Incorporating OSINT into the Defence Intelligence Environment

Media Thumbnail
  • 0.5
  • 1
  • 1.25
  • 1.5
  • 1.75
  • 2
This is a podcast episode titled, Incorporating OSINT into the Defence Intelligence Environment. The summary for this episode is: <p>In this podcast episode, Harry Kemsley OBE and&nbsp;Sean Corbett CB MBE talk to Terry Busch, Executive Advisor at Capax Analytics and Former Chief Technology Officer for the DIA's High Priority Machine-Assisted Rapid Repository Program (MARS). </p><p>In this discussion we cover the increasing relevance and utility of Open-Source Intelligence (OSINT) in support of the defence and security community. Specifically, the expert panel discuss challenges and opportunities of incorporating OSINT into the defence intelligence environment.</p>

Speaker 1: Welcome to the World of Intelligence, a podcast for you to discover the latest analysis of global military and security trends within the open- source Defense Intelligence Community. Now, on to the episode with your host, Harry Kemsley.

Harry Kemsley: My name is Harry Kemsley. I'm the president of Government and National Security for Janes. A huge privilege to be at the helm of what is one of the greatest organizations of this type in the world. We're here today to talk about open-source intelligence, a topic extremely close to my heart, my professional heart with my role at Janes, and how it can be incorporated, better incorporated into the defense intelligence environment. To help me in that discussion, I have two extremely expert colleagues with me who I will introduce in just a moment. Our intention is to consider the challenges and indeed the opportunities of incorporating open- source or publicly available information and intelligence into the defense intelligence environment. First of all, allow me to introduce Shawn Corbit, who is a long term colleague and friend of mine. He is the founder and CEO of Insight Global, a consultancy specializing in strategic thinking, open- source intelligence and business optimization. That's since he finished working briefly for a satellite defense analytics company. For those of you that know Shawn, will know him to have retired from the Royal Air Force in 2018 as a Air Marshal, after a 30- year career as a professional intelligence officer through which I knew him and encountered him many times. He is a fearsome man to cross when it comes to matters of important mission focus. His last appointment, I think is worthy of note, in the military was two years in Washington as the first non- Us Deputy Director of a major US intelligence agency. Shawn, a real pleasure to have you with me. Thank you for joining.

Sean Corbett: It's great to be here, Harry.

Harry Kemsley: Now, let me introduce Terry Bush, a legend in his own time, a man who I have seen and followed many times. I'm genuinely privileged to introduce to you as an entrepreneurial and innovative former government executive with over 25 years of experience, developing new government capabilities, programs and enterprise programs. He is known and well regarded as the builder of leading edge artificial intelligence technologies, big data and data science analytical programs. At the moment, Terry is developing new platforms for government and industry focusing on leveraging open- source data and software. I hope you can see as clearly as I can, two true experts to help us on this conversation. A real pleasure to have you both here. Thank you, Terry, for joining.

Terry Busch: Thank you for having me.

Harry Kemsley: I think what we should do first, gents, is to frame the discussion by being clear about what we mean by open- source publicly available information becomes intelligence, so- called OSINT. I say that and I ask you to help me define it because it will provide a platform for conversation as we go forward in the next 20-25 minutes. For me, I think it's sometimes the source of misunderstanding of the value of open- source intelligence. I use that word intelligence very carefully and very deliberately. Terry, if you don't mind, I'm going to come to you first, Shawn, I follow to you, in terms of how do you define, what do you think of when you hear the phrase, " Open- source intelligence."

Terry Busch: Sure, absolutely. Thanks. In an old paradigm, we would call something... The difference between information and intelligence is that information is assessed. It has been assessed to have intelligence value and a judgment has been made about the data. In today's world with open- source information, I re- define that to say that we've contextualized that information, we've extracted it for its veracity, its confidence in being right. We've got to the point that we can use data as intelligence, meaning that it's not just some sort of data that we've just pulled and inserted in the process. But we've put it into some sort of validation chain to make sure that the data can be used in a highly interoperable and a way that we trust the information.

Harry Kemsley: Thank you, Terry. I will come back to a couple of points you made there about contextualization, perhaps later, but Shawn, can I just to throw it across to you in terms of how do you think about what people call as open- source intelligence?

Sean Corbett: Yeah. Thanks, Harry. Obviously, I agree with Terry in terms of it has to be that value added. But for me and it's a really important start point, actually, that I think there's a perception out there that open- source intelligence is what you scrape off the internet and false persona and all this sort of clever quite dark stuff that happens. But for me, open- source intelligence is not that actually. I guess I would say it's the source of value added that any member of the population or public could be able to legitimately and legally access that information, whether that's commercial or it is just open- source. That could be anything from commercial imagery in all its forms, news reports, AIS data, that sort of thing, academic articles, brochures, interviews, even legal intercept. That's one of the beauties, but also one of the pressures that I'm sure we'll talk about in a moment of open- source intelligence. It is any form of information that can be used to support an assessment. It's that open, accessible and legal part that's important for me.

Harry Kemsley: I agree with all that. I had the privilege recently of walking along Hadrian's Wall, a very, very old structure that the Romans put in in the back end of their time here. I can tell you, it occurred to me that what we talk about as open- source intelligence is kind of traditional intelligence. It was around for a lot longer than the explicit stuff that we're talking about now in a classified environment. But I'll move on for fear that I stopped boring you about my love of ancient history. Let's move on then with that definition of the contextualized validated information that we can pull from public sources and indeed commercial sources and bring them to the utility of the analyst in the national security and defense arena. I'm going to put to one side for now, a really important topic about the other parts of government that also need access to open- source, because we're here to talk about defense intelligence. I'm going to be a little bit precious to focus in there. But let's not overlook the fact that it's not a unique requirement for defense or national security. It is actually a hand government requirement, as much as it is for many of the commercial parties that support them. The need, let's move on to the need for open- source. Why do we care so much that people aren't using open- source as much as we think they could or should? Again, Terry, I'm going to start with you on that. What's the need? What's the driver here for us to be moving that way?

Terry Busch: Sure. Absolutely. I think from a defense intelligence perspective, if we go back in time and we look at preparation of the battlespace, this has been a difficult process and it's a time honored process. And every commander has always known that they're seeing some version of reality but not the complete depiction of reality. It was expensive and hard and it took us a long time to put those plans and that understanding of the battlespace together. And then about 10 years ago, there was this big data revolution that sort of happened overnight. We started using Google Maps and Wikipedia and open- source information became part of your consumer life very, very quickly. You can go to almost any place in the world and there are data deserts that are important to us where this isn't happening. In C, every function and every building and every piece of infrastructure. We had to transition from this older way of knowing which took us forever, to this new way of knowing. Now we just use as analogy is think how long it took us to collect and understand a bridge 25 or 30 years ago, versus today. Every bridge in the world is in an open- source database and we can confirm it. That's a transition that we're bringing into defense intelligence today. We're still in that process.

Harry Kemsley: So that battlespace preparation, Terry, which I recognize is one way which we really should be using open- source. I think the point that we might come back to again, if we have time, although I'm sure we'll burn that quickly, is that technology has unlocked the door to that publicly available information. I do distinctly remember the time when, in my own roles in military life, I knew the information was out there somewhere, I just couldn't get to it. I really, really knew it was out there. The known- unknown or the unknown- known, it's out there somewhere, but I just couldn't get it. Maybe that's the change in the last 10 years, Terry. The technology has unlocked that door, allowed us into the Aladdin's cave of open- source.

Terry Busch: Right. It's this collection and an ease of access and use, all came together in the last 10 years. But they came into a setting with legacy technology and very well established business processes that are still changing and adapting to that.

Harry Kemsley: Yeah. Shawn, over to you. What about the drivers here for the power of open- source and why it needs to be brought forward?

Sean Corbett: Again, I'd agree everything that Terry said. I don't think there's a particularly new requirement for OSINT. I think it's always been there. But I think the availability of it and the what you can do with it has made it more powerful now. But if you step back once or strategically looking about the state of the world today, I just wrote down just a few of the challenges that the poor or the intelligence community is faced with now. It's always been over tasked and under- resourced. But you've got a Russian resurgence, you've got the rise of China as a regional or global competitor. You've got the rogue states that are becoming more rogue, you could argue. You've got pressure on supply chains caused by COVID and other elements. You've got violent extremist organizations coming up and then you've got increasing environmental threats. That is a huge amount for any organization and even multiple organizations to look at. But there is a requirement to do so obviously, because that threat is so pervasive now. If you think there's 1, 200 petabytes plus, it's probably a lot more than that now on the internet available, it would almost be criminal not to use that to not just fill in the gaps actually. But I think in many ways, the sort of less urgent, the less exquisite requirements, operational requirements to actually fill that gap with the stuff that's clearly available to everybody. But I do go back to it's got to be off... We'll go on to talk to this, no doubt, it's got to be of use and helpful for that poor analyst that sat there churning out lots of lots of products with not all the information they need and they have to have the tools to be able to do so.

Harry Kemsley: I do want to spend just a second though talking about where OSINT could or should be used, its power. I'll give you a couple of examples from my experience working at Janes and kind of things that we have seen ourselves do, because I think it might open up a slightly different channel of this conversation. I know that Jane's information can be used to sort of prime the pump. It's a foundational intelligence that allows people to see things quickly, find things quickly and move on to them more detailed analysis. That's relatively straightforward. But they also see things like indicators and warnings coming out of the open- source environment. I remember with horror, the scenes around the Boston Marathon. I'm an ex- runner and I do remember the scenes that you saw on the TV of all the telephones that were up in the air videoing, recording what was going on immediately after the incident occurred. To me, that was a bit of a revelation that in the open- source environment at that moment is where the understanding of what was happening in the" battlespace" of the horrors after that event, it wasn't in the minds, it wasn't on the systems of inaudible and the other systems around it. It just wasn't yet. It would be soon, but it's in the open- source. That indicator of warnings piece comes through. Then there's also the plausible deniability. If an organization in the open- source environment, and there are quite a few of them now, can put something into public domain that allows an agency to talk about it that otherwise might not be able to, that's also very useful. This capability primer, this indicators of warnings. And also there's plausible deniability, just some examples of where open- source I think has real utility in the defense environment. Are there any others? Or do you have any experience of any of those, Terry, that you'd like to share with us?

Terry Busch: Yeah, I think there's definitely the breaking news cycle of it. The ability to look into the world in practically real time and understand breaking events. And strategic surprise and defending against strategic surprise is a very hard thing for the intelligence communities of the world to do because it's generally popping up in an unwarned situation. That's the first. The second is contextualization of what's going on and adding to our understanding of any environment. It's not just the window of the world, it's looking at the world over time to understand those patterns, trends, enablers and anomalies, so that we can get ahead of some of these cycles, that we better comprehend all the forces. And that ability to see comprehensively I think, is new in those realms. It's not just INW, it's the entirety of your assessment, of your stare. Very important that we incorporate OSINT into those processes as well.

Harry Kemsley: Yeah, Shawn. Any additional thoughts on that?

Sean Corbett: Yeah, I would focus at this stage because this has got to be an incremental... You've got to start increasing the trust in the open- source intelligence. And I would look at foundation intelligence as being really the bread and butter in which OSINT right now can play its role. It's not that stuff that needs to be right now, this has got to be brief for senior decision makers on which to base and although exactly as Terry said, there's legitimate use for that because you can just see with your own eyes if there's a press reporting going on. But beyond that, I think that I've got the more boring stuff, but it's not actually, it's the really foundational stuff, order of battles, that sort of thing. Not even that actually, if you're doing something like a noncombatant evacuation operation or disaster relief, you're going to want to know about the road networks, you're going to want to know about port facilities, about airfields, etc, etc. All that is out there and it can be easily gained. Now, by concentrating people on the open- source side of doing that or even letting commercial providers do that for you, that leaves the analysts to do the really exquisite, so- what piece that they are very, very good at but get frustrated, because they don't have the opportunity to do.

Harry Kemsley: Just to kind of put a punctuation mark in the conversation so far, we've agreed that open- source is a pervasive resource that we really must use. It would be criminal not to use it given its depth, breadth and potential utility. We have talked about some of the utilities that we have seen, we can see for the use of open- source in the defense. I'm going to pick up on a question that's come into the Q& A box from Paul Benfield about the bias of the IC toward the classified environment and that's what I wanted to go to anyway. I'm going to use your question Paul as a bridge, the segue into that. I agree that from my own experience years ago, if somebody brought in, " open- source intelligence" for which most people thought that meant a quick read of some social media, which isn't open- source of intelligence in itself, it's part of the huge pot that's out there, there was a tendency to say, " Yeah, that's not very well. I can't believe that. I'm going to go with the signals intelligence, I'm going to go with the inaudible that I've got and I'm going to work with that." So what are the barriers preventing the shift? Where do we have to start chipping away at the wall between where we are today, where we want to be. Shawn, I come to you first, because I know you and I spoke about so many times and Terry, I'll come to you after that. What are the things we got to get right, in terms of the shifting focus towards the utility of open- source. Shawn?

Sean Corbett: I'll let Terry go and talk about more about the technical side and the actual practical limitations, but for me, the biggest thing is cultural. That starts with leadership recognizing the legitimate use of open- source intelligence. That's generally I think, started to happen now. So quite good articles out there where people that are either former ex senior IC or still serving saying, " Yeah, we need to use this." But of course, it doesn't stop there because you've got to get down a level and get to the analysts, the people that do all the really work and I always think about it, about what's in it for me. As you know, one of my main roles over when I was in DIA was to increase the amount of Five Eyes intelligence sharing. It was hard enough to get even quite senior people to see the legitimacy of other Five Eyes partners intelligence. This really frustrated me actually. There was a... Call it an institution arguments if you like, but it was more than that actually. So that was one. I think we're breaking through a lot of that actually, as people realize it. But there is still a philosophy that if it's not classified and it's not collected by exquisite sources, it's not intelligence. Of course, that's not surprising, because people have been indoctrinated but also trained to an extremely high level on this exquisite stuff. Of course, that's their bread and butter. Back to the okay, what's in it for me? Well, it's got to add value, obviously. And that goes with the validation, it goes with the assurance and all those other good things. But it's also got to be practically easy to do. If I'm an analyst again and I always think about the analyst sat there working away, I've got to look at 10 different systems, I've got deadlines coming out of my ears, all of which are competing. And the last thing I want is someone to say, " Oh, by the way. Go and have a look at that computer in the corner, because you've got shed loads of information out there that you also need to filter all the way through and find out the little nugget in there, if you can. Off you go." So it's not surprising and I'm not being rude about the cultural challenges, but they are really there. It comes with education, but it also comes with a degree of facilitation. And then tradecraft as well. I think trade craft for open- source intelligence is a very big area that I'll say we might talk about that later, but because it is different because of so many sources of it. But for me, the big one is cultural.

Harry Kemsley: Okay. Cultural and just going to touch very briefly, from my perspective, the tradecraft piece is one that I have seen firsthand. An individual staring at a signals intelligence feed gets a pretty much pre- formatted, intelligible response to the question they're asking around signals intelligence. That is a relatively slim piece of intelligence. Open- source if anything inaudible is incredibly broad, deep. It's very variable. It's changing continuously. I think that's a different tradecraft. I think that is a different skill set that's acquired. Actually, some of the time, I wonder whether analysts just don't know where to start with this morass of information that is just churning in front of them. Where do I begin? I think the tradecraft training that I've seen in recent time is starting to emerge that helps people navigate their way into open- source is something that again, needs to be taken up, it needs to be seen as an opportunity to really break open the box, called open- source and make it useful. Terry, let's step across then from the cultural thing that Shawn's described and I've added a piece there about tradecraft to the technology, the more technical barriers we've got to overcome. Terry.

Terry Busch: Sure. This is a interesting story. If we look in the last five years, in the United States, our foundational intelligence holdings have increased 25 fold. 95- 98% of that is open- source information. That is the significant change. To some degree, the big data revolution is already happening. The technical side of that is you're right, you're overwhelmed. The first thing is we have to let the machine take on some of the burden of watching the rest of the world because we don't have enough analysts to cover down everything in the world, the human labor involved to watch every significant event in the world is too great. Using that data, we can say, " Hey, look. I'm not looking at this particular place in the world but I'm going to let that collection continue." I've had enough say in that collection that I can trust. We can put enough into that automation to say, " Hey, we have the analysts thoughts and rules into that collection," so that when the endless does pivot to that place in the world, the data is already there. We can start by saving time through technology and then building out a better understanding without a lot of labor. The second part of that, though, is in that transition from that cultural trade craft into the machine. That's where we're very early in the story. There are great capabilities developed out there that do some of that today. But at the end of the day, there is a shift from a very analog trade craft that we're used to and it's time honored to a quantitative trade craft, which we're at the very beginning of that story. We begin, as we've mentioned here, by using the technology to create ease of use. I can find the data and ask questions of the data better, faster, easier than I used to. That's followed by more advanced things. Once I've learned how to do that, I can move into my hypothesis making, my traditional intelligence process. I can use ACH on a computer. I don't have to formulate that in the best supercomputer ever made. And then I can get to very advanced technologies. The key to that though, is all that data at the beginning has to be well- organized, well- constructed, well vetted and well placed so that it can enable all these other missions.

Harry Kemsley: Yeah. And that trade craft you're alluding to there, it's supported by technology is a piece that we're going to come on to. I can see some questions coming in about that, which we're going to talk into. Let's go to one of those right now. We've just talked about technology and how it's starting to unlock things. A question from Madeline, " How will AI and data analytics assist in shortening that time?" You mentioned time in your piece just there, Terry, between collection of OSINT and turning it into actionable intelligence. Once I've asked you that question, Terry, Shawn, I'm going to ask you on the cultural side is, so why would that not be a good thing? Why would an analyst just sort of throw up their hands in horror, if they saw the AI starting to do that? Terry, give us your thoughts.

Terry Busch: Yeah, sure. To me, it's a tale of two cities. The first city is the data management city. The AI can collect, organize, control, assess confidence for the data and we're doing that right now. We can use very advanced concepts to get to very high statistical significance in our data collection using AI. We use computer vision, we use machine learning and we use multiple points of inputs to say, " Hey, I'm not going to trust a single set of data. I'm going to use a multiplicity set of data to get the confidence I need." We use UCA adages. If Apple, MasterCard and Google all say that that's a restaurant, it's highly probable that that's a restaurant. We need to get to the same place in our world. The second part of that is the analytical side and that's where the most potential is. To me, that is beginning... A, data has to be well managed. It's 80% of the job, if anybody ever says it's not that, take your money and go home. But the analytical side is the promise. What I would say is treat that as a stepping stone towards moving your business processes. Start with asking Boolean questions of the data. Start with descriptive and inferential statistics, don't run to AI, move into your understanding of that data. Because you're building trust, you're building confidence and very importantly, you're also slowly and very methodically bringing that trade craft that you already have into that automation process. I think there's where industry can provide us the greatest benefit to open- source and we're just touching on that. But we shouldn't go from zero to 60 on this. We should do this deliberately over time and there's some great, great, great examples of that happening.

Harry Kemsley: Can I redefine the AI to augmented intelligence, is that a more useful version of that?

Terry Busch: Yes. I do. I don't want to speak critically about AI but I worry that we're watering down the definition of AI. It's expanding into any sort of automated quantitative assessment, not the true meaning, which where's the computers learning and then making its own assessment. Many times we conflate AI to mean a great many things.

Harry Kemsley: Yeah, so let me now walk that scenario towards you Shawn. So you're back in your role as the Joint Force commander's J2 lead and intelligence is pouring in and it's been sifted through the machine augmented systems and it's presented to you. How are you feeling about that?

Sean Corbett: If only, if only. I have to say that I don't think we're there yet or we're nearly there yet actually. The Excel spreadsheet, sadly, is still king. It's all about trusting the data, of course it is. But it's all about trusting the data anyway when your intelligence inaudible and having that confidence in the data. A lot of people say, " Well, if it's open- source, how can you assure it? How's it validated, etc, etc?" Well, for me, it's just like any form of intelligence. If you look at human to even inaudible and I'm being blasphemous here, it's only as good as the source that comes from. For instance, if you talk to somebody in a covert manner, you're hearing what either they want you to hear, what they believe or what their perceptions are. It's the same with any form of intelligence at all. So how would you know its actual true? Well, you know that by testing and adjusting. If you start getting a load of rubbish or intercepting a load of rubbish, you know not to trust that. It's exactly the same with open- source. This is where the clever stuff, the artificial intelligence, I do agree, Terry, there's a lot talked about it, does come in, because that can start to cross- refer to different sorts of intelligence, it's all about making sure you've got two or three different sources that validate each other. For me, in that scenario that you gave me, I'd be absolutely delighted because what I want my clever analyst to be doing is doing that. You've heard me say this a million times, what if and answering the commander's question, not sifting through huge amounts of data of which 95% will be totally useless. Time is of the essence, as always, with these things and you've just got to get through it. So you end up giving assessments that are only partially validated anyway. Anything that can filter down that information that is more useful, saves your time, it gets you to concentrate on the actual exam question. That is going to take time and but it's also trust. There will be mistakes, but just as they are with analysis anyway. I don't buy this, you can't use OSINT because it's not validated, it might do a load of rubbish. That's the same with any intelligence, frankly.

Harry Kemsley: Yeah, the comment is made by Anthony on the chat room about the tripod method, finding at least three cooperating open- source data points is a method that they use in evaluating open- source information and I fully concur that. That ability to triangulate is so intrinsic to tradecraft and intelligence, why wouldn't it be useful in open- source. Terry?

Terry Busch: Yeah, I wholeheartedly agree. We live in a world today where it's acceptable for the human being to be fallible. The human being can make a mistake, but the machine cannot. I've fought this battle a lot. 99% confidence interval, but you still can't break through. And they'll use exceptionalism and other things. You call that a bridge, but it's a roller coaster. Yes. But out of the 2 million bridges, only five are wrong. I agree with you. The validation process to get to trust and there's many methods for doing this. I think that is one area the intelligence community does well, it's never relied on single source to make a judgment call. It doesn't matter which phase of intelligence you're in. If you're a targeter, if you're an analyst, if you're a inaudible person, you need that quality. The advantage in defense intelligence, is that. It is. Does a tripod have to come from three open- source methods or can one be GEOINT? Or can one be something new? There's a proliferation of data happening across what we would traditionally call INTs going on out there. And I think we should expand our horizons to look at that, because the volumes of... There are other independent signals we can use to get that validation yet still stan OSINT.

Sean Corbett: crosstalk Right, right. Sorry. inaudible Put into that just just to throw something in there is that some of the best assessments I've ever heard were low confidence ones. We don't know, but we've got some information. And then based on the analyst experience, knowledge, tradecraft, I guess, but I think this is going to happen. And those are the ones you really listen to actually, not because... I do think sometimes when he tells us words, we're too safe. I want to be so corroborated and so validated that I'm not going to put anything out there until that happens. You can understand that. If you get it wrong, the consequences are better than if you get it right late or are they, don't know on that one. I'm just going back to sorry... I'm going on now. But briefly, inaudible, say I wanted to do an air order battle, I could use commercial satellite imagery today that is good enough, very good enough to identify not just aircraft but specific types of aircraft. You can apply machine learning algorithms to that automatically, that will count your numbers of aircraft instantly as long as you've got the data and you've had the training data to train the algorithms in the first place. You can just keep that rolling along, day to day every day. That's when it can alert you to say, "Hang on, there's two less than the normally I hear." That is when the analyst gets engaged. Why is that? So that's the real power I see of... It's just an example I use in terms of the power of open- source information, that it just helps the analyst, doesn't replace the analyst, it helps them.

Harry Kemsley: I like the analogy used earlier, Terry, about start gently, get those Boolean searches out there, start to trust the data that's coming to you and then start to increase the analytical power you apply to it. I also like the idea, Shawn, that you've introduced there where you're enabling the analyst to get to the so what question so much more quickly with a much more complete and one hopes, more accurate data set for them to start doing that so what analysis on. If you use the four factors of information quality of timeliness, relevance, accuracy and completeness, well, I think I'm hearing is, with the right technology applied to the open- source environment, you get a more complete, relevant, accurate and therefore usable intelligence that gives you the context and thing to talk about. I am conscious of time because I know that we will run away with ourselves if we're not careful. I've got three more questions I want to bring in which are relevant to what we were just talking about there. Francis has asked about defense intelligence, as an all source, organization and discipline, has it become less relevant in this COVID related environment? All source of organization with discipline, has it become less relevant in the COVID related environment? Is it likely to be devalued in importance to senior policymakers and operators who can access OSINT easily and in real- time?

Sean Corbett: crosstalk Okay. Being a Brit, I want to address that one. Hi, Francis, it's great to see you. I don't think it's become less relevant, I think it's become less listened to perhaps. Why is that? Well, because everyone's an instant expert. We've all done it, we just google the answer to everything. The problem with that is that, of course, 50% plus probably in the significant more of that is not right. But this is where it comes down to the augmenting piece. If the DI and other intelligence organizations should be focusing on that which is unique that cannot be taken from open- source, but to an extent, validating that open- source as well and adding the extra value added to it. If you see something going on in the world, COVID is a great, a great example. Ebola, that was a really good one. We were really badly equipped to actually deal with that because it was an enemy we couldn't see, hear, feel or fight. It was all in the open- source domain. But we didn't have the information, we needed to actually come up with an assessment. Now, it took us a while, but we eventually got there. In terms of okay, so what from that? But if you take that, then add that to this sort of exquisite stuff, the collection, you very quickly then get really into the detail that might be the bit that's required for the policymakers and the decision makers in terms of where to use their resources most effectively. I can understand the frustration that everybody's an instant expert now. But it's the job of the intelligence community to make sure that we really or you, not me anymore, really are the experts and that layer of value added that open- source simply can't do for me.

Harry Kemsley: Just before I go to Terry, Francis, I noticed that I moved one of your questions to the answered section. That was my bad, I'll bring you back into the to- be answered section in a second. Terry, your views on this sidelining.

Terry Busch: Sure, absolutely. We've debated this for a long time pre- COVID is that there is enough data and there's enough apparatus in the world, that you can conduct some intelligence outside of the all source environment. We acknowledge that. What does the all- source environment bring? First is accountability. This is something that's very important to understand. If a fly by night open- source Intelligence capability, and some of them are fantastic, I follow them as well as you do, make a call and they're wrong and they had been wrong at times, there's no accountability. In our context, we have accountability. We are part of the government system. We have to make amends and we have to reinvent and this has happened to us. Intelligence failures occur and we reinvent processes and respond and do a better job as a result. This is something you can't throw money at. This is why it's inherently governmental. That said, there are functions and there are going to be functions that can be completely done in the OSINT side. I honestly believe that we should use OSINT as a basis for all of our intelligence research. So I'm with you on that. I don't think it diminishes it though. I do think that there are still certain things that we need to hold under the National Security tent no matter what the issue is that gives us an advantage. It's important that we always are assessing what those reasons are. And if they don't need to be held on their national security umbrella, then that's okay. Then we can move that out into open- source.

Harry Kemsley: Very good. I've got two great questions here and I want to go to the one I inadvertently put into the wrong section on the Q& A box. We'll come back to that one, Francis. William has asked to you, Terry, " What data architectures help with contextualizing the breaking news that you described with the foundational information found in something like Janes?" What are the architectures that are going to help that?

Terry Busch: We're moving very quickly through this. 10 years ago, plus, when we started, we were trying to collectivize the data, get into a singular understanding so you wouldn't be lost in perpetual search and get to that entity level, even though it was some sort of conflation of inputs to an object and get that out to you as fast as possible. We serviced ease of use by creating architectures that were focused on delivery at high speed. They were not focused on analysis. They were to get the information. Now, we're seeing we're moving into graph technologies, we're moving into other understandings of data, that A, do a lot of the heavy lifting for us. They're going to make those connections to help us do those hidden anomalous connections that we don't see ourselves and B, make it easier for us to ask questions in the data. So I would say we're advancing into graph and other higher order data architectures right now and it's its own subject area. That's a great thing for us right? As long as... I love all the progress being made there.

Harry Kemsley: Let me ask the question to you, Shawn. Do you have confidence in the ability of the UK Defense Intelligence, DI, to move properly into big data, AI explosion space? Do you exploitation space? Do you see that as something that the UK going to do successful anytime soon?

Sean Corbett: I think my answer would almost without being impertinent be equally applicable to the US although the US can scale so much better. I think there is still a procedural process bureaucracy that prevents that happening to its optimum now. I've been working around the sort of open- source AI world now for three, three and a half years and I'll offer my expertise for what it is to anybody. Nobody has approached me from my old colleagues. Now, that might be just because it's me. But the engagement is just not there. If you're lucky, you get five minutes at the sort of Innovation Center, have a chat and they think you're trying to sell them something. So they don't want to know. Now, that is changing, there's no question. Certainly the narrative and it's the same mistakes. You've got brilliant advocates like inaudible in the US that really get it and want to work with industry, but the mechanisms just aren't there yet to facilitate that. It's so usual for defense and government to go, " Right, we need to reinvent the wheel, do it ourselves." And getting to the, " Hang on, there's one that was created earlier. We'll just partner with that." We're just not there yet. That's not just in intelligence but it is a really big area, which has always been compartmented anyway. That is starting to happen. And it will be a slow evolution but it does get frustrated when there's just not the... There's the appetite there, there's the will, it's just not necessarily the mechanism yet I think is the way I would describe it. Of course, many people are like me, middle aged and slightly past it. We need the bright young things to be given their head to say, " What are you talking about? No, no, no, no, no, no. This is what you're doing. This is how you do it." I think we're trying to inculcate that as well. There's a wise debate on whether we are and can recruit the right sort of people to do that. This is where I do think partnering with industry is going to equally be important.

Harry Kemsley: I'm going to squeeze in a couple of more questions and as I do so on the second one, gents, I'm going to ask you to start to bring together for me your vision of what good looks like when OSINT is properly deployed and integrated. So the first question before we get to that, though, is from William, publicly available information, intelligence when it's assessed and relates to capability of managed attribution capabilities, helped certain organizations to remain productive during work from home periods. I think that's a very interesting statement and implied question. For me, watching the Janes' analysts overnight switch from being teams in offices, contributors sending in their staff to suddenly doing that from home in remote places was a very interesting and very fast transition and one that I think is actually quite an interesting topic in itself, but to the point, publicly available information related capability of managed attribution capabilities has helped certain organizations to remain productive during work from home periods. Your thoughts on that Terry, because that's going to be a factor that we're going to have to deal with for some time here.

Terry Busch: I've been thinking about this a lot for a project to have lately. So some of the problem has been Op Sec. It has been being secure at home. So the manager attribution systems are great, because it gives you a connection to the world with some anonymity. However, you're still compiling information and there's always a risk when you're outside the closed environment. So the opportunity there is, we can begin to understand the world better using OSINT. So the nice thing is, is if I'm studying a particular phenomenon out in the world for intelligence's sake, if the entirety of what I'm looking at is open- source, is it a surprise to anyone that we're studying these issues? Of course not. We have to get over that. There is an old paradigmatic way of thinking that is, if I'm looking at it, it's there for all, 100% classified. The fact that I'm looking at it makes it classified, when we know we're looking at these things. I think it's leveraging those sources as your basis of this and that's what we learn through this sort of going home manager attribution environment is we can do a lot of this research online. I still have security concerns, don't get me wrong. There's still things that give me butterflies in my stomach because as the guy who's collected a lot of this information, I sort of know how it works in reverse. That said, it is a new paradigm. Let's just use OSINT as the basis of our understanding for that research.

Harry Kemsley: Very good. All right, we're into the last minute or two. So we're going to use the last a minute or two, leveraging off Simon's question about how we completed the IPA at where OSINT can and can't should or shouldn't be used. Where can we safely use it successfully? Indeed, on that route, Terry, I might ask you just to touch on the technology piece, again, about where we feel more comfortable or less comfortable. I think we talked about it, but by all means, reemphasize it. Using that as the platform, my question to you in turn, I'll start with you, Terry, is what does good look like when we've got OSINT to where it should be in your own words?

Terry Busch: Yeah, I think good looks like to us is that when we are seeing dynamically a level of comprehension of what's going on, that is ethical and responsible and legal, but it is giving us a dynamic view of what we're studying. I think we are not even close to what's there and we have to continually reassess what is responsible if we clearly want to make sure that we're doing this in an appropriate way. But the level of comprehension that we need to get to, I think we're 20% of the way there at this point in time. We will be disrupted by investments in technology that are yet to fully be realized, a small set world alone, if you look at the billions of investment in that market. What is going to happen when that market gives us complete persistence of staring at things in the world? To me, it is you can't make me escape my all source mindset. To me it is the combination of OSINT replicating and then advancing what we once considered traditional INTs into this fuse understanding. I think we're just at the beginning of that story.

Harry Kemsley: Yeah. Thanks, Terry. Shawn.

Sean Corbett: Yeah, the key for me, as Terry alluded to, I think, actually is the ability to integrate the filtered the right open- source into the rest of the intelligence community, however difficult that is. This is one of the real practical constraints. Even if you have to lift a air gap and put it in there, you've got to treat all the information in the round. You can't just ignore bits, because then you get both conscious and unconscious bias. That leads back into what you're talking about in terms of working from home. That's about training and education. You can put security protocols on a computer but you've got to actually behave and use them in the right way. But you've also got to make sure that you are looking at the right things in the right way and not unintentionally as everybody does, including the IC to going to that source either because it's easy, because it's accessible. But you've got to... For the success for the IC is to bring it all into one place and to be able to manipulate it together. That's another discussion about standards, data standards and all the rest that we done for today.

Harry Kemsley: Right. Well, because we have overrun, I'm obliged to say sadly, that we've brought this conversation to a close but as we said, before we started, gents, I knew this would go for as long as it wanted to, it could have been hours. So let me first of all, thank the people who have joined us for this conversation and answer some really good questions. Terry, thank you very, very much for bringing your expertise and your outstanding insights to this. Shawn similarly, as ever forthright, straight to the point and accurate. Thank you both. Bye bye.

Sean Corbett: Thank you.

Terry Busch: Thank you.

Speaker 1: Thanks for joining us this week on the World of Intelligence. Make sure to visit our website janes. com/ podcast, where you can subscribe to the show on Apple Podcasts, Spotify or Google Podcasts so you'll never miss an episode. Uncover the threat landscape with Assured, an interconnected threat intelligence from Janes covering military capabilities, terrorism and insurgency, country risk and CBRN. Support your threat and capability assessments and enhance your situational awareness with Jane's Threat Intelligence Solutions. Find out more at janes. com/ threat.


In this podcast episode, Harry Kemsley OBE and Sean Corbett CB MBE talk to Terry Busch, Executive Advisor at Capax Analytics and Former Chief Technology Officer for the DIA's High Priority Machine-Assisted Rapid Repository Program (MARS).

In this discussion we cover the increasing relevance and utility of Open-Source Intelligence (OSINT) in support of the defence and security community. Specifically, the expert panel discuss challenges and opportunities of incorporating OSINT into the defence intelligence environment.

Today's Host

Guest Thumbnail

Harry Kemsley

|President of Government & National Security, Janes

Today's Guests

Guest Thumbnail

Terry Busch

|Executive Advisor at Capax Analytics
Guest Thumbnail

AVM (ret’d) Sean Corbett CB MBE MA, RAF

|CEO and Founder IntSight Global Limited