever tried to, you know, find something in your local area, you thought, "You know what? For the next few months I've got a bit of extra time on my hands; I'd really like to, you know, go and read in a—in a community center or a hospital." It's actually really tricky to—to find those. Um, and so the Open Data Institute are looking at how they create uh a sort of open layer for how you would pull in all of those volunteer opportunities.
[25:21]And I think on top of that, there's a—a really interesting opportunity for: what does it look like to also be able to open up um the—the credentials required? So if I'm going to go into a care home, for example, I need to have a DBS check. Um, and so it's not just about pooling the volunteer opportunities, but it's actually about saying, "What—what are the requirements for meeting this, and can I create an automated pathway to pull in: does this person have an up-to-date DBS check?" And I think, you know, we strengthen communities, we prove public—digital public infrastructure, um, and we make it easier for people to take a credential that belongs to them and port it across, you know, public-interest use cases.
[25:59] Ravi Gurumurthy:Credentialing is an interesting use case in its own right, isn't it? Could go much broader than that.
[26:04] Mallory Durran:Yeah. And I think there are some really big proposals for things like a digital ID, for example, um, but some really good use cases that fall below the layer of, "We need to legislate for this standard." So, GOV.UK One Login—I don't know if you have updated your passport, driver's license...
[26:22] Ravi Gurumurthy:Um, you know, I did actually. I was on the sofa thinking, "I've got to do my passports," and my wife was going, "Are you coming to bed?" And I went, "I'll be a while; I've got to do this for the kids. We're going on holiday." 3 minutes later I went up to bed and she went—and she was actually annoyed that I'd given up. And I said, "No, I've actually done it all." And it was—that—it was amazing.
[26:41] Mallory Durran:Yeah, my—my most recent favorite example was I had updated my driver's license and then went to apply for my first British passport, um, and was able to use the same photograph because that was all stored within an account, and that didn't require the implementation of a unique ident—identifier across all systems. It's a brilliant technical layer, bringing things together for citizens.
[27:00] Ravi Gurumurthy:In a sense, that's where I was going to go next, which is: we've obviously had the latest attempt at um at digital ID. I'm actually not even sure where we've got to, if I'm completely honest, on—on digital ID, but to what extent do you need to do digital ID as the layer before you can then do the others, as India did actually, or um is there a different route through this?
[27:21] Mallory Durran:Yeah, I'm sure that there are really brilliant people super in the detail of this that might disagree with me. Um, but I don't think you need to do digital ID first, and I think GOV.UK One Login is a really good example of that. And the reason I think that is that even if you create a unique identifier tomorrow for every citizen, there's a significant amount of fixing within existing systems that needs to be done in order to carry out implementation of that sort of unique ID system.
[27:52]Um, you know, one example of that would be, uh, because I had a student visa and then moved to a working visa, I actually am two people in the NHS, and I once showed up to a hospital um, having, you know, significantly cut my hand, bleeding everywhere, and they're like, "Well, which of these two people are you? Which of these addresses is correct?" And we have NHS numbers. It just so happened that I ended up with a—a dual account. So even within systems that have an identifier code, there is work to be done to improve, consolidate, update infrastructure to allow for what is the promise of that sort of digital ID system. So having an identifier does not automatically mean you have joined-up systems with interoperable technology.
[28:36]And so I think we get into this unnecessary chicken and egg where we, you know, don't update our infrastructure to be able to improve join-up because one day we might have a digital ID. But even if we decided and went hell for leather on digital ID tomorrow, there would be base work to get that right as well. And so I—I think it's a false dichotomy, yeah, and that there is more and better that we could be doing—One Login as an example of that, um, like the—the one GOV.UK stuff, um, that would improve citizens' experiences and interactions with applications before you get to the point of implementation that, if we were to decide to create a unique digital ID, um, that would be easier to implement because we had done that underpinning work.
[29:21]So I think we're unnecessarily tying ourselves in knots over the real questions from citizens about why and how you would do a digital ID in a way that was, you know, fair and reasonable.
[29:32] Ravi Gurumurthy:Yeah. And also one challenge we've got in doing in this country is that with India, um, there was a real, huge value in creating their digital payments layer and—and wallet, but, you know, we've kind of solved some of those problems in a different way, possibly suboptimally, but it's quite hard to sort of um have the same transformation here because it's not a greenfield site in quite the same way. Is that—is that fair? Is that an impediment?
[29:57] Mallory Durran:Totally. Um, yeah, I mean, I—I waxed lyrical there about the brilliant life sciences, you know, the history of this country and, you know, the advantage that that brings us. Um, there is a flip side to that, which is if you've got a thing that sort of works some of the time, that can be harder to build upon, or especially where you have 15 different versions of a thing that have sprung up quite organically, actually how you consolidate to a single thing. So yeah, in some ways I do think building from legacy systems is harder than starting from carte blanche. Um, and it really does come back to vision and leadership and ability to create that coalition. So the—the problem isn't actually a technical one; it's a human one.
[30:43] Ravi Gurumurthy:Okay. So um let's pretend that you're now um Andy Burnham. Sorry, Andy. Um, what would you—what would you actually do? What would be, you know, the three to five things you would prioritize now?
[30:58] Mallory Durran:Yeah. So I mean, we've—we've talked about digital sovereignty, and I think making really clear the government position on that. I think lots of private-sector organizations, and in particular infrastructure companies, are trying to think about how they respond to a call for sovereignty. If I'm Andy Burnham and his team on day one, I am going to come out of the gate and say, "The United Kingdom is moving to a standards-based system for procurement, for data models, and we will, you know, appoint the right technical people to make those decisions sector-over-sector in the way that we have done with open banking. Um, but this is how you will play in the UK digital sphere, mhm, unequivocally across sectors forever."
[31:40] Ravi Gurumurthy:So standards-based procurement system, that just means if you're going to play in X market, these are the kind of common standards that will always apply for a long period of time.
[31:52] Mallory Durran:Exactly. And the—the benefits of that are preventing vendor lock-in. So if there's a transparent architecture from your sort of base data model through storage, through APIs, then A, if a provider isn't doing a good job, then someone else can come in and follow that blueprint and replace them. Um, so it raises the quality of standards.
[32:13]It's also true that, you know, in—in most places in the public sector you don't have people who are experienced engineers or architects with the sort of final procurement decision on the systems or services, sometimes because it's not sensible to have, you know, that kind of technical knowledge in every team everywhere that might need a digital service. Um, but it means that, you know, companies trying to sell things have loads of bells and whistles on, and the person trying to buy doesn't know which ones are going to be a meaningful value-add versus, you know, what extras aren't necessary.
[32:46]And so being able to set those standards both means the quality of the service should be better, you have the ability to replace a service more easily and transparently if something better comes around (either in a value-for-money perspective or in a provision perspective), protects you against that lock-in, and will lead to better value for money.
[33:05] Ravi Gurumurthy:And does that require a standards body um either across multiple sectors or in each sector?
[33:10] Mallory Durran:Definitely. Um, I think it's probably sector-specific, but you probably do want a—a single governing body to make sure that the linkages across things make sense. You know, there are common variables between health and welfare and tax systems. Um, and this is boring stuff like how you even...
[33:34] Ravi Gurumurthy:Yeah.
[33:34] Mallory Durran:...define an address or something.
[33:36] Ravi Gurumurthy:Yeah. I mean, like, have you ever, you know, opened a new spreadsheet in Excel and had the dates column turn into exclamation marks? We're basically trying to prevent that at a really base level, right? So that's the first—that's the big thing.
[33:47] Mallory Durran:That's the big thing.
[33:48] Ravi Gurumurthy:What else?
[33:48] Mallory Durran:Um, so I think I—I alluded to it there, but getting the right people into those decision-making positions um and treating talent as a national emergency. I think the public sector cannot...
[34:02] Ravi Gurumurthy:You spent so much time in—in—in government, in Number 10 and in uh the Incubator for AI. You're saying talent is a national emergency.
[34:11] Mallory Durran:Talent is a national emergency. There are incredible people um that exist, but they have a really hard job. Um, it is not easy to be the lone technical voice at a table full of people who want good things to happen, but have absolutely no technical grounding um or—or knowledge. And I think if you look at, you know, procurement systems as an example, you do have to have people who understand the difference between uh federated and fabric-based data service. Um, and that isn't something that can be upskilled in a civil servant overnight; it's genuine knowledge and training and, you know, a need for exper—experience of implementation of that at scale. So you need your risk and decision-making owners to have real technical skill.
[34:59]And then we've sort of not talked specifically about AI, but having a good cadre of engineers is not the same as having frontier AI talent. And I think a—I mean, I'm an immigrant. The UK is a great place to live and to, you know, care about frontier research. I think that we have a moment and an opportunity coalescing around some of the big bets that the government is making on sovereign AI and some of the brilliant companies that have been, you know, built or seeded here, but we've got to make it easy to get the best and brightest talent into those places.
[35:39] Ravi Gurumurthy:How—how optimistic are you on those big bets on sovereign AI? To what extent are they—I mean, the kind of outside critique would be: "This is tiny and we're late."
[35:49] Mallory Durran:I think that that's true. Um, but I think it would be silly to say, "We're small and we're coming late, and therefore we shouldn't try." I think that we're going to need to catalyze private-sector investment and think about commercial models in a different way; it can't just be state investment. Um, and I think that there's a different risk appetite around consolidation of, you know, picking a player and running with it um that needs to happen so that, um, you know, the sovereign AI players don't fall over because we don't have the investment in data centers or compute, which is something we've already seen happen.
[36:24]So I don't think that the journey is easy and that it will happen tomorrow, but I think that it would be um—I think it would be an underestimation of the ambition and capability of AI engineers and researchers already here, and the ones that might be attracted especially to a nation that says, "We're here for openness, you know, open-standards AI, um, and we're here for sovereignty." And I think that there's a huge appeal there, um, but multiple systems have to come together in order to make it possible to get that talent in, to pay them well, um, and to keep them in the UK.
[36:59] Ravi Gurumurthy:If I was um—if I was Andy Burnham now hearing your pitch a bit, we would go, "Okay, great. That's all really wonky plumbing."
[37:05] Mallory Durran:Yeah.
[37:06] Ravi Gurumurthy:"Uh, what's the—what can I promise the country uh if we do all this? You know, give me a sense of in three years or five years, what might be different?"
[37:16] Mallory Durran:Yeah. Yeah. So I think health is probably the—the area that speaks most to people. Um, and if we take that example of having an amazing health tech sector on the back of this, um, I would have more choice in how I access information and support and care um for my health. I think that would matter. I think that, you know, better, faster, more efficient treatment um within our NHS and operational systems and services matters.
[37:48]But I think it's also about economic growth and opportunity. That health tech sector is about providing, you know, better access to different kinds of care and services for people, but it's also jobs, and it's also, um, you know, value creation...
[38:02] Ravi Gurumurthy:When you talk to them enough and say, "Yes, scale-up capital is a problem, but also access to data, clinical trials, is also a big constraint," and this would solve that.
[38:10] Mallory Durran:Exactly. And I—I think, um, if you—if you give access to the data, the case for investment will be so much stronger. So I think some of the early health AI ventures really did struggle with scale, not only because of capital, but because the path to deploying that capital into the kind of model training or application that was required in order to then go back to investors for more cash just wasn't there.
[38:36] Ravi Gurumurthy:Make it the best place to do the—do the thing, and then the money might flow. Make Britain the best place to have data-driven innovation.
[38:43] Mallory Durran:That's a great place to end. Um, Mallory, Britain is the best place to do data-driven innovation. Thank you very much.
[38:49]Thank you, and good luck to Andy and his team.
[38:53] Ravi Gurumurthy:If you enjoyed this episode, please do like, share, and subscribe wherever you get your podcast. We've got more coming. For those of you who don't know us, Nesta is a research and innovation foundation based in the UK. We design, test, and scale solutions to the big challenges of our time. We're funded by a charity endowment and are politically neutral. For more, do visit www.nesta.org.uk.