In this episode, Tom discusses the future of driverless cars, the rise of digital twins, the outcome economy, and the Industrial Internet of Things (IIoT) with best-selling author and industry pundit Joe Barkai.
Joe Barkai is an industry analyst and consultant, public speaker, blogger, and the author of The Outcome Economy: How the Industrial Internet of Things is Transforming Every Business. JoeBarkai.com
"TK: [Music] This week on Foresight Radio, driverless cars, digital twins in the industrial internet of things with best-selling author and industry pundit Joe Barkai. I’m your host, Tom Koulopoulos and Foresight Radio is brought to you by Wasabi Technologies. Find out more about Wasabi at Wasabi.com.
Joe, one of the things that we hear a lot about lately is the internet of things, but you focus more specifically on the industrial internet of things. Talk a bit about that and how that differs from what we call the IoT, the internet of things.
JB: Fundamentally, they’re not different. Fundamentally, we talk about pieces of assets or pieces of information sources that provide more information you can aggregate, you can analyze, you can draw conclusions. The reason why we want to preface this with the industry is because, A, it’s in the industrial setting, that is usually pieces of capital equipment or cars or medical equipment, but also to focus there on the outcome. I always want to know about who cares, why is this important, why is the IoT important, and to whom and how, and it’s easier to explain those to create these outcome-based scenarios in industrial settings, but this is not to say there’s no value to personal internet of things.
TK: It all creates data, obviously. The industrial internet of things involves much more data created by machines that’s consumed by machines. Give us an example of an IIoT that maybe helps to better understand what the industrial internet of things is and how it works and how it creates value.
JB: One example that I really like and actually the book I wrote on IoT opens with that, the use case is the Renfe high-speed rail in Spain. If you buy a ticket to ride on a train between Malaga, Barcelona, and Madrid, you actually buy a contract from Renfe, the initial train system that provides you with the guarantee of on-time arrival. If the train does not arrive within 15 minutes of the designated time, you get all your money back. It’s a very significant contract. No wonder everybody wants to take the train. Now, when you think about it - this is a losing proposition. Trains are always late 15 minutes, all trains, certainly here in Boston.
TK: How would we make money in the US with that model?
JB: Exactly. I just came here by train. It was only five or seven minutes late. What Renfe, the Spanish rail system did was they contracted Siemens, the provider of the train and the maintainer of the cars and the tracks. They created a contract with them to maintain the trail and provide availability. This is a classic paradigm of IoT. You have two contracts, one between Renfe and the passengers, the other one between Renfe and Siemens. Each contract is phrased very, very differently; they’re phrased in terms of different outcomes. One is on-time arrival, the other one is off-time. Yet the two contracts have to work together to provide the value, to provide the outcome. This really ties to the other observation, which is really to get value from industrial IoT data. We have to consider the human element.
Many organizations are totally myopic about their products. When the product leaves the door, gets installed, sold, what have you, they lose sight of the product. They don’t know how it operates. They don’t know what outcome it provides. They don’t know how users interact with it. What they do know, it’s usually that loose, warranty claim, complaints, and so on. The IoT provides opportunity to understand the machine but also the users. Often when users of equipment run into a problem, they don’t usually call the number the first time because they don’t want to be talked down at, they don’t want to be offended. They don’t want to be put on hold, et cetera. It’s surprising how often they go to the internet to look for advice and it’s very, very strange how they get advice from strangers. They download patches from total foreigners they never met before, but they trust them more than they trust the support. In the process of doing it, they don’t miss an opportunity of course to give advice, “The product sucks, the service is no good, I don’t like your company.” This information is also out there as information source for the organization to make better decision. It’s really now humans become sensors.
TK: The industrial internet of things is much more expansive than just machines talking to other machines, in other words?
JB: Absolutely, and I think that if a company is thinking about an IoT strategy and all they care about is instrumenting information assets, they’re missing a huge piece. What about the users? What about the environment? What about these instruments talk to another instrument? It’s really a more complex ecosystem that we never had an opportunity to look into. Now, we do.
TK: Your book, The Outcome Economy talks about two very important things that I want to touch on here. The first is that we now have the ability to sense our customers in a way we never could before. In the past, we would do a focus group, and the focus group would tell us what we wanted to hear in many cases. Today, if I have a Tesla, that Tesla is communicating enormous amounts of information about my behavior back to Tesla, so you have an ability to know me at a level that was incomprehensible just five or 10 years ago. The second is what that then does, again, going back to the point of your book, it creates an outcome-driven business model where I will pay you based on the performance, the degree to which you are actually able to do what I anticipate or expect that you can do with that product or service. Could you talk a bit about that shift from the focus group to a sensor-based customer service/customer support model? Then talk a bit more about the outcome economy and what that is and how that works.
JB: We still do a lot of focus groups, and we try to get an understanding of the market and the customers but in reality, these are very difficult to do because it’s very hard to get the right sample of people. They tend to be long and expensive efforts, and more importantly, even if you wanted to do that, from the time you conducted the focus group until you actually use the data and release the product, in a case of a car, it could be five years. In five years, customers have changed, competitions have changed, and technology arrived, and you don’t see it. I often argue that the whole notion of the innovation funnel, if you implement it as prescribed, and most companies do that, it’s really wrong. It’s really false or flawed I should say, because you lose connection with what’s happening out there. Now you have the ability, using this Tesla example, to have an ongoing focus group. It’s really the best ever focus group you can get because customer too has changed, whether it’s because of technology, because of competition, because they just aged. A young person buys a Tesla. Now he has a family, a very different approach and this gives you the ability to be continuous with your hand on the pulse.
TK: It’s a real-time focus group in some way?
JB: It’s real time.
TK: A focus group of one in real time.
JB: At the same time, you look at the data from all of them, so you actually have a better focus group because now you can segregate it much more precisely by age, by geography, by demographic, by weather pattern, by location.
TK: Where is that being done? Tesla is the example that comes to mind because it is one of the most connected cars today. Are there other examples of where that’s being done today?
JB: Other makers are doing it but they do it very sporadically. It’s too dependent upon initially from an individual product line. It is not too propagated to become a mindset, a culture. We don’t have a culture of IoT. We have technology and we have users. We don’t have a culture of IoT.
TK: Let’s come back to how that changes the business model and how culture plays a role in that, but I first of all want to talk about the outcome economy. It’s a wonderful term. I love it. However, it creates a lot of risk for a provider of a service or a product because now you’re on the hook, as in your railway example, of delivering a service within a certain very strict set of parameters. How many folks are actually doing that? How many providers of services are doing that?
JB: It’s getting more and more common, but it’s true that usually it’s the large industrial organization conglomerates. It’s not common small product/small companies. It will change. It has to change because you can no longer compete on price alone. It’s getting harder to compete on features and functions, especially if you’re looking at advanced economies versus off-shoring and low-cost manufacturing. You really need to provide better, more comprehensive service. It’s really moving from products and features and functions and you said it yourself earlier: It’s the outcome. What is the value that the product provides and to whom? Also, once you start looking at this as an ecosystem, your product becomes a platform. Renfe could potentially put more services on top of their IoT platform. You become now an orchestrator of services, not only provider of seats in a car.
TK: The example in the aviation industry that I heard you talk about for jet aircraft or engines and what’s called “time on the wing”, the actual time that that engine is in service, brings up this next question. Who finally owns the engine, and how do you charge for the service of the engine? Do you charge for it by the air mile, by the hour, as opposed to buying the engine outright or leasing it?
JB: You charge it by air time or lift cars. What was the availability? What was the uptime? Typically, the Rolls Royce or GE or Pratt will own the engine. It’s not very dissimilar from leasing, but you measure the performance based on uptime.
TK: You talk a lot about this term “digital twin” which I’m fascinated by. Define it for us. What is digital twin? What’s the easiest way to understand what a digital twin is?
JB: The easiest way to look at a digital twin is a digital model or replica of a physical product. For example, we’re talking about a car, so the digital twin would be the 3D model and all the other pieces of information that describe and represent the car. Now, why is this important other than it’s very cool, because you can look at stuff in real time? The ability then now to have better insight as far as the performance, you can understand and simulate behaviors. You can test different scenarios and say what if and you don’t really need to be there.
The car sample perhaps is too complex, but look at a wind turbine which is slightly simpler. If you have a model of a wind turbine running digitally in the back office, you can then really observe behavior. You can understand the impact of different changes in the field and draw conclusions. You can again simulate. You can test the impact of changes on the digital model and then move it down to the field, but saying that is not enough because to me, the digital twin is not yet another model. It’s not like there’s a 3D model of the turbine and then there’s a manufacturing build of material, maybe there’s service. To me, it’s really the ability to follow the life cycle of the turbine from cradle to grave and continuously evolve the digital model. I want them all to evolve and change through this IoT. I want them all to, for example, respond to wear and tear to the actual wear and tear of the equipment. I want it to be more sensitive to weather conditions, so you throw in simulation on top of 3D model.
The main conversations on the digital twin are missing it. I hear too many people talk about digital twin and you can substitute twin for model, digital model, 3D CAD model, simulation model. No, these are missing the point. The point is I want this to replicate the entire life cycle of the equipment.
TK: The volumes of data required for it, Joe, as calling a digital twin, are difficult for us to fully understand or appreciate today. One example I often use is that of an autonomous vehicle, AVs today, even Waymo’s autonomous driverless car, for example, will capture about 11 terabytes each day. Typically, you can count on AV to capture somewhere between 2 and 20 terabytes to be fully autonomous. It’s going to be closer to the high-end of that range, but let’s take the two terabytes to be conservative. Multiply that out by 365 days, you get about 7.3 petabytes of information generated yearly by just one autonomous vehicle. What that means in today’s terms is that it would cost about $300,000.00 to $400,000.00 to store all that data in the clouds. Your Tesla Model 3, which goes from $30,000.00 to $40,000.00, now suddenly requires 10 times as much investment to store the data.
You can see how today digital twins are making a lot of sense, in terms of how we demonstrate the capability, but to scale digital twins in a way that will be required for industrial applications creates an entirely new set of issues and challenges from the standpoint of where all that data will be stored. We have metaphors that we’re accustomed to using. When you talk about digital twin, the first thing that pops to my mind is a simulator, but as you say, they’re so much more than simulation because a simulator is built once and then it’s done. It doesn’t evolve with the actual physical object that it’s simulating, and that comes back to this issue of culture. Culture often ends up being the thing that most delays technological adoption. What are the cultural impediments when we talk about things like an outcome-based economy when we look at the industrial internet of things? Because all of these make perfect sense; they save money, they allow us to do all these wonderful new – build these wonderful new business models. What’s standing in the way of that, culturally?
JB: What’s standing in the way is really the need to change everything we do. We have to change the entire way of thinking, the entire way we approach product development. We need to be able to look at entire product life cycle earlier in the process. We need to build features and behaviors and maybe even architecture that reflects the new business model. One of the most commonly used examples of IoT is for service, so we connect pieces of equipment to remote areas so they alert us if there’s a problem because we know what it is, because we have the digital twin, we know the exact configuration or we can even figure out maybe the parts, et cetera, but there’s more to it because if I really want to improve my contract – you asked earlier about how do you make money out of this. I want my contract to be optimized to my needs. If I guarantee uptime, I want to know that I can reach it. Therefore, in the case of service, I want to optimize my service delivery. To optimize service delivery, the design, the architecture of the product should probably be larger field-replaceable parts because each part represents many failures. It’s a split up that came and undone. That makes sense from an uptime perspective. It makes little sense from an inventory perspective. It perhaps makes little sense because now I need two technicians to remove a large part.
Now we have a set of constraints with solutions that contradict each other, they with each other. The whole idea of designing for IoT is really considering all these downstream activities, and organizations are really not built to do that. Very often service requirements show up late in the design. I remember many years ago when I had a real job, I was actually working with a large heavy-equipment manufacturer. At some point, we started looking at these feasibility aspects of the equipment, and we said, “No, we need some changes,” and product management goes, “No, you can’t do that. Design is frozen. We have no more REM.” These were the days where REM was a scarce commodity. It was too late. Now, you release a product that perhaps meets all the functionary requirements but does not meet downstream activity requirements such as service. It requires the entire organization to change everything differently. It requires function that tends to be triggered later in the process to be brought forward. It requires us to invest upfront in order to reap the benefits down the road. It’s so foreign.
TK: I often talk about the fact that we’re moving out of an industrial era model/paradigm. The way we think about the problem and the potential solutions and the opportunities is tethered to that industrial era model. We’re building in the IIoT around an old model and the reality is the model itself has to change, but that is why we think about the model has to change as well.
JB: Exactly. We’re talking about a situation where you have additional twin that can mimic future activities, that can mimic behaviors so we can really optimize decisions short-term but also long-term, but the organization is not as flexible as the digital twin. It has a set of very rigid tools and the long time to change. Granted some areas are not as easy to change because of perhaps compliance and risk. I understand that, but we have to start this journey. We cannot just wait until this happens, and companies will take it sooner I think it will be the winners because they have greater flexibility, greater agility, they understand behavior sooner and then respond faster and better.
TK: Part of the answers that you decide experimenting with this seem if you don’t know exactly where you’re going to apply, it’s to build the competency and the knowledge base to be able to make those decisions later on.
JB: That’s a great point. This is really amplified by the fact that many features and functions knowing from the software, which means changes can be made very quickly. You mentioned Tesla. Tesla is known for updating functionality using software updates over the year. They’re not the first one to do that. People think that Tesla were the first ones. No, they were not, but they were the first ones that showed the value of doing that, and to bite the bullet as it were, to really go ahead and offer this flexibility. Now you have the ability to detect issues sooner, to issue fixes sooner, and instead of taking the car to the dealership, just issue it as a patch over the air. We have the mechanisms. Once again, it’s not the technology. It’s the culture that needs to change.
TK: A natural segue to what you’re just talking about was the topic of autonomous vehicles or AVs, and I love this topic because the reality is none of us know what that world’s going to look like when vehicles are fully autonomous. What I do know is that today we hold the AV technology on much higher standard than we do the human being because also a supply chain is threatened. It’s not just me as a driver that’s threatened. The supply chain, the dealers are threatened, and all of that threat is going to translate into obstacles to adopting AVs despite the enormous benefits ecologically, from a safety standpoint, the logistics standpoint that we can achieve. What are your thoughts on AVs and the direction they’re going in? What do you see as the evolution of that marketplace?
JB: There’s no doubt that AVs provide huge opportunities both for passengers and perhaps even more so to e-commerce and logistics and so on. Adoption is slow because like you said, we hold the cars at the higher standards and perhaps we’re not quite willing to change our behavior and our lifestyle. It’s a matter of time before we can ever convince that it has value. I do think that perhaps we’re going the wrong way about showing the value. We’re overemphasizing the risks and so on. The way perhaps some of these carmakers are going about proving the value perhaps is wrong. This certain value culture of “let’s try to break it and see what happens” is probably not the right mechanism when we’re dealing with cars and human lives and so on.
TK: Do you think that’s because of the risk factor and the potential incidents where cars would be involved with a human being and…?
JB: Yes, human fatalities and damage, but also there’s a little bit of “we can do it” type of attitude, which rubs some people the wrong way. I believe that the way to do it is really to find better managed applications. The example that I always use in this context is, you go to the airport and you want to rent a car unless you take an Uber, and you wait for this bus to take you to your car. Here comes large, noisy, polluting bus, and you and two other people are on the bus, they take you to your destination. Two minutes later another bus from another rental company. It’s wasteful. It’s really ridiculous. What we all have is autonomous cars in a semi-protected environment, a semi-dedicated lane. The route is well mapped. The chances of interruptions or surprises are very small. You have an app that summons you, summons this pod, takes you to the car place.
We need to find those well-managed special applications. Think about factory campuses, universities, hospitals, ports. This is the way to, A, fix the technology. B, it’s important they lower the cost of some of these technologies, especially in AVs, and develop trust on the part of the public. That is something we are not – we are moving in this direction, but I think we’re not doing enough. The other part, which goes back to your question about the value, I think there’s greater value in commercial applications, not for passenger cars. When it comes to passenger cars, it’s really not about you and I because we may like to drive them. It’s really about the elderly, people with disabilities. Those are the people who benefit, but for that, we need to get much higher level of cars and behavior.
I like to see more autonomous trucks, for example and they don’t have to be fully autonomous. How about a platoon of trucks? The lead truck is either autonomous or semi-autonomous or even not, even driven by a driver, but the following trucks should really be autonomous. Again, it’s okay to have the driver as the backup, but she or he doesn’t have to drive, so they’re not tired. The whole thing becomes so much more efficient and well managed, not to mention the fact that you save on fuel because if the trucks follow each other very closely, they benefit from reduced drag. Again, [Laughter] it’s so much of a Silicon Valley mentality. All we can fix, we will put it out there and don’t worry about it. I disagree.
TK: There are a lot of high-value, very specific narrow areas where we can apply the technology, in other words, where we can see benefit with relatively low risk. However, here’s one thing that I think about, and I’d love to get your opinion on this. When I consider the applications of autonomous vehicles, and I think about where globally they can have the greatest impact, we might very well see autonomous vehicles get greater traction, pardon the pun, outside of the US because there’s so many interest in the US that will prevent that, for example, trucking in unions.
There’s been an enormous amount of discussion around this point regarding autonomous vehicles and where they’re most likely to get, pardon the pun, traction in terms of their adoption. The fact is that there are a lot of regulatory hurdles and labor hurdles in the US that might very well create a momentary pause or a lapse in our ability to keep up with the rest of the world when it comes to AVs. Delphi Group did a research report last year in 2018 that looked at the future of transportation in the year 2050. In one of the conclusions in the report, which was co-authored with Ideafarms, an Indian consultancy, identified the opportunities outside of the US as being the more probable early adopters because you have a landscape there that is not as rigorously governed by transportation regulation and a need, which from an economic standpoint could present incredible opportunity to create efficiencies in an otherwise relatively dysfunctional transportation systems. It’s an interesting thought and one way in which the US may very well quickly lose ground if we don’t identify ways in which to make AVs a more viable alternative, especially in commercial applications such as trucking.
There are developing countries in the world that can easily leapfrog us where they don’t have those obstacles. In the same way that former Soviet countries that are Eastern Bloc countries leapfrogged us with telecommunications, the same thing can happen with autonomous vehicles. Do you think that’s a real threat? Is that a genuine concern?
JB: I think it’s possible especially because they can create an easier on-ramp, pardon the pun…
TK: This is an area ripe for puns.
JB: I know, and most of them are going to be terrible. For example, we in the US cannot have dedicated lane to autonomous vehicles. I’d love to see that because one of the things I dread the most is between now and the time we have many autonomous cars, I dread the time where we have the smart cars and stupid drivers on the road sharing the road.
TK: That’s my concern as well. We all have to go through it. I don’t see how we avoid that, so we will have an era – how long it will be, we’ll have to wait and see – when there will be smart cars and we can call them stupid or even normal drivers who don’t drive so well, both sharing the road. That’s a fascinating thing to observe and consider.
JB: I don’t know if it helps at all, but I think the time for that is a little bit further out. The average age of personal cars in the US is about 12 years, and they get older because baby boomers no longer buy as many cars. Cars are very reliable. There’s no need for new cars. Let’s pick a number. When 30% of the cars are smart and safe, we’ll start seeing the impact, in terms of reduced fatalities, lower congestions, and so on. The rate in which we refresh the fleet shows that it will take about 10 years to get to 30% penetration from the time cars are so capable. It’s a little bit out there. Back to our other geography’s example, I think they can build highways with dedicated lanes. They can guard those lanes to allow autonomous vehicles to truly be so much more effective and no other passenger cars. Again, autonomous public transit, I’d like to see more public transit on the roads.
Taking the segue a little bit, I also want to see better coordination among the different modalities. Today the different modalities, even within the bus system, they are not coordinated and one more time, it’s not about you and me because we can figure it out. We can Uber if we need to and so on. I’m thinking about the person, especially a woman that needs to leave home at 4:00 AM to reach a domestic-type job. She often takes one bus that arrives exactly two minutes after the previous bus had left. This is wasteful. Running buses on fixed route, fixed schedule is wasteful. I want to see a time where these different buses are coordinated. They don’t go to places where there’s nobody waiting. They wait for this person who just arrived on another bus to make the whole thing more efficient, less wasteful.
TK: The term that I’ve heard used recently is the IoV, the internet of vehicles. In fact, what we have is the internet of everything, but we want to categorize each of these, and we’ll talk about them in more specific and parochial terms. The internet of vehicles, explain that one to me a little bit. It sounds as though all vehicles should be talking to each other, but the reality is there are many cars that will not be able to talk to any other vehicle for quite some time because some of them don’t have the capacity, and I can’t see us retrofitting that capacity or mandating it into every vehicle. What is the IoV? Is that just the internet of vehicles among autonomous vehicles? Does Tesla have one IoV and GE have another IoV?
JB: The reality is that each car company has its own IoV, so I don’t use the term. I don’t see the value in saying it, and I think it’s misleading, and it’s pointing in the wrong direction. We talked earlier about the internet of things, how we create ecosystems of content and service providers that come together. IoT is about platform that allows you to onboard and offboard these services and content, so I don’t want to see IoVs. I want to see internet of everything even though it’s a bit of cliché term because I want to see cars talking to each other but also talking to the public transit, talking to the infrastructure, helping smart to diminish traffic, connect to smart road signals, and so on. It’s got to be everything has to be connected, everything and everyone because you are connected, I’m connected. I’m pulling in my Fitbit. We’re all connected. We need to benefit from the fact that everything and everybody is connected, which of course leads to the other question, “So, what happens to privacy?”
TK: Let’s get into that issue of privacy because one of the things that we’re doing no matter which “internet of” we’re talking about is giving out more data about ourselves, about where we are, what we’re doing, our behaviors in the moment. I saw a study that was just released by JD Power a few days ago that said 75% of all consumers who have connected homes, smart homes with Nest thermostats and what have you would be willing to share their information with their insurance company if they could get a better rate. What that says to me is on the one hand, we say we’re concerned about privacy but on the other we will willingly give it away if we can save a few dollars. Explain the dichotomy of that to me. Where are we going? Privacy, has it left the station? Is that something that we shouldn’t be worried about anymore? Where should we be vigilant, if we should?
JB: It’s all about its value. Am I willing to give some privacy knowing my information will not be abused, and do I get value from it? How do I trust the entity I give the data to, to use the data properly? It’s a very open question and every time there’s a failure, there’s a breach, there are concerns, but again, we have to, as a society and also in terms of policies and maybe this is the role for government is to help us understand this to create transparency to be able to manage to some extent.
I’m certainly not for government managing our lives, but this is one instance where perhaps it is the role of government to provide the policy and the guidelines, and by the way, it ties back to the question of AVs. Who is liable for an AV making the wrong the decision and causing damage? AI is very difficult to understand today. Some algorithms are totally unexplainable, so no one knows how they reach decision and the only way to move forward without risking too much is truly have an open conversation. We have to have an open discourse. We have to have transparency of algorithms, and this is, again, perhaps where we need government to do that because perhaps we cannot trust individual providers.
TK: I often use the example of GPS and how Ronald Regan decided to take the military GPS, which up until 1983 was only used by the military, and open it up, and the reason was Korean Airlines 007, which didn’t have GPS and couldn’t tell whether it was in Soviet airspace or not at the time. Today, none of us could imagine in the developed world being without a GPS in our pocket, in our purse. It’s a necessity. At the same time, we’re giving up a lot of data by using Google Maps or whatever GPS you happen to be using, but I don’t think anyone questions privacy in that case because it is such an essential part of modern life. Is that the direction we’re going in? Does this technology become so essential that we somehow don’t question the privacy piece of it anymore?
JB: I think so. I think that parts of it, yes, would become open and then public domain as it were, and the example of GPS is really a good one because if you think about Waze where our driving is helping others, it also creates chaos in neighborhoods that used to be quiet. [Laughter]
TK: Like every technology, we gain it. People are gaining GPS, and those in the neighborhood who don’t want traffic to go to their neighborhood are gaining ways by – for writing "