Value Driven Data Science: Boost your impact. Earn what you’re worth. Rewrite your career algorithm.

[00:00:00] Dr Genevieve Hayes: Hello and welcome to Value Driven Data Science, brought to you by Genevieve Hayes Consulting. I'm Dr. Genevieve Hayes, and today I'm joined by Mikaela Ledwidge to talk about the ways virtual reality and data science can be combined to create interactive data storytelling experiences. Mikaela is the co founder and CEO of Mod, a studio specializing in real time and virtual production, And the creator of Grapho, a virtual reality platform that lets non technical users examine and manipulate graph data.
[00:00:40] She's also the writer and director of A Clever Label, a world first interactive documentary. Michaela, welcome to the show.
[00:00:50] Michela Ledwidge: Ah, thanks for having me.
[00:00:52] Dr Genevieve Hayes: In the 2002 movie Minority Report, the future of data interaction was depicted as Tom Cruise standing in front of a computer monitor and literally grabbing data points with his hands.
[00:01:05] Data interaction was shown to be as easy as interacting with physical objects in the real world. This vision of a world where data is accessible to all was considered to be science fiction when Minority Report was first released. But, over 20 years later, we're now at a point where technology has become good enough for this to soon become fact.
[00:01:28] And it's data science that's making this possible. Or more accurately, it's the intersection of data science and art. Now, like many of the guests on this program, Michela, your background is in computer science, but you're also a hands on creative in your own right. I mean, I think you're the first guest I've had on this program to have their own IMDB page, which is incredibly cool.
[00:01:54] To begin with. Could you give us an overview of your career to date, and in particular, how you came to combine technology and art?
[00:02:03] Michela Ledwidge: Sure. So I was fortunate enough as a kid to have access to a Commodore 64. I was supposed to not be using it without parental supervision, so it became this very fun illicit thing that I would sneak in and learn to code when my folks weren't around. But as a result, I was a kind of a beneficiary of the, sort of, the mid eighties explosion in Competency, you could say, when it comes to yeah, extracting interactive experiences out of computers.
[00:02:33] And it wasn't really a focus for me until university days. And to be honest, I had no intention of making my career, I was much more focused on music as a kid. But I found myself at university doing an arts degree, majoring in computer science and French, and that led to ending my university time doing an honours degree in computer science, right smack at the point where the web was kind of the rappers were coming off.
[00:03:01] So I was one of the, I guess the first hundred or so people. I set up a website at Sydney University, which turned out to be the first in the state. It was probably about the third in Australia and just basically had this incredible fortuitous opportunity to be paid to discover a new medium. So I went from corporate work as an architect.
[00:03:23] I started at the National Library of Australia. I set up their web services. I worked at the first web publishing company in Sydney next. And with that kind of in my back pocket, I set off to become an interactive filmmaker. That was kind of my B in my bonnet was that I'd sort of read a lot about virtual reality and played video games since, the mid 80s.
[00:03:43] And I was looking for ways to tell stories in new and interesting ways. Which I think is the number one thing you have to do as a director. So fast forward to now, I've sort of had a career that's really spanned science fiction and science fact. I use my computer science training to design architecture and tools but typically in aid of particular projects and productions and the Grapho product that.
[00:04:09] We're talking about today has really come out of that. It started as a Nietzsche. I wanted to scratch in terms of telling a particular documentary story. And now we've found a market in enterprise.
[00:04:21] Dr Genevieve Hayes: So, do you see yourself as a computer scientist first who's also creative, or as an artist first who happens to be incredibly technical?
[00:04:28] Michela Ledwidge: I see myself first and foremost as the director. And I often have to qualify that as saying I'm both a creative and a technical director. But yeah, the computer science is a qualification. I mean, I think it's a really exciting time to be a developer. I've been a hands on developer, most of my life now.
[00:04:46] But I've typically only sort of sold myself as a developer when times are really tough. It's more, it's just something that I do to, I like to work with much better developers than me and, build teams of more talented folks than me so I can, you know, I do messy prototypes and sketches and get functional experiences and then work with other people to clean them up.
[00:05:07] But right now, as you know, there's a real. see change into how you can produce functionality with code and all the tools and support services that are available. So I'm leaning into applied data science as an individual in a way that I haven't really done for decades.
[00:05:25] Dr Genevieve Hayes: So, All of these data science skills and computer skills and coding skills, they're just tools of the trade for your artistic
[00:05:34] Michela Ledwidge: Yeah. And I mean, it's not just about the art. I mean, I've been making tools. , since the 80s when I made a little basic app to roll the dice as a Dungeons and Dragons Dungeon Master. Like I've been building and supporting web services for over 30 years now.
[00:05:49] So the artistic work I do as a creative is very much one side of me. But I am a toolsmith as well.
[00:05:57] Dr Genevieve Hayes: endeavours. Which programming language is your favourite, by the way?
[00:06:00] Michela Ledwidge: Look, I moved to Python probably, gosh, mid noughties, and Python and C are sort of my main languages day to day my studio is an Unreal Engine service partner, so we do all that stuff together. A lot of work with that beast of a game engine where C is kind of core to how you extract stuff out of it.
[00:06:22] But yeah, for scratching stuff out and web services and data science, Python is my main tool of trade.
[00:06:28] Dr Genevieve Hayes: So, can you tell us a bit about your company, Mod? The website says that Mod's a real time and virtual production studio, but I don't understand what that means.
[00:06:40] Michela Ledwidge: So Mod is a 14 year old business that was set up after a startup I had in the UK kind of ran out of puff called Mod Films. And that was actually started with an invention award for the idea of remixable film. So the term mod actually comes from the games industry.
[00:07:00] In the early noughties the most popular online video game was a game called Counter Strike. And a lot of people just, you know, they don't know how games are made. They just buy them or you just play them. So, It was basically a modification of a commercial game called Half Life.
[00:07:18] Two guys had basically hacked together a kind of goodies versus baddies counter terrorist, you know, in big quotes shoot em up based on this very critically acclaimed sci fi First person action game made by Valve Software called Half Life. So basically what they did was they actually made a modification of a commercial product that transcended the original product and in some ways became something completely different and took on its own life.
[00:07:47] So being part of that world, I'd always wondered why doesn't this happen with film and TV? And the answer is because the system is set up to really prevent that from happening or certainly more back in those days. But I basically won an invention award in the UK for the idea of remixable film the technical architecture and the kind of creative and business architecture that would need to support it and spent six years of my life trying to prove out that model.
[00:08:14] Dr Genevieve Hayes: What do you mean by a remixable film, because I can't envisage that. Oh,
[00:08:18] Michela Ledwidge: yeah, so you mentioned Minority Report. So, what comes to the screen as Spielberg's film Minority Report is an edit of a production, and most of the assets and creative and technical labor that go into making it, it's like an iceberg. You never see, most of that material. The idea of a remixable film is that it's legally and technically possible for the audience to take the film and craft something new.
[00:08:45] So it's almost like you were distributing films like a Lego kit. You were still making a film that you know, it was designed to just, you hit play and it plays. But if you were a creator yourself, He could disassemble it, you could access the rushes, you could access all of the 3D assets basically every aspect of the production.
[00:09:06] It was an experimental business model, the idea that everything that goes into a production was licensed under the Creative Commons IP structure. And you actually create creative and business incentives for audiences to remix. In the same way as the music industry had to be dragged kicking and screaming to the point where they now have sampling departments that make enormous amounts of revenue from the licensing of remixes.
[00:09:34] So that model hasn't really ever happened in film and TV for more reasons that we can go into on this podcast. But I was involved in a lot of exploration of that space. And so MOD today is a much more traditional organization. But under the hood, we still go via the notion that any production should be as remixable as possible for the creators.
[00:09:59] And if we decide to make it remixable by the audience, we can, but some of these ideas of remixability are now. Completely mainstream. You look at what's on TikTok, you , you , look at how social media works. People remix without even thinking about it. And that's due to the medium itself and the platforms of today support remixing like never before.
[00:10:21] I've spent my career building pipelines that allow me as a creative technologist to repurpose, refashion, and iterate really quickly. And there is, I think, some real excitement when you bring that to interactive experiences like video games and even in the enterprise tools that allow the user to be more empowered by the experience.
[00:10:45] Dr Genevieve Hayes: , I remember a number of years back seeing it was a trailer for the movie The Shining that had been remixed so that it looked like a romantic comedy.
[00:10:55] Michela Ledwidge: Yes, yes, I remember that, Shining. That was a great example of editing.
[00:11:00] Dr Genevieve Hayes: Yeah, so is that the sort of thing you're talking about?
[00:11:03] Michela Ledwidge: So, the idea that if I made the Shining That rather than that just being produced as a fan thing, that the actual product, The Shining, had some kind of tool or functionality that allowed the audience to remix it. That was the concept.
[00:11:19] Dr Genevieve Hayes: And turn it into a romantic comedy.
[00:11:21] Michela Ledwidge: Exactly. Exactly. And you know, that fills a lot of storytellers with horror because a lot of storytellers you know, the prime focus is to get their version of something out.
[00:11:30] And in a way that's the same for me, but where I'm a little bit different is that as a tool Smith, I'm really interested in when somebody loves your work so much that they want to transcend it into something else.
[00:11:42] Dr Genevieve Hayes: So in addition to production services, you told me when we first met that you've also recently pivoted into offering data science and machine learning based services. I'm guessing that's not a standard offering for most production studios. How did you end up moving into that space?
[00:11:59] Michela Ledwidge: Yeah, it's a good question. I mean, in some ways we always have been developing these services internally. I mentioned that my first. Job after university was setting up the national library. As an indie studio, we live or die based on the original IP we create. And throughout the history of my companies, there's always been a real effort to.
[00:12:21] archive and store and maintain the work that we create, in a way that we can internally reuse it as much as possible. And when it comes to creating video games and complex narratives that might go out into film and TV, storing the elements of the story world, has got some value and it's often hard to see unless you're like Marvel or some giant organization, but , I've found over the years that I've been very influenced by my early years working in large companies like the National Library and the BBC and Reuters where I was doing a lot of content management systems.
[00:12:58] And so I've actually spent a lot of time thinking about, well, how do you store media productions and data of any sorts in a way that helps you revisit it down the line and then craft new narratives out of. And so having done a little bit of neural networks at university many moons ago I was really quite astounded to discover in last five years or so that a lot of the boom in, gen AI , and data science, the new age we're in now, is really such a clear sequel to the processes I learned at university.
[00:13:31] So it was like, it wasn't a huge stretch to go back to what we already had in terms of systems for storing topologies of story universes or ontologies of how, a particular training simulation might carry out and go, well, with a little bit of rejigging. We can make all this data available as a training set for some new application.
[00:13:54] So that's kind of what led to the Segway, the realization that there was an opportunity to really focus on just because you have the data doesn't mean you necessarily got the clearest insights out of it. So we use the phrase data science plus storytelling because I feel like a lot of tools for data science have been traditionally a little bit siloed away from non technical people.
[00:14:19] And it feels to me that in an age now where so many decisions and so much potential is kind of tied up with this. Machine learning revolution. It's never been more important for coherent storytelling to happen with data. So there's an opportunity to revisit things in terms of accessibility, but also explainability.
[00:14:42] And that's where storytellers come in.
[00:14:44] Dr Genevieve Hayes: So what sort of data science techniques are we talking about here?
[00:14:48] Michela Ledwidge: so we're currently working with the National Film and Sound Archive on basically rebuilding their. Fantastic collection of the history of Australian multimedia, essentially as a knowledge graph and as part of that are you familiar with graph databases?
[00:15:05] Dr Genevieve Hayes: Yes, I am.
[00:15:07] Michela Ledwidge: So, we've sort of jumped very deep into using graph databases over the last nine years.
[00:15:14] Mainly because of the additional context. That we can capture with modeling relationships, not just information, but the connectivity between information and with a collection like the National Film and Sound Archive we were brought in specifically to look at topic modeling, so to be able to automatically create hierarchies from unstructured data.
[00:15:36] So here is the history of Australian media, what topics exist in the collection based on, what's in transcripts, what's in the content of an image, as well as all the more formal data like archivists have put in categorization over the years.
[00:15:51] But it's about using multi module things like, Whisper for audio extraction, Lava for the content of an image to be recognized. But also using semantic chunking to basically look through an old database. I mean, it doesn't matter how well you set up enterprise data systems over time, things become messy things.
[00:16:13] So just taking a view that A database structure of any source is always going to have a mix of well structured material and unstructured messiness and chaos. Using some of the new techniques to basically analyze, as, data scientists would typically do in, Explorations in Jupyter notebooks, for example, looking through and just identifying, telling the story of what you have.
[00:16:40] So looking for gap analysis, looking for opportunities to improve how information is stored, looking for ways where. Certain information was too expensive to be captured, but potentially now, there's new ways that you can do it. So it's about keeping track of the state of the art of, across a lot of different areas, but also working out some practical way forwards, which have to assume that, Well, there isn't all money and time in the world for human archivists to re categorize things.
[00:17:07] So where can models help? So crafting the story of, well, this is what you have, this is what you could do. And this is the roadmap of how you can actually build a new solution. That's the kind of tooling and processes that we're involved in.
[00:17:23] Dr Genevieve Hayes: What you've just described at the Film and Sound Archive sounds very similar to a project I worked on in a previous organization. Except we're using, it was data from a social media platform. So we had images, video, audio, text. And some of it was foreign language. So we had to extract all the metadata from that, get transcripts translated into English do summaries of what was in all the images and videos.
[00:17:55] And then we created a knowledge graph from that by connecting things like, who sent the message What were the key things they were talking about, things like that. So, it sounds like what you're doing is that, but on a much grander scale.
[00:18:12] Michela Ledwidge: Yeah, I don't know about more grander, but that's exactly the kind of work that we're doing with them. And the fact that we make and sell XR visualization tools are not so relevant right now, but it's part of their interest in us that, we're sort of at the moment we're working at the coalface in how do you Move everything into this knowledge graph paradigm, but knowing that at the very end of the process we have other customers like Neo4j who make the property graph database, Neo4j who have licensed our XR tools.
[00:18:45] Specifically to help their customers and potential customers understand what a property graph is itself. So they have a little activation template now that they use all around the world. So that if they've got two minutes to get someone interested in the idea of their product, if they want to go into VR and literally grab the graphs like Tom Cruise did in Minority Report they can do it because our experience is pretty good.
[00:19:12] Polished like a video game. But it's as generic as a web browser. So the data could be anything, so it works best with smaller data sets. But I think the trap and the trick with this new generation is that just because you can work with these incredibly large data sets and the complexities and the interconnectivities have never been more sophisticated.
[00:19:36] Human beings still can only hold a certain amount of information in their head at any one time. So that's where I think traditional tools of storytelling, from the old campfire analogy through to this minority report one, which I find very useful for those who've seen it. Not so useful if you haven't seen the movie.
[00:19:54] But just the idea that, oh, I'm interested in this relationship between two bits of information. Oh, can I just reach out and grab it with my hand and look at it? I find that a really powerful analogy and that's why I set out six years ago to make it a real tool. And yeah, it's been quite satisfying watching we're not a big company, so we don't have huge marketing or, incredibly fast velocity of rolling out tools, but there's something there and it's the simplicity of it, I think, that's What people are reacting to.
[00:20:24] Dr Genevieve Hayes: Did you have Minority Report in mind when you started building your virtual reality graph tool?
[00:20:30] Michela Ledwidge: No, in fact, what I had in mind was Al Gore's An Inconvenient Truth because he famously had this PowerPoint on climate change, and then, you know, fast forward, he had a Academy Award for a documentary, and , the most visual effects y bit in the movie was him getting into a cherry picker and going up in a physical cherry picker to show how high the climate change graph spiked.
[00:20:55] And I just thought, oh, that's a cute trick on stage, but as a VR practitioner, it'd be kind of cool to have my own virtual data world. It'd be cheaper to set up, and I could tour it and I could stand in virtual worlds physically in different parts of the world. So I was basically developing a traveling show.
[00:21:13] The technical term in our trade is location based experience. And having done a lot of video installation, interactive media, interactive art installations, and also activations, I thought,, I think there would be an opportunity to tell a data story live. But within a virtual world and tour it and then this COVID thing happened and it was like, oh, well, that's not going to happen.
[00:21:38] So we pivoted to focus on releasing a clever label as a kind of more traditional download a VR product , from an app store model, but it started with the idea of, yeah, Al Gore's slideshow. And having, like, a Jupiter notebook attached so that if the data scientists or journalists wanted to dig deeper, coming back to that remixable film analogy, all of the data was licensed under Creative Commons.
[00:22:04] So the idea was that, if you didn't like the way I told the story, for me, the main thing was that people were able to access the data and draw their own conclusions.
[00:22:13] Dr Genevieve Hayes: Do you do anything with generative in all this data science stuff you're doing?
[00:22:19] Michela Ledwidge: Yeah, it's a really, really interesting question. So as an artist, I've spent quite a lot of time looking at what's happening, what's coming and but I actually spend I guess we don't make much use of generating multimedia at the moment. The main thing I use Gen AI for at the moment is generating summaries, analysis, so text.
[00:22:47] And yeah, it's an interesting time because I actually find, , there's a huge schism, as you can imagine, within the creative sector about the use of Gen AI to create content. I'm actually, as a writer I'm very interested in using Gen I as supporting tools, but yeah, at this stage it's still playing a secondary role I could say.
[00:23:10] The thing I'm generating the most, actually, is Code I'm generating an enormous amount of code, whether it's troubleshooting or a refactoring of work. So , I'm definitely making a huge use of Jedi, but we're not putting Jedi content front and center. Partly because I think maybe it's ego, but the role of a, creative studio and artist and into creatives is to find their own way through.
[00:23:36] whatever technologies exist. And yeah, there's nothing, nothing to date makes me want to yeah, replace the content as yet. But when it comes to visual effects and, invisible effects like a lot of studios, we've done a ton of experiments and we've used Gen AI in productions we've used deep fake techniques.
[00:23:58] We do a lot of cleanup work. But typically it's not as kind of that naive thing of, text to content that is sort of gets a lot of the attention at the moment. That's not really our focus.
[00:24:11] Dr Genevieve Hayes: Yeah, so you're not going to type in ChatGPT produce me a movie where, I don't know, Humphrey Bogart interacts with Tom Cruise and Taylor Swift or something like that.
[00:24:25] Michela Ledwidge: Yeah, it's, I mean, it's, it's interesting. I'm not, I'm not as interested in that. And maybe it's just because I've, I've juggled. I've worked as a technologist, worked as an artist my whole career, and for me, art is the exploration of concepts, and I don't have any issue with people doing it.
[00:24:43] A friend of mine sent me a music video last night that was completely created from all of these tools. And it was pretty weird and, my feedback was, keep going, that's great, but it did remind me of my early days as a club VJ remixing, films, TV, whatever I could get my hands on or I was interested in that week remixing stuff live so in a way I feel like, Pattern recognition and the kind of multimodal activity that these models are doing yeah, in a way for folks who don't have certain skills and trade, I think that it's almost more fun to do it.
[00:25:22] But for me, it's always about, okay, what am I trying to do away from this? Where can this play a supporting role? But yeah, I'm less interested in coming up with the idea from scratch at the moment.
[00:25:33] Dr Genevieve Hayes: I know a lot of creatives in Hollywood are getting very nervous about Gen AI and there was a lot of blowback when Ashton Kutcher made comments about Gen AI being the future of Hollywood, what impact do you believe Gen AI will have on creative industries such as film and video games production in the future?
[00:25:54] Michela Ledwidge: Yeah, look, it's a great question. I think the kind of work that we're doing at the National Film and Sound Archive, where you're essentially preparing and fine tuning your data so that you can take advantage of AI tools of any sort. is the first step. So I think regardless of people's visceral reaction, there's a really strong argument for knowing where your data is and keeping it.
[00:26:18] in a form that you will be able to take advantage of some of these approaches in the future. Whether you choose to do so, there are fights looming and there are fights going on already about, unsanctioned use. And ultimately there's been a bit of a coup.
[00:26:34] I mean, a lot of our data has been taken and used without our knowledge. And so, there is a huge danger of these tools not actually empowering that many people through it, but at the same time, I think the ball is in every creator's court and every organization's court to understand, to get practical experience about what is involved in retaining ownership of your data.
[00:26:58] and leveraging it in a way that's useful for you. And it may end up just being, okay, I'm going to license it to these larger companies and, try and get some revenue from the world at large being able to access the DNA of what I've created. But think what we're looking at now is so transformative. Most creative. Teams and organizations and individuals will build their own relationship. They won't just be able to have a knee jerk.
[00:27:24] Oh, I don't want to touch this in my life. I mean, even just this week, seeing Demis Hassabis, who I remembered as a young game developer in London, getting a Nobel Prize for using the deep mind approach to visualizing proteins. I mean, this stuff is so much more transformative than just here, generate a music video for me.
[00:27:45] I think. The key is to be able to separate private and public domain information in a way that is fair so that the creators of work are able to actually retain ownership. And that's going to be super tough. And it's so hard to retain control of your own digital IP in a world where, every company wants to give you, you know, sign up to a service that then might have dark patterns that automatically upload your data to their servers.
[00:28:15] And even just this week, realizing that OpenAI is training on my data, but even though I pay them 20 US dollars a month, but it's the plus account, not the business account. And all of their statements are around, we don't. Train on the business account so they haven't made a single public statement about the paid for plus account So, of course, they're training on it.
[00:28:37] You know there is a level to which we're being swept along by this, sea of greed and enthusiasm and there's empires moving as we speak, but I do think there is a role in individuals and smaller teams in carving out intellectual and creative and business space that suits them.
[00:28:58] And then, putting down stakes for a fight to be able to do that. But it's a really tricky time at the moment. Now, I think for me, it starts with the organization of the data and I'm a data architect. So I am going to always say that, but I do think before any applications, it's really important to just know what you have and try and structure it in a way that it's useful to your teams locally.
[00:29:22] Dr Genevieve Hayes: One thing that I. What I find very positive is if you compare the way people are behaving now with 20 years ago when the internet was first starting to become a big thing, 20 years ago, people were very naive about internet companies. I think a lot of us believed that Google was just a nice company that wanted to provide search services to us out of the goodness of their heart.
[00:29:50] And now we realize, well, They were getting all our data and selling it to advertisers and et cetera, et cetera. And people are awake to that. So they know what to look out for when it comes to these companies that are selling generative AI services. So they can start putting stakes down a lot earlier and say, I don't want my data harvested, and if you're going to take my data, you better be paying for it.
[00:30:15] So I think that's one very good thing. And the other thing that I think is really promising is, you've seen with the coverage of the current U. S. presidential election, There's been a lot of people who are saying, no, we do not want to see deepfake ads or deepfake images of either of the presidential candidates because that is misleading.
[00:30:42] So I think people are a lot better educated now and know what they do and don't want. And I think that'll help get a much better solution for the world than had this come along 20 years earlier.
[00:30:56] Michela Ledwidge: Yeah, no, it's interesting. I agree with you that, yeah, we have been forewarned a bit. I think there's so many machinations though I think You can spend all your time worrying about how your data is going to be misused and, how do I choose ethical providers that I sometimes find myself in a bit of paralysis and I've actually just got to move forward.
[00:31:17] I've only got one life and I need to get things done. And like, for example, I've had a personal boycott against Facebook for quite a few years. And Found myself in this situation where I'm running a business that can't avoid Facebook products because they bought Oculus. Basically we can't operate as a VR studio without using Facebook owned, I mean, I should say met us.
[00:31:39] I always say Facebook, but Facebook owns technology and, amazing engineering teams, amazing products come out. Do I like their business model? No, I don't. But I have this uneasy relationship with the fact that their VR product lines, are the most pragmatic way of us reaching an audience at this point.
[00:31:59] And do I wish there was more competition? Yeah, absolutely. But in a way Yeah, just trying to keep track of the dark patterns in the technologies that we have to use in order to survive as a business is it's like a full time job. And I find it quite draining, but sometimes you just have to, Close your eyes and push forward.
[00:32:21] Dr Genevieve Hayes: So I'd like to go back to Grapho, your graph data science storytelling tool. So at the start of the episode, you mentioned that in addition to using this for creating creative stories. You're now also offering this to businesses. And it sounds like the work you've been doing with the film and sound archive is somewhere between creativity and a business offering.
[00:32:48] Can you give us an example of how you're going more in that business offering direction?
[00:32:53] Michela Ledwidge: Yes. And look, to qualify, it's not really a creative role at the National Film and Sound Archive. We've been doing prototypes and rolling out production services. So that's very much on the data science and the data engineering side of things. So Grapho , as a product line started with the idea of VR visualization of graph data.
[00:33:14] And also another qualification is there are very few creative studios using these processes at the moment. So these were invented for original productions by my studio. We've found them to be successful ways of working. We'll continue to do that. But in a way, our pivot to enterprise for Grapho.
[00:33:35] this is going where the market is creative media production is a very high risk area and typically the choices of production technology are incredibly conservative people wait until there's, an academy award winning project using a particular method before they dive into it.
[00:33:52] So we're under no illusions that, we kind of need to prove this model around creative storytelling first, but with enterprise, it's a little bit different. So we didn't start off looking for an enterprise customer base. What happened was Clever Label came out and because the topic of that show was LGBTQI rights and dark patterns in the dark money trails that fund the opposition to LGBTQI people in Australia.
[00:34:23] We had quite a few people say, look, I'm not interested in this topic or, I don't even want to talk about this topic, but I really liked what you did with the technology. How do I get hold of it? And so it was a bit backhanded because the whole point of doing the documentary in VR was to take quite boring, depressing data that's quite obvious to LGBTQI community members like myself but People outside of those communities, don't necessarily know what some of the things are going on are.
[00:34:49] So, in a way, that was a big tick for me. It's like, okay, cool, this is actually helping with the storytelling. So, from that point on, we were like, oh, let's just keep our ear out to who has issues with communicating. Their data, and the first client we had was the Asia Pacific Network Information Center.
[00:35:09] Have you heard of APNIC all the NICs around the world? So this is kind of interesting. So, Most people use the internet and are aware of, old school protocols like SMTP for, email, and of course HTTP from the late 80s, early 90s hypertext transfer protocol.
[00:35:28] But there are these other protocols like border gateway protocol, which are the protocols that actually allow the internet to constantly retain shape. And you heard the old Cold War analogy that, the internet can self organize and repair damage and all that. Well, how does that actually work in practice?
[00:35:47] It's not a topic that's appears much in the media for two reasons. One, it's a bit complicated, but also it's actually a real Area of weakness, if you like, in the system and there are these organizations for each region of the world. They're like the systems administrators of a particular region.
[00:36:04] So APNIC looks after the Asia Pacific section of the internet and then there's other regions around the world. If you're a telco or someone who's setting up a giant organization, you need to get new IP address ranges. You go to the local NIC and you apply to get internet address space, and your credentials are checked, they do due diligence, and, if something ever goes wrong, or there's activity on your section of the network that is breaking the terms and conditions, then these are the forensics analysts who actually go in and have to basically work out what's going on.
[00:36:40] And, why is. All Genevieve's traffic going through Ghana, and so it's really interesting complex data mapping. And they said, look, we've got every tool you can imagine. Cognitive load is a real issue. Could you build us a forensics tool using your GRAFO system? And at the time it wasn't called GRAFO, it was just the VR component of our indie documentary.
[00:37:04] And we said sure, and three months later we were able to deliver them a system that allows you to basically navigate the 43 million nodes of the Australian Pacific internet. You could say, bring up Fiji, and in our little system, , bringing up the internet of Australia, showing all the nodes, there's too much information.
[00:37:24] But for a smaller country like Fiji, Fiji was a great example, because someone who worked for APNIC, who was on a visit to Fiji next week, and just wants to brush up on, well, what is the state of the internet in Fiji? The 20 or so. nodes, from the banks and the telcos. So seeing the overall shape of the country's network and being able to see the relationships becomes really non technical and accessible.
[00:37:51] So that was a pilot that we did three years ago, and that really opened my eyes to the potential of us just focusing on, if you like, a generic capability to bring data out of this kind of data science silo where, you need to a computer science trained individual to get their hands on the data and then through them the insights can be delivered to management.
[00:38:17] Really to latch onto this idea of what can we do to allow the end user or the stakeholder in this startup who may not have any technical training, what can we do to give them more empowering data experiences than they currently have. We all know about dashboards, we all know about the traditional approaches, And so this idea of being able to reach out and grab data and use haptics and body language to explore things we're on a bit of a mission to say this should be as boring and as powerful as the mouse, I don't think we need to sing it from the rooftop as all data is going to be accessed in this way.
[00:38:55] But why shouldn't all data be able to be accessed in this way? And if we can continue building out the capabilities to bring your data off the desktop, off the laptop screen, into XR within a few minutes for the purposes of whatever, then I think that that is a worthwhile pursuit.
[00:39:17] Dr Genevieve Hayes: And how does storytelling fit into this?
[00:39:19] Michela Ledwidge: Well, storytelling for me is leading people on a journey. And the way I approach storytelling as a director is telling a story in interesting ways, being aware of your audience, being able to keep them engaged when, intention is flagging and making it relevant to them. So with data storytelling, I think there's a real creative tension between making powerful tools that give the operator incredible scope for repurposing data, but that's usually at the expense of leaving non technical people with or time poor people behind, because you actually need to invest a significant amount of time in order to be able to develop your storytelling abilities with this tool.
[00:40:11] And we use, I mean, you mentioned virtual production was an unfamiliar term before. Have you heard of those shows that get filmed now on giant sound stages where the actors have got basically video walls behind them.
[00:40:26] Dr Genevieve Hayes: Yeah, I've heard of those. I haven't seen any of them, I don't think.
[00:40:28] Michela Ledwidge: So that's an example of virtual production that's quite popularized, where instead of like rear projection, you're actually using video technology. And so the actual backgrounds are coming from video game engines typically. So this idea of using real time graphics and audio capabilities for filmmaking, that essentially is what virtual production is.
[00:40:52] So we, as a small studio, we sort of sit between film and game. And so we're taking advantage of interactive technology paradigms that are honed in game development, but have really useful places in traditional storytelling. The audience may never know that they were used, but those tools are available on set.
[00:41:13] So you take that idea and extend it to the enterprise. What if I've got a story to tell, and For whatever reason, the traditional dashboard or my notebook approach or my reporting capabilities is not doing the job. It's not capturing the attention the insights aren't clear enough. What other tools are in my toolbox?
[00:41:36] So the idea is that being able to slurp in, perhaps it's just always sitting alongside these more traditional tools, or perhaps it's just an alternative being able to bring your data. into new kinds of visualizations that aren't limited to 2D display on screens. That's where, for me, the storytelling comes in.
[00:41:58] So we have found with younger audiences that they want to reach their own conclusions. So there's less interest in just being presented with the data statically and, as you know, data visualization has got to the point where there's very sophisticated tools that might allow a little bit of interactivity on the page, like move a slider or there was a fashion for 3D visualizations where you could rotate things.
[00:42:23] Now, for me, that's a very sculptural approach. If you create a 3D structure out of your data and allow the user to zoom in or rotate it's still like a sculpture. You can't touch it. What we've found is that if you can present the data in a way that it may have an original shape, like a sculpture, But if the audience can actually reach in and mold it like they're actually doing the sculpting themselves, that becomes incredibly powerful.
[00:42:49] So you can see this all relates back to remixable film in a way. It's always about saying, No matter how complex the possibility spaces are of your interactive product or experience, we still live in a time based universe. So we still receive information in a linear way. The value of being able to get from A to B to C And hold the meaning through to a conclusion is really powerful.
[00:43:12] So should we just be limited by, whichever data visualization tool has the most market share, or is there an opportunity to, get the data off the page into a more malleable form? That's kind of my quest with this.
[00:43:25] Dr Genevieve Hayes: So it's allowing the end user to effectively take on the role to a limited extent of the data scientist.
[00:43:32] Michela Ledwidge: Exactly. And this is where it becomes really important to manage expectations. We've found probably the most successful use case with our tools. Recently has been Neo4j's decision to roll out the use of graph OXR to all their field engineers and marketing folks for the simple reason that, and you might find this hard to relate to because both you and I are practitioners in this space and understand graph databases, but not everyone understands what they're talking about when they, talk about property graphs.
[00:44:07] So just giving people a property graph that's floating in the air around them. We use mixed reality, not virtual reality at these events. So the audience is still seeing the real world.
[00:44:19] It's so they've basically got a few nodes and relationships hanging in space and the way we've crafted the experience that they're running with is that the audience has filled in a form already. So not only do they have a very accessible collection of nodes and relationships, they can literally grab.
[00:44:37] With the hands, but they're in the data. So they can see a node with their name and it's attached to a relationship to a skill that they've identified and they can grab the relationship and it'll say, Genevieve is skilled at data science and lives in Australia. So it's not about dumbing it down, but I've used my tool And lots of conversations and this great relationship we have with that company to kind of lead an exploration over the last couple of years of how can we take a really powerful concept, but make it really simple and engaging for someone who.
[00:45:16] May not give them time of day for more than three minutes. Can we give them a one minute experience of something that's going to be really powerful? And this is the kicker. 99 percent of the people who come to these activations have never used VR before. So we've got a doubly hard job because we're not just introducing them to Neo4j's product line and the concept for a property graph.
[00:45:40] We're sharing an experience in a medium they've never experienced before. But because we've done our job reasonably well, And VR was first created in the 60s. So even though market share wise, it's still considered fairly niche, but we have a lot of, robust methodologies and practices to build on.
[00:46:00] So we're certainly not the only ones doing this, but it is possible to introduce someone who's never heard of VR to VR and actually give them a good experience within a couple of minutes. And so that's where I think enterprise use of XR is really growing steadily.
[00:46:17] There's companies like Ford who were really early adopter to VR. Um, And Ford had these famous rules, you know, the rules of production. They added one that said no change the customer may observe can now be deployed without an immersive design review. In other words, everyone has to get into VR and look at that change to the car before it goes into production.
[00:46:43] And so, just little things like that in the history of XR mean that, we're not selling snake oil or a a gimmick. We're talking about trying to identify really discrete user cases and fill them. And for me, the best way to fill them is not to say, we've got an all singing, all dancing, super complicated tool that we've made available for sale that people can build.
[00:47:05] It's about saying we've got a toolkit. And when people come to us, we do want to make the toolkit more readily available. But at the moment, because it's all bootstrapped, the most successful way we've found is to basically have these conversations with companies and build bespoke versions of what we've got that are very finely tuned for a very specific use case.
[00:47:27] And once people have got their head around the use case, they can customize it and use other data because, It is essentially a web browser for graph data. It literally is that. You can connect to any Neo4j database through it. But the starting point has got to be a good experience in those first couple of minutes.
[00:47:45] Dr Genevieve Hayes: Hmm. Otherwise you've lost people for life.
[00:47:47] Michela Ledwidge: Absolutely. I mean, we have to deal with things like, you know, people wearing too much foundation who, don't want to have their makeup mucked up or perhaps the owner of the hardware doesn't want them putting their face into it because they don't want to get the foundation on it. So we have to basically have, masks and all kinds of hygiene related activities that don't just exist and are accessible, but our customers understand the need to think about this stuff and, hygiene is not something that typically purchases of digital products worry about too much, but when it comes to VR, it's a very tactile, it's a different kind of medium.
[00:48:26] So it's a really interesting space to work in.
[00:48:29] Dr Genevieve Hayes: So what final advice would you give to data scientists looking to create business value from data?
[00:48:35] Michela Ledwidge: Look, I think as we all lean into this incredible, you know, frankly, astonishing power of machine learning, I think it's fair to say that a one size fits all automated approach isn't going to cut it. So as much as we all use automation, as much as we all are exploring and leveraging the humans driving the outcome isn't going to go away.
[00:49:01] And because of that, coherent storytelling isn't going to lose its value. I think it's going to increase its value as people say, explain to me how this works. You've told me something, now show me why that's the case. Because I think as you said before, we're a little bit more wise to the snake oil sales people.
[00:49:21] We're wiser to the dark patterns and the, conflicting drivers for why people are offering services. So a key opportunity in storytelling is to show rather than telling. And showing is all about explaining.
[00:49:37] Dr Genevieve Hayes: For listeners who want to learn more about you or get in contact, what can they do?
[00:49:42] Michela Ledwidge: I think the best URL to go to is Grapho do A-P-G-R-A-P-H-O-A-P. At the moment, that's just a deep link into the MOD website, but as I said, this has started off as a bootstrapped endeavor of an indie studio, but we will be spinning this off fairly soon.
[00:50:00] We will be releasing a demo version of the XR tools. Next month. And yeah, there's plenty of contact details there. And I'm just Michaela on a lot of the channels.
[00:50:11] Dr Genevieve Hayes: And thank you for joining me today, Michaela.
[00:50:14] Michela Ledwidge: Thanks Genevieve, enjoyed that.
[00:50:16] Dr Genevieve Hayes: And for those in the audience, thank you for listening. I'm Dr. Genevieve Hayes, and this has been value driven data science brought to you by Genevieve Hayes Consulting.