#133 Building a Safer Internet with Data Science

Hi, I'm Maggie, VP of Curriculum at datacamp, and I know firsthand how overwhelming it can be to keep up with the rapid changes in data and AI. But staying ahead doesn't have to be hard. That's why our curriculum team has curated the best learning content to help you and your team stay ahead, covering everything from data and AI literacy for busy executives to cutting edge generative AI for developers, and everything in between. Whether you're upskilling yourself or your team, we make learning practical, engaging and flexible. It just works. Power your data and AI transformation journey on datacamp.com.

The areas that I think is really different for me working in this sector that I haven't experienced before is then say, how do I make sure that I've got that really close tie in with our legal team and the public policy teams to be able to make sure that everything that we do is within the remit of what OFCOM is legally allowed to do.

Hello, welcome to Data Framed. I'm Adele, data evangelist and educator at DataCamp and data framed is a weekly podcast in which we explore how individuals and organizations can succeed with data. Today we're speaking to Richard Davis, Chief data officer at Ofcom. For those who are not aware, Ofcom is the UK's communication regulator. While quite different in what they do for our US audience, the US equivalent here would be the fcc. As part of Richard's role, he's overseeing the data transformation at OFCOM and looking at ways at how data can unlock further value for the regulator. So I wanted to talk to him about just that, how he sees the mandate of the Chief Data Officer in government, how data unlocks value for ofcom, how to balance between technology, culture and process initiatives, and how data science can help make the Internet a safer place in the uk. Richard also shared what Chief Data Officer should probably be doing in their first three months. So make sure to listen to that. If you enjoyed this episode, make sure to subscribe to the show. And now, onto today's episode. Richard, it's great to have you on the show.

It's great to be here.

So you are the Chief Data Officer at the UK Office of Communications within the uk, or OFCOM for short. So maybe to set the stage, walk us through ofcom's mission and more importantly, what does data mean for ofcom?

Yeah, so ofcom, as you said, the Office for Communications is the converged regulator for all communications within the uk. It has a mission to make communications work for everyone around the uk, and that covers everything from public sector broadcasts. So TV Radio, the radio waves, the spectrum, the licensing, the operation around that and then looking at networks and communication. So the resilience of the communication networks. But then tele, broadband, mobile as an economics regulator on that side. And then more recently, look at, we're setting ourselves up as the regulator for online safety. So where's data in that? Well, data's really at the heart of everything we do at ofcom. We pride ourselves in being a evidence based policy regulator. So the way that we look at this is to say how can the data teams support the regulatory decision making for policy for licensing by putting the data and information in the hands of those that are making the decisions.

Preparing for this interview is actually pretty challenging because there's so many things that we can cover in our chat. Right. One of them is the use cases that you've been unlocking at ofcom. But what I wanted to focus on in today's conversation is how you've been approaching the data strategy at Ofcom and some of the use cases, data unlocks within government. So maybe to start us off, walk us through the key components of the data strategy at ofcom.

Yeah, so we have been creating a new data strategy for Ofcom as we go forward. I'll test our mission statement with you and see what you think. So currently we're aligning to good data, good people, good decisions. How's that sound?

Sounds great.

In terms of what that means on the people side, it's really about how we set up the culture. On the good data, it's about making sure that we've got excellent data in the hands of people that need it. And then on the decision side, that's about making sure that we've got the data platforms that are able to create that self serve capability. But then from the delivery side, making sure that everyone's speaking the same language when it comes to data.

That's really great. And I love the use of the term good people, good data, good decisions. I think it's really important to create a engaging communication strategy about a organization's data strategy. And we'll discuss that a bit more detail. But let's maybe roll back slightly to the beginning. You've been at OFCOM for the past eight, nine months. I think at the time of the recording. As a chief data officer looking to build the data strategy for an organization like Ofcom. Walk us through the first two months on the job. What do you think constitutes a successful first two months and what are common pitfalls to expect?

Yeah, that's interesting. So I started in the first couple of months with these grand ambitions of what I wanted to do and how I would get this business unit set up and we'd be delivering. And as you go into a new role, you write yourself some okrs. You're like, this is what I want to do. But honestly, in the first two months I was still talking to people and getting to know the business. And this is something that I think I've learned that it takes a lot longer. Especially with a business the size of Ofcom or certainly with the remit that it has. Trying to understand all of those different areas takes a lot of work. We've been going for 20 years. The data innovation Hub, the central team was set up in 2019. They've delivered a huge number of products, products that are then being delivered into all of the different areas of the business. And a large part of it was trying to understand, well, how is the processes working, where is it working, where is it not working, what are the main key challenges that the business has that a data team can add value to and how do I enable those teams to use data more effectively. That said, I think from those conversations I was quite quick to see that certain things were missing and within the first few months it was really setting ourselves up for success with future. So key things, there wasn't a data strategy, so people didn't have a vision for the value that data could give or how they would look to use data, and how we could bring the community of data professionals together to understand and how they could grow their careers within of beyond. That is then looking at how do we make sure that we got the data dictionary, the asset registry, so there wasn't one person who knew all of the data across the organization. And another key thing for me which helped define the data strategy, we do governance really well within Ofcom, but it's sometimes done a little too well and there's a bit of duplication process to just cover the gaps and it's done in multiple teams in slightly different ways. So it wasn't a full end to end thing. And that's not me saying that things will fall down through the gaps. It's more saying that it just wasn't efficient and so people were spending more time on doing the governance than they were on the delivery. And so really keen to say, how do I make. Let's data people do the things that they can do by having a more efficient governance process.

I definitely don't envy the Chief Data Officer's role, especially going into a new organization trying to Understand the lay of the land. But maybe as a mental model for you as you approach these transformational projects and setting up a data strategy, what would be your advice for data leaders looking to get a quick win, for example, after they learn the lay of the land within the organization within the first two, three months, what do you think is a good framework for trying to understand what is a quick win within the organization?

Yes, I think this is where for me it's understanding where the pain points are. So those conversations I was having, I was talking to people across all different business areas and they start to see common areas where they're not getting the data, they're not getting the information they need in the most efficient way or in a way that enables people across the business to use information as effectively as they can. So for me, a quick win was really to come in and start to understand where some of those opportunities were to be able to understand that, I think is really about understanding some of the value drivers that you have in a business and then working through how you can record or measure those to be able to have effective prioritization.

And you mentioned part of the data strategy, really the key components of the data strategy on Ofcom is good people, good data, good decisions. Right. So let's maybe work back, let's work back from these components and tease them out one by one. Right. I want to start off with the data you mentioned as well, the importance of understanding the value drivers of data, trying to align data with organizational value. Walk us through the component of data. What does good data look like? What are the standards that you look to set within your organization as a data leader?

Yes, I think understanding the value of data is really important. So I, this is probably something within a regulator that's a bit more difficult compared to where it was before. So I used to work in a bank and you, you could look at data, the value from data, as they're saying, are we increasing the revenue? Are we looking at cost reduction? Is there an efficiency saving, is there a risk mitigation? And all of those can be translated into a pound amount suddenly working in an arm's length body. Those drivers aren't there anymore. Or not to the same extent we have cost reduction and efficiency, certainly. And we've got cash, flat budgets. So we've got to make some of those efficiencies to be able to do more. But we've then got to make more conscious decisions on how we measure the value of data. And really as part of that good data, it's about how we start to understand where we can drive some of those. So absolutely still looking at the cost and the efficiency savings. But a large part is also saying how do we enable teams to use data more effectively? So looking at where the alignment is with the data, the analytics, the decision making process, to our three year plan, to some of the operations that we have, we've consciously across the organization, made choices of things we're leaning into and things we're leaning out of. Interestingly, data helps both of those. So if there's something that a part of the organization, say for example, a Spectrum team is saying, actually we're going to lean out of that as a conscious choice. But it's still some of the BAU activity, the business as usual activity that they have to do. We have to then say, well, how can data support that? Whether it's a data automation, a pipeline, getting that engineering done, or is there a way to have a reproducible analytical pipeline that's able to get the information, the insights to people in a more efficient way effectively, so they can lean out of that piece of work that they were doing, or certainly reduce the amount of effort that it takes and then be able to have that capability to lean into some of the key business areas that they'll be looking at as part of their three year plan. And I think trying to measure those as a value is really key to being able to help prioritize the activity. Beyond that, I think we can talk about the advantages and the future technology that data's giving, but as the technology is changing, that means that we can have more capability that we can give to people. I think some of this goes back to what I'd almost term as resilience. Whereas something would have needed a team of people and with years of coding experience, with some of the no code, low code environments, we're able to now offer tools that will enable people to move faster and potentially potentially improve the resilience of the organization as a whole. And that's a key part of some of the value drivers that we're looking at as well.

That's awesome. And there's a lot of things that I would like to tease out from your answer here. You mentioned kind of the data engineering part, building the data pipeline, making sure that data is reproducible. What are some of the challenges associated with that when revamping or improving the data infrastructure in a government agency such as Ofcom?

So I think I mentioned before, OFCOM is now entering its 20th year, so it's our birthday this year. Big anniversary. And it was set up from a number of different regulated areas. And we're empowered by acts of parliament and those come to us to be able to regulate different sectors for communications. As that's happened, these things are coming in, they're being set up and they're set up in a siloed way. Those will all have their own systems, their own stores of data, their own methodologies for how they look to produce in insights and use those in decisions. And so a key part of that data engineering plank that we're looking at is say, how do we actually start to use the platforms more effectively to create a single data store where we have that information? We're able to then look across the full gamut of data that we have available, where it's being collected in a way that is able to be used in a manner so we can start to say, where's the converged consumer? So how are people using telecoms, broadband, mobile? How does that impact their online safety? And is there methods where we can start to understand some of that convergence of the consumer across all of those different areas of the regulations that we have? But to facilitate that is part of the, I mentioned the data excellence banner of our data strategy. We need to have really strong metadata that's able to support that. And this is where we're starting to build up some of that information, so we can say where's the data? And being able to just have a really good searchable store of data, understanding the lineage that the data has been taken through, so what transformations have been made on data, who's the owners at different stages of that data. If we have raw data that we've ingested for a very specific purpose from our legal remit, we can only use that data for that purpose. We need to maintain the metadata as that data is aggregated and goes through different stages of reporting, so that the final reports and the insights maintain that level of security, confidentiality and purpose throughout, so we can't then misuse data. And that's really key to how we're setting ourselves up. So we're maintaining that full governance, but also enabling people to move faster by using the metadata around it to be able to support people.

And you mentioned as well in your answer, like the importance of understanding the value drivers of data and aligning data with the business value. I think this is probably one of the biggest pitfalls data teams face when getting started with data. Walk us through maybe why that pitfall is so common. Because a lot of the times you see data teams Especially early in their journey, kind of approach, really complex use cases that don't necessarily unlock business value, but are just really cool to have. And what do you think as a data leader are mental models data leaders can adopt to avoid this pitfall?

Yeah, I've seen this throughout my history and I've probably been guilty of it myself. You get excited about a new technology and you're like, I want to see what's the potential of this. And before you know it, machine learning is very cool. Yeah, the number of times when I was in the bank, I don't see it so much now, but it's certainly there. Someone go to a conference, a senior leader, and they're like, ah, I want this machine learning. Like, okay, what do you want to do with a machine learning state? Don't argue with me, I want machine learning.

Exactly.

Artificial intelligence that's needed. And we're probably going to see this now with people saying, I want a large language model, give me chatgpt in my business. And you're like, okay, what do you actually want? What's the business problem? I've seen this with some of the data teams and people get into a, this is something that I'm interested in. People go and they spend years doing research PhDs, they're constantly learning. And I want to encourage that because it's great because the more that people go and they learn something new that might be applicable to what we might need in the future. So having that capability and the curiosity should, should never be discouraged. At the same time, part of the process of being like that data leader is to say how do we have strong business partnering? And this is part of the process that we're looking to set up is to say how do we go into businesses, understand their problems. So we're a cross functional team. How do we treat those regulatory teams, the people who are making policy decisions, enforcement decisions, the operational decisions. How do we understand where their pain points are or really understand what they're looking to do and then suggest methods. The technology and the data can be used to do that. And sometimes we could be using some of the really cool advanced techniques, other times we just have to rein it in. But say, trying to keep that balance, we're saying I want to maintain the curiosity, keep people alive and interested in the subject and potentially then finding routes to be able to deliver on those, but in a way that adds value to the business.

I completely agree and I think it's a lot of times as well, a lot of the value drivers that you see within data within an organization come from descriptive analytics and diagnostic analytics and not necessarily a lot from complex machine learning models. Right. And being able to align that and create that partnership between the business functions as well as the data functions is extremely important to keep that alignment of value. But I do agree that sometimes machine learning is very cool and stuff like large language models and ChatGPT are very interesting and they tend to attract especially technically minded folks because of how interesting of a problem set it creates. Right. For you to solve. So let's talk about the data culture component of your data strategy. Right. You mentioned here we talked about good data. Let's talk about good people and good decisions.

Right.

And what does that mean? So you mentioned here the cultural change aspect of Ofcom's data strategy. This is probably the most important and most challenging aspect of a data strategy. Walk us through why it's so important and what does cultural change look like at ofcom?

Absolutely. Culture, I think, is quite often one of those, the hardest pieces to change. It's more ingrained. Luckily, I think Ofcom's in a really good position. One of my take homes you asked about my first two months, the first week I came home and I'm like, wow, there's some really, really smart people that are here attracted, a great range of technical experts and people who are curious. So part of that cultural piece is a lot easier to fix in terms of saying, how do we move into more data driven area? So our strategy for data looks in, I say, two main dimensions. So the first is saying from that kind of cross area, how do we support data literacy? So this is somewhere where we want to try to help people from across all of the business, everyone within Ofcom, understanding what data is, what the methodologies look like, what the tools are that they could be using. So if they're hearing about, I don't know, some kind of data lake, a data lake house, wherever it would be, how do they start to know what these terms are and be able to have that conversation with more technical people? Beyond that, we're also like, these are people that are going to be given insights and they're going to have statistics and we want to be able to get them questioning those statistics and understanding where they come from. So it's kind of like how to lie with statistics in reverse. So how do you spot that the people that might have given you information that are trying to hide something through the gaps and trying to work through that within that data literacy bracket? Interestingly, we just had our chief legal counsel and a Number of his team of lawyers going through this. And that's been absolutely brilliant because we work really closely with the lawyers in terms of what's within our remit, what data are we allowed to use? So having them being trained on how data could be used has been really important for us. And certainly not a traditional area that you'd usually say train the lawyers on how to use data. But it's been great to see that. And then beyond that, you got the in depth technical experts. So this is where we have our data profession. And currently we've got an organization which has been set up with people who have a line manager who might not be a data expert. So they're having their career development conversations from someone who's not not a data expert. And so we're trying to say we've got this big wide community of data people. How do we make sure that every single person in that community is supported in their future career development? So trying to understand what their specific skill set is. Are they a data scientist, are they a data engineer, are they a machine learning expert? And trying to work through at different levels what skills and capabilities that they need and then linking where there might be gaps in their skill set to specific learning courses, whether that's an internal learning, a web design learning or potentially professional certification. And helping people to build their career so they can reach their full potential at the positions they're in, or as they're looking aspirationally to move to another area of data delivery, or if they're looking to progress their career, how can they start to build up those skills and apply them within ofcom? So that's really key to what we're doing.

I want to tease out a few things as well you mentioned. I want to focus on this career development side of things for data scientists and data practitioners. We will talk about the data literacy side of things shortly. But data science I think has a problem, which is that career development is not necessarily as codified within data science as you see in other technical professions like software engineers. If someone joins a data scientist position right now, at most companies, unless you're a Google or a Meta or a Amazon, there's no data scientist, level one, level two, level three, staff data scientists, et cetera. It's really dependent on what the value that you bring in as a data professional and the organization needs to, on an ad hoc basis, adapt that. I wonder how you view that perspective. What do you think are ways organizations can codify the career development and the laddering of a data scientist within Their teams.

Yes. I've almost got a challenge that I think organizations could. If they haven't, it's either a lack of a conscious decision to have done that. And I think this is very key to what we're doing, is to have that laddering approach. So you can start to say, is someone entry level. We have a talent development pipeline that we're looking at, say how we gain people in where they're grads and we have interns that are coming in and we're looking at a diverse network of people to be able to support that. So whether that's women in Daisy, the bright network for some of the internships to be able to support the diversity that we have, but as we grow through that, that's kind of like entry level. So they were saying for the apprenticeship, how do we get them onto an apprenticeship program? So they actually then have that full career development. We're then looking at even internally taking people who might not have data science skills, for example, or data engineering and working with them to do internal apprenticeships there at higher levels, they can build their career journey and cross functionally trans. But then even as people roll off from the grad program, we got the associates, senior associates, the principals, each of those stages, we're looking to say how do we make sure that they've got some of the skills that we need? That's both a technical skills, but also then looking at some of the wider leadership and capability skills that we would expect people to have at different stages of that. So the business partnering, the value assessment, all of those kind of things that you start to have, whereas you do more senior work as well as more in depth capability build that we have. So yeah, we do have the laddering and we're starting to make that work by saying what skills we're expecting to see from people at different parts. One of the key distinctions that we're also looking to set up is the difference between what's a technical expert and a people manager. And I think sometimes that's overlooked because you kind of say, oh, if you progress your career then you've got to now become a people manager and not do any of the technical stuff that you got into the job because you loved. So how do we start to enable people to grow in their careers from a technical leadership point of view and a thought leadership within that space and really encourage them to do that, but then enabling people who are more of the people people and being able to develop and coach team members to be able to do that effectively. And I Think they're different skills that people might have different capabilities on. Some you can teach and some more of a desire to be able to do that.

I love that and I love how you've approached lettering as well from a people manager perspective versus a distinguished data scientist creating a technical track, but also a people management track track. Let's talk now about data literacy and data culture in more general in a lot of ways. You mentioned that the data culture at OFCOM is powered by data literacy. You want to be able to scale data literacy within the organization. What does data literacy at OFCOM look like in context of day to day operations?

There's a number of different areas we can look at for that. So part of it is we're an evidence based regulator. So for us to be able to make decisions, to make an enforcement action, to be able to have effective supervisory relationships, we need to be able to say how do those people able to assess the information that's provided to them within that context. Also about if we're standing up in court and making a decision, are we able to actually evidence the quality of that data, to be able to show the provenance of it, the lineage, what processes it's gone through? So we need that understanding of how data works flowing throughout the entire OFCOM and what we're doing. So that's really key for the work that's done so far. But now as the tools are progressing, as the methodologies are changing, it say how do we upskill people to continue to learn and continue to develop to understand those future capabilities?

And you mentioned here at the beginning, you mentioned in one of your answers on the importance of resilience. Right. And how no code tools and a lot of different new tooling innovation necessitates the need for resilience within the organization around using these tools and unlocks a lot of the value drivers. Walk us through some of these no code tools that you've employed at OFCOM and how it can connects to the data literacy conversation.

Yes, I think the simplest ones will be things like power, bi tableau. Some of the dashboarding tools that you'll see there's obviously technical requirements that are needed for those. I think actually some of the biggest requirements on a dashboard is how to tell a story with data and how to effectively display it. And this kind of takes me into Dave McCandle's kind of books and things like information's beautiful, it's like that's how I want to see it. But those kind of low code environments that enable you to be able to more effectively display information. So as an example, my team supported our spectrum team at the Commonwealth Games in England. So when you get a large event, you've got a lot of spectrum radio that's required. So teams of broadcasters are talking across to their camera people. You've got the director that will be giving communications there. They're also then broadcasting signal back to the shared across the world. And each of those different broadcasters, if you've got the Commonwealth Games from every single country, will need their own specific part of the radio spectrum within that geographic area contained. So we have people who are then monitoring those spectrum board ranges that are being used. And my team then helped develop a dashboard to support some of that. To understand which radio waves are being used where there could be interference across the different bands, but then also understanding who's been given licenses to use different bands within that area. And that's the case where you had the low code, no code being written into the dashboard, taking very technical information, presenting it back to then people who aren't necessarily data experts, but they are spectrum experts to be able to use that information, to be able to make decisions on the fly in real time.

I love that example, especially in how it unlocks your subject matter expertise, like being a spectrum expert. Right. And with data you're being able to improve and make an operation on the day to day much more efficient. Now one thing that you mentioned as well that we talked about is the importance of communication and being able to evangelize the importance of upskilling and data literacy within the organization. Walk us through. As the Chief Data Officer, how have you approached this aspect of change management and what do you think are best practices for engaging the organization with data upskilling and data literacy?

Yeah, so I think I mentioned that we're in the stages of finalizing our data strategy and actually getting out to everyone. So there's a large part of the comms that goes alongside how we work to get that data strategy out to people. So I've always looked at a comms strategy from a. Almost like, I don't know, I was obsessed with the west wings. They're looking at their messaging grid and how they get political messages out. It's almost the same thing where we've got channels of being able to get information out. We've got certain stories that we're trying to land at different times and try and say how do we reach people within the organization to be able to share that? So let's take go through some of those channels. We've Got an internal intranet that we use as a news site. We have yammer that's then going out. There's a weekly broadcast in Ofcom that goes out to all of the different sites and that's filmed and shared and then you have team meetings and beyond that we have things that we're trying to look at like ideation sessions and hackathons to be able to say how do we get some of this out? And also trying to say we've got data champions that exist within the entire Ofcom. So in different business areas, people who are part of the data profession, they're part of that wider community and they'll be able to share information out to their teams. So all of those channels exist. But I think there's also then that messaging piece in terms of what the key messages that we're trying to get out. So I came aboard those early days as an introductory message like this is who I am, this is what my team do. The facilities and the capabilities that we can add, that can add value into your business areas. Beyond that, there's a part where we're trying to collaboratively develop the data strategy. When I came into this, it was from a position where I didn't want that to be a top down dictat. It's not a this is the data strategy, you will follow it. I needed that to be from that organization. How do we break down the silos and make sure that there's one data community and we're working together. So we helped build four forums, were able to pull in some of that information. There's a lot of surveys and like trying to work with people to tease out some of the knowledge. And then that culminated in a large workshop where we had over 100 people. I think it was in the end, but from every single business area was represented there, whether data experts or technical experts to say how they use information, how they'd want to use information, what their business problems were and where there's some of the blockages in that sort of. That whole piece is part of the thing of building a larger community and being able to demonstrate the value that data has. And from that that's been able to build up some of that internal community going forward. I think it's then a case of some of those ideation sessions. We want to take some of those future technologies, some of those more advanced capabilities and talk to people and start saying, well, let's imagine we're 10 years into the future and data's great and everyone can get all the information they want. Now what do you want do with it? You're seeing examples of large language models of image generation, of whatever else it is, video classification. Now let's try and use some of that new technology that's coming out. And what would that mean in your business areas?

It's amazing. I love the approach to community here and I love how you've been able as well to incorporate maybe new advanced technologies as a way to draw a vision right of where we could be. What do you think as a chief data officer here, what is the vision that you're trying to communicate to the wider organization and how do you get people on board with that vision? I know that you mentioned here the importance of community. What do you think is a message that resonated quite a lot when incorporating new technologies as part of your vision?

Yes, I think this probably goes back to Ofcom's mission statement. So there's obviously, how do we make communications work for everyone? But then underneath that there's things around an Internet that's safe and reliable, a broadcast like all of the different areas that I've mentioned before, the OFCOM regulators. And I think part of it is helping people to understand the link from the work that they do to the mission and the vision of Ofcom overall. And that really shouldn't be underestimated. I think in the current environment we've got a huge shortfall of technical and data people across the world. Wages are much higher than a government organization would normally be able to pay. And don't get me wrong, we're definitely trying to keep up with it. But you see some of the wages that are paid by some of the largest tech and data companies. And so we're looking at what our vision is and our value that we offer. And I think part of that is around the value that Ofcom has as a whole. And you certainly see that from the Gen z, who by 2025 they're going to make up over a quarter of the marketplace and their value that they put into the wider community. And how the benefits that a company has in terms of some of what we're doing for the wider society is so much more important than just a what's my pay packet? And so trying to rely on the messaging around how important communication is, how it breaks down some of those barriers, how it enables people to live better lives, is really important for us. And then some of those value drivers, especially like the online safety is a huge draw where people see some of the dangerous aspects of online, some of the rabbit Holes that people can get drawn into. So how do we then start to protect children online? That's a big draw. To get people who maybe wouldn't think about working in this space into ofcom.

I love that. And Richard, this connects to my other section that I want to discuss with you here is the specificities and the challenges of working in data and government in the public sector in general. I don't envy the CDO's position. There's a lot of change that needs to be managed. It's a heavy burden from managing data infrastructure to technological change, cultural change. However, in a lot of ways the CDO's role is even more complex in the public sector and government. One of them, for example, you mentioned here, is being able to attract talent. Right. And being able to create a vision for attracting talent that is really relevant, which I think you do excellently here. What are some of the key government specific challenges data leaders can face when executing on a data strategy in the public sector?

I think one of the biggest challenges that I realized coming into quite quickly coming into OFCOM compared to where I'd been previously, previously was I'd been able to use pretty much whatever data that I wanted to within the reims of GDPR and everything else, but I'd been able to use the techniques and the data and information that I wanted to, to be able to make decisions and say, how do we use those? We're constrained by acts of Parliament so that we got our powers from the Communications act and then subsequent Act. You've got the online safety bill that's coming on now that will define what is within ofcom's remit of what we're able to do. We can't just then go and pull any data or request any information unless it's in line with the bill. The information that we request needs to be proportionate and there's a number of different areas. So the areas that I think is really different for me working in this sector that I haven't experienced before is then say, how do I make sure that I've got that really close tie in with our legal team and the public policy teams to be able to make sure that everything that we do is within the remit of what OFCOM is legally allowed to do. And that does create some challenges. And probably it goes back to some of the conversations we had about the curiosities that people have that we're employing now have come from academia, they've come from smaller enterprises, some of the startup tech companies, companies and places like that. Where they've had a lot more freedom, so they feel constrained. And there's a question now as to say, how do we make sure that they're not being seen, that there's just barriers that are put in front of them and someone saying no to them constantly. But how do we make sure that we're having those effective communications? So it's not just with the business, it's also then trying to integrate some of the legal and public policy teams that maybe other places wouldn't have to consider. Consider.

Yeah, I completely agree. And I think the flip side of this. Right. And you mentioned the importance of communication. Right. Is that when we talk about the challenge of managing data in government, I think it's also important to recognize the other side of the coin here, which is the responsibility government agencies bear when leveraging data. Ofcom has an extraordinary mission, as you laid out, of keeping the Internet, TV and radio safe, ensuring consumers get what they're paying for when it comes to telecom services. Walk us through how data science keeps the Internet safe.

Yeah. So the online safety bill is coming online this year and Ofcom's been selected as the regulator for that. So what does that mean in terms of the data team? So we have a group within Ofcom that's being set up, the online safety group that's going to manage the codes and the recommendations of codes. They'll be going out to. To public consultation. We also will then have a supervisory team that will be managing those closest supervision with platforms and enforcement teams. So all of those different areas will require different types of data when it comes to some of those examples. So just being able to understand the remit or the universe, the size of what there is in scope of the bill. So I think when it was originally set up, there was a. They thought that. I think it was 10,000 platforms within the UK would be in scope. As we did some initial analysis, that kind of increased to around 50,000. I think. We've recently done some work where we started pulling in information about all of the different web servers and what comes into different categorizations, whether that's a user to user service, a search service, adult services, to be able to say what could be in scope of the regime and suddenly we're up to potentially hundreds, thousands of platforms to be able to then understand what those platforms do, what the features are they have. Is it a user to user, is it search, is it an adult service? And then within that, trying to understand if it's an adult service, does it have age recognition on it so we can help make sure that children aren't accessing adult specific content. Things like that we can start to look at from a data science point of view. So it's just a trying to, to pull in that information to be able to understand who the platforms are, what their features are. Some of that information's early doors start of some of the capabilities that we're looking at beyond that, trying to understand what are the terms, the conditions that the platforms have in place that they're using to for instance, protect children online. And then on the internal side, I mentioned that we've got the policy that will be going out to consultation that's likely to generate a lot of responses. So there's a huge part of this to be able to say how do we understand what those responses are, how they link back to the policies and the code. So if we've got, I don't know, a thousand different responses that come in via email, web form, whatever it is, and they all relate to section 3.4 of the codes, how do we make sure that we can identify that? And that's partly using some natural language processing and trying to work through with that to be able to link individual responses back to the code to enable the team to respond to that consultation more effectively.

And Richard, with this overview of the challenges, but also with the responsibility that a data team and a Chief data Officer has within a government agency, maybe as we close out, I'd love to learn from you. What would be your advice for a newly appointed Chief Data Officer in government?

Have a very clear calendar and speak to as many people as possible. I think literally that communication cannot be underestimated. To understand what's the value that data can drive. And you can only find that out by having the conversations with people.

That is great. And as we close out, Richard, what are some of the exciting innovations maybe that you've been looking out for in this space? You mentioned here the vision of connecting technology to the mission. What are some of the technologies that you've been excited about?

Especially, yes, I think as every other person's looking at some of the capabilities that some of these foundational models will have, the large language models, but then also for me specifically some of those multimodal models, when you start looking at the impact that you can start to have with video, text and images, and being able to use all of those effectively together, that's going to have huge impacts on how we look at online safety. But just generally, I think there's that capability to make things faster. I saw someone today, someone emailed me in a video of someone who'd drawn a website on a napkin and shown that to ChatGPT and it had written the code for them. And I'm just saying, what does this mean for our team as we move into the future? And it's replicating what we always did. We'd go into stack overflow and take bits, chunks of code and copy them and paste them and kind of go, is this working? And then pull it together. It's doing that, but really, really effectively. So I think it's just going to make us more efficient for the future.

In a lot of ways as well. I think it will pose a lot of challenges for online moderation. If the cost of creating content goes down to zero, what does that mean for online safety as well? And there's going to be a lot of data science techniques that are needed as well to combat that velocity of content creation down the line, I think.

Yeah, the risks of algorithms and the algorithmic assurance is definitely something that will be looked at more broadly.

Yeah, definitely. So, Richard, as we close out, is there any final call to action you have for listeners before we wrap up today's show?

No, I think the big thing for me, as mentioned a few times, is keep the communications going and stay curious.

Curious, yeah. That's awesome. Thank you so much, Richard, for coming on with Data Frame.

Thank you very much.