Gene Kim (00:00:00): This episode is brought to you by IT Revolution, whose mission is to help technology leader succeed through publishing and events. You're listening to the Idealcast with Jean Kim, brought to you by IT Revolution. If you haven't listened to the last two episodes, where I interviewed Steven Spear and played his 2019 and 2020 DevOps Enterprise Summit talks, I recommend you go listen to those first. In this episode, I do a follow up interview with Steve, where I got to ask him some even more basic questions, where I continue some of the explorations I started with Elizabeth Henderson. I asked, is it really true that leaders have fewer and fewer knobs to turn as they get higher up in the organization? And what are the implications? Why do we see this ever increasing complexity in the need for specialization in the world and what does it mean for leaders? Gene Kim (00:00:57): I asked him to share more personal accounts of his interactions and observations with the honorable Paul O'Neill, who was CEO of Alcoa, who built an incredible culture of safety and performance during his tenure. I asked why do we see this ever increasing amount of complexity in the world and need for specialization? And we talked about and explored further, the structure and dynamics of the famous MIT beer game. I'll just say now how much I am dazzled by the clarity of what he sees the world and I think it is so relevant for all leaders. Here's the interview. Steve, one of the most surprising things I've really learned and has been reinforced over the last couple of months is, in some ways the limitation of the leader to do what needs to be done. I think in a certain part of my career I thought that if one were king or queen, they would have control over everything. Given all the authority and broader the perspective, that they could make anything so. Gene Kim (00:01:59): However, what is obviously more true is that in general, the person best suited to making decisions is the person closest to the work. And they tend to make better decisions than distant authorities who have a more and more distant view of the actual problem. Can you validate that claim that leaders as they get more senior, they actually have fewer and fewer knobs to turn? Does that resonate with you? And if so, why? Dr. Steve Spear (00:02:24): I think some people like to think that we exist on a flat world. And in fact, we exist on a world with a very, very varied rugged topography. What I mean by this is that a flat world is homogeneous. And that no matter where you observe that world, you can have a full enough and complete enough understanding of how the world works. And with that, the complimentary full enough and complete enough reach into the world at any point to affect how it behaves. And of course, since we're all egocentric affect how it behaves in ways that accord with our desire and will. But the problem is the world is not flat and smooth and homogeneous, it's actually really heterogeneous with this very varied and rugged topography. Now here's the problem with that, is even if you can see a location, seeing it and understanding it may not give you the necessary insight to act effectively in another location because the other location is different. Dr. Steve Spear (00:03:31): So if that's the situation, if you need a varied action reaction to match the high degree of variety variation that's characteristic across topography, then rather than having one actor who's acting on the whole thing, what you need is actors everywhere tuning the localities based on what the local circumstances are. And here's the other part, tuning not only to what the local circumstance are, but consistent with what the system objectives might be. Now, you start thinking about this is a CPU problem, which is if we did, we don't, but if we did live in a flat topography, then the person who's looking down at the whole thing would have the computational space to figure out what to do in one location and then just spread that out. But as the topography gets varied as you go from one topography to two, you've doubled the computational problem, when you go from two to four, you haven't doubled it again. Well, maybe you've doubled it again, but it's only if the four nodes are now still independent. Dr. Steve Spear (00:04:43): Once the nodes become interdependent, you haven't doubled and doubled your computational problem, you've gone up orders of magnitude now. It's two to the end growing very, very quickly. And so when you start getting into a very, very topography, there's actually a very limited amount relative to everything that has to be known, that someone's sitting at a way elevated level can possibly know. And in fact, speaking of elevated levels, the metaphor and I've been working to see how this topography one rings up the issue of mountain. So it reminds me, there's an account in the Bible. So I'm going to go back to God on this one. So it's after Mount Sinai and Moses is sitting there and this long line of people waiting for his decisions. And his father in-law comes along and says, "Moses, what the hell are you doing? You got this whole line of people." And Moses said, "Well, they're coming to me with this problem and that problem and this thing and that thing." Dr. Steve Spear (00:05:44): And they're lined up all day and Moses is worn out and his father in-law says, "What the hell are you doing it by yourself? Why don't you teach some other people to solve local problems." And they get into this. And the Bible even says. He says, "What you really need to do is you got to get groups of 10 and a leader for the 10. And if that guy can't solve the problems for the 10, take 10 groups together so he can elevate the problem to the 100 and then 100 to the 1000 and Moses, you worry about the general stuff. You worry about the general stuff and as you get closer and closer, let those other folks worry about the specifics." And so you start seeing, Moses's father in-law, this guy Jethro, he's already thinking about this idea that you got a very, very topography, you got hundreds of thousands of people wandering around in the wilderness and what's going on in one tribe, the tribe of Reuben is different than what's going on in the tribe of Judah and Benjamin, who knows what the hell they're up to. Dr. Steve Spear (00:06:43): And so, you got to have this distributed decision making problem solving. And your job Moses is to set the tone, what are the terms by which you define a problem? What are the terms by which you accept the solution? What are the terms by which you move between problem and solution? That's your job. And let all these other folks behave to the local idiosyncrasies, according to the general guidance you're providing. And anyway, this goes way back in society. So, this is not a novel situation where we've invented things on silicon with aluminum and all of a sudden, whoa, the dude at the top can't see and do everything. It goes way back. It gets back to this basic issue that the world is a varied place, really varied. Dr. Steve Spear (00:07:28): And that's what gives it its beauty and its interest and all of that, but what it does mean is that that senior person doesn't have nearly the bandwidth nor the capacity nor the speed to determine what should happen everywhere. All they can do is create conditions in which people everywhere can arrive at a set of conclusions, decision, solutions, that are not only locally appropriate, but collectively consistent. And that's what the leaders got to do. Is create the environment where you get the local appropriateness in the collective consistency. Gene Kim (00:08:03): So you've spoken so beautifully about your experiences with Paul O'Neill, the recently departed Paul O'Neill and his work. But before we go there, let's take it from the Moses example to... Let's say the assembly line of Toyota. And one thing that you paint very vividly is that the person who is experiencing a problem is at that moment the world's expert in that problem. Dr. Steve Spear (00:08:25): Right. Gene Kim (00:08:25): Can you talk about how is it possible and to what degree it would be difficult for the first-line supervisor or the second-line supervisor to know what the right answer is just by experience? Can you even dispel the notion that the first and second-line supervisor just by experience would know the answer better than the person actually experiencing the problem or solving the problem the front line? Dr. Steve Spear (00:08:45): Right. So, here's the thing about problems. If someone had a complete and full understanding of a situation, there wouldn't be the problem because it's tautological, like a problem is a bad thing and if you had complete understanding of a situation, then you would make the right actions not to have the problem. The fact that you do have the problem reflects your lack of understanding. All right. So, the problem emerges. Now, at that moment who understand something? Well, it's the first person to experience the problem because they're aware of its existence. And perhaps they're aware of some of the subtlety of its presentation and so on. And so when they call for help, they may call for help from someone who's got a deeper experience and deeper expertise and that sort of thing. And their ability to gather data, process data arrive at a conclusion maybe better. But at that first moment, the person who understands the situation the best is the person who's stuck in it. Dr. Steve Spear (00:09:44): Now, at zero plus delta, the expert might start recognizing what's going on and say, "Whoa, I've got expertise experience to [inaudible 00:09:53] at zero." It's the person experiencing the problem. And Gene, just by way of metaphor here, look, we've all had issues where we've had ailments where we feel it's necessary to go to get clinical care. I've never really walked into a clinical office and said, "Well, I'm the expert on this or that ailments." I am the expert on the experience though. So when I went to a doctor because my ankle was bothering me, I knew more about how that ankle was bothering me than anybody else in the world. Now, why it was bothering me, I didn't really have a clue, which is why I went to the doctor for an examination, diagnosis and treatment plan. Later I had to go to physical therapy to put that treatment plan into practice for implementation as it were. But the experience, I was the world expert on the experience of having that pain. Dr. Steve Spear (00:10:45): And so when we end up with this, it's not that it's either or, but it's both and. When the problem is detected and experienced, that person has a lot to add to the conversation which is a compliment to and complemented by the person who has sort of the expertise and the kind of the basic thinking of the fundamental science to help with the diagnosis and help the generation of a treatment. Gene Kim (00:11:13): So, suppose you have the best first-line supervisor, the second-line supervisor, one can easily get to a point where given enough surface area, people having problems it would exceed their ability to solve on their own that at some point, you need distributed problem solving even at that first-line, let alone second, third, fourth line. Dr. Steve Spear (00:11:33): Right. Gene Kim (00:11:34): Does that resonate with you? Dr. Steve Spear (00:11:35): 100%. Gene, this is why you have organizations like Toyota which are so deliberate on developing their group leader core. And just to put it in context, the way this works is you've got your associates who are doing the wrench turning and the welding and all that stuff and there's a huge investment in getting them to be capable of seeing and solving problems that are very, very local. And you have team leads who their skill is developed to not only solve more sophisticated problems, but at least be capable assisters in solving problems by the associates within the Toyota system. The group leads, these are people who in the military would be a senior non-commissioned officers, chief petty officers, that kind of thing. 15, 20 years experience. The burden on them is not only to have really deep sophisticated problem solving skills for problems that have sources which are more complex and varied, but they also have to be great educators in terms of being able to develop the skill set of the team leaders and the associates for whom they're responsible. Dr. Steve Spear (00:12:44): And the reason Toyota feels that is such a source of power for their system and the reason they I think, worry quite a bit about group leaders or the process of developing group leaders, isn't only a source of power, but it consequently a source of vulnerability, is that in the absence of really skilled group leaders who have sort of experiential knowledge, technical knowledge and are really capable coaches, the system falls apart and then all of a sudden... Not all of a sudden, but you get to this point where if problems aren't being seen and solved at the local level and they're not being seen and solved at team lead level, and they blow through the group lead level, if that level is not well developed, then you start getting the managers having to deal with everything. You get back to the Moses problem, right? Fortunately, Moses took the advice of his father in-law, Jethro, when he started cascading his knowledge and skills down to much greater numbers. Dr. Steve Spear (00:13:45): And because of that, then he got to deal with the really big issues like succession to Joshua. That's the kind of a big issue. How to deal with the [inaudible 00:13:53] and the Amalek and that kind of thing. But those are system issues. If on the other hand, he had been unsuccessful in developing and cascading expertise and responsibility down to the 10s and the hundreds, he would have been back to where he started, which was a line of many, many people queued out front his tent every day waiting for him to tell them what to do. Gene Kim (00:14:19): Gene here, I have two thoughts I'd like to interject. One is around what managers know and don't know. I remember a time early in my career at Tripwire, around 1999. It was my first time being a first-line manager and then being a second-line manager. I remember one of the first challenges was shipping our first couple of commercial releases out the door. And the challenge of figuring out what we could ship by a certain date. Looking back, I remember thinking how many times I thought I knew better than the team about what could be shipped by a certain date. You could dismiss some of it as just the naive wishful thinking of a new inexperienced leader, but some of it went far beyond that. So two decades later, I'm thinking about just how much more the person working on a certain code base knows than the person who doesn't. In fact, on so many things, I'm the only person working on a particular set of code and if I leave it alone for four months, I basically don't know anything about it at all. Gene Kim (00:15:15): So, with that perspective, I'm a bit stunned by the notion that the first line manager thinks that they would know better how to do task estimation than the people doing the work. And then of course, a second-line manager thinking they might know better than the first-line manager or the team itself. This is belied by the fact that whenever it comes to actually implementing something in code or fixing defects due to that code and of course, estimating tasks related to that code really should be and can only be done by the people doing the implementation. In other words, due to the rugged topography of the problem, people best suited to solving a problem are the people closest to it. This validates a concept from the book, Team of Teams where you put authority close to where the information is as opposed to bringing information to where authority is. The second thing I want to interject here, is this amazing NPR Planet Money episode that recently aired called Summer School 2: Markets and Pickles. Gene Kim (00:16:12): In this episode, they interview numerous members of a national food bank, in this case is the person who was in charge of a food bank in Alaska. She was complaining that they got a truckload of pickles and it was because the centralized system sent it to them specifically because of the high shipping costs associated with shipping fresh fruits and vegetables. So, often they were stuck with things like truckloads of jars of pickles. And it turns out they weren't the only city having this problem. The head of the food bank in Idaho was complaining that they received a truckload of potatoes, which is pretty funny because potatoes are primarily grown in Idaho, so they always have a surplus of potatoes because they're donated to them all the time. Long story short, new management comes in who proposes creating an internal marketplace to better match supply and demand with the help of some academic economists. This causes some problems because many people who work in the food bank system feel that it is capitalism and markets, which failed them and are what created the need for food banks in the first place. Gene Kim (00:17:12): But even they are eventually won over by the internal system they create. Imagine a system like eBay, where each food bank is assigned a number of credits based upon the number of hungry people they need to feed. Each week when food becomes available, each food bank and then bid on the food or save it for the next round. It's an amazing story to hear about how everyone seems to benefit and how thoughtful people are about which foods they want to acquire and how they save and bid accordingly. Some foods like pickles that are undesirable actually have negative prices, signaling that no one wants it, but will take it if paid back in the appropriate number of credits. Cereal commands high prices because it has a long shelf life and is so versatile. Gene Kim (00:17:52): What caught my attention is how effective pricing signals enable the right allocation of resources, getting scarce resources to where they are needed most and how it's so dramatically outperformed the centralized planning model. They even said exactly what Steve just said. The maximization problem can be computed, but it's one that is in practice almost impractical because of the computational space required. They even use those words. The person at the top never has enough information to make all the right allocation decisions, that the internal topography is so complicated, so full of nooks and crannies of hidden information that things like internal marketplaces need to be created. This is a topic that is not totally clear in my head yet, but I think it's so important that I will be revisiting it again soon. When you go back to the interview, I'm going to ask Steve a question about Paul O'Neill and his achievements at Alcoa. I mentioned Paulo O'Neill a couple of times and I thought it warranted describing specifically what some of his breathtaking achievements were. Gene Kim (00:18:51): I'm going to read some of what I wrote in The DevOps Handbook, which extensively cites Dr. Spear's work in his book, High-Velocity Edge. Consider the following example that improved workplace safety at Alcoa, and aluminum manufacturer worth $7.8 billion in revenue in 1987. Aluminum manufacturing requires extremely high heat, high pressures and corrosive chemicals. In 1987, Alcoa had a frightening safety record with 2% of their 90,000 employee workforce being injured each year. That's seven injuries per day. When Paul O'Neill started as CEO, his first goal was to have zero injuries to employees, contractors and visitors. O'Neill wanted to be notified within 24 hours of anyone being injured on the job, not to punish but to ensure and promote that learnings were being generated and incorporated to create a safer workplace. Over the course of 10 years, Alcor reduced the injury rate by 95%. The reduction in injury rates allowed Alcoa to focus on smaller problems and weaker failure signals. Instead of notifying O'Neill only when injuries occurred, they started reporting any close calls as well. Gene Kim (00:19:58): By doing this, the improved workplace safety over the subsequent 20 years and they have one of the most enviable safety records in the industry. I quote Dr. Spear, he writes, "Alcoans gradually stopped working around the difficulties, inconveniences and impediments they experienced. Coping, firefighting and making do were gradually replaced throughout the organization by a dynamic of identifying opportunities for process and product improvement. As those opportunities were identified and the problems were investigated, the pockets of ignorance that they reflected were converted into nuggets of knowledge." This dynamic of learning helped not only increase safety, but also created competitive advantage, leading Alcoa to become one of the most admired organization in its industry. And during O'Neal's tenure, the market cap of Alcoa increased from $4 billion to $28 billion in 13 years, a 900% increase. All right, with that knowledge, let's go back to the interview. Gene Kim (00:20:57): You got firsthand observation of seeing this theory put it into practice with the honorable Paul O'Neill and what he did at Alcoa as he set out to create an environment where there were no workplace injuries. And you share some revelations about his philosophies and how he put that into practice. If the job of the leader isn't to make all the decisions and make all the right calls, he viewed his job as different than that. Can you talk more about what he viewed his job as and how it evidenced itself in terms of what you saw? Dr. Steve Spear (00:21:26): Yeah. I think Paul probably viewed his job was to set standards. Standards in terms of what the goals were, standards in terms of not only direction of the goals but distance for the goals, one of them being safety for everybody all the time and then setting the terms by which those goals were pursued. And I think in Paul's mind was that of, he clearly articulated. Not only in word and memo but in action and practice that the system goal was a perfection and that perfection should be relentlessly pursued and the means of pursuing it was through the regular energetic recognition of problems and the resolution of problems, they would keep moving in the right direction at the right pace. And that was too big an organization for him to have solutions to even a few individual problems, let alone all of them. But he could create the context in which the brainpower of the workforce was engaged and not just the brawn power of the workforce was engaged. Gene Kim (00:22:36): And just to chime in, what are some of the elements of that rugged topography? High heat, corrosive chemicals, huge, heavy pieces of round things moving around the floor, right? This is a very dangerous environment. Dr. Steve Spear (00:22:48): [inaudible 00:22:48] this dangerous environment, the thing I would offer is that it was dangerous in different ways, because you had plants doing different types of work, creating different types of product from the way, way upstream, operations of mining raw material through smelting it, refining it, et cetera, et cetera to giving it shape and then shipping it. So you had many, many different locations. Inside a site, you had many, many different processes, those processes were being run by many, many different people, because you got into the thousands and thousands of operators. And so the experience of one operator in one location was different from the operator across the aisle, let alone the operator across the world. It was a very, very rough varied jagged topography. Gene Kim (00:23:38): This podcast is brought to you by the 2020 DevOps Enterprise Summit, Las Vegas, which will again be a virtual conference due to the global pandemic. Holy cow, last month we held our first virtual conference, which was amazing. I'm so proud of the programming, we were able to deliver for you and I was so blown away by how many people said that it was even better than attending the physical conference. We are planning to put together an even more amazing program for you, which will be on October 13th to the 15th, that's only 11 weeks away. I will have some more exciting news to share with you in the weeks to come, but I am confident that it will be the best programming we've ever put together. For seven years, we've created the best learning experience for technology leaders, whether they're experience reports from large complex organization, talks from the experts we need or through the peer interaction that you'll only find at DevOps Enterprise. You can find more information about the conference @events.itrevolution.com. Dr. Steve Spear (00:24:34): If you start thinking about what Paul O'Neill did, is first he set context. He said, "What we're going to do is pursue perfection on workplace safety and as collateral benefit for doing that as we see and solve problems that present physical hazard and risk, we'll recognize what we don't understand about these very dangerous processes and as we come to a better recognition of what we don't understand and convert that into competency, not only will we move the needle significantly on safety, we'll move it significantly on yield, quality cost, et cetera." So, that was the context setting, which is, "These are our goals and the way we're going to achieve those goals is make sure that when we have problems, they're recognized, they're respected and they're addressed." All right, that's context. Paul did something else. And this I think, starts getting into the structure and the resulting dynamics. Is that Paul insisted that he find out what had gone severely wrong directly from the business unit president, not through a report, not through a publication, not through a memo, but through a call. Dr. Steve Spear (00:25:59): So, the first structural element was Paul being connected directly to his direct reports. And what that cascaded into is that if one of his direct reports could know quick enough, fast enough to tell Paul, let's say within the 24 hour limit that something serious had happened, he had to be connected directly to the shop floor. Now again, if you start thinking about the layers and layers and layers of authority and supervision in Alcoa, a business unit president couldn't be directed to all of the shop floor all of the time. Again, computational overload. So what he had to make sure is that he was connected directly to his direct reports and they were connected directly to their direct reports and so on and so on, all the way down to the associates on the shop floor being connected to and able to signal essentially immediately their immediate supervisors. Dr. Steve Spear (00:26:55): So in terms of structure, Paul recognized that he had to set up a structure which was not a horizontal connection of start to finish, but a vertical connection of start to finish. With the start being what's going on on the shop floor and the vertical connections up layer by layer by layer by layer until boom it connected to Paul O'Neill. Now, once that structure was in place, that he was connected to the... If he's the level end and then the end minus one, end minus two down to the shop floor, then that creates the opportunity for the dynamics he wanted. And the dynamics he wanted was the near immediate recognition that something had gone wrong and the near immediate response that recognition to figure out why, so that things can be made right. So anyway, creating the vertical connectivity, I think was the structural response in order to create the dynamic that he wanted of fast, frequent, uninterrupted, highly distributed problem recognition and problem resolution. Gene Kim (00:28:00): It is interesting to expound upon that. So not only did he set up a structure, but in the previous episode we were talking about certain structures that potentially suppress or extinguish weak signals. He put out that right, "If you don't get the response from your leadership chain, call me directly." In one of his talks, he said that this was during a time of immense tension between unions and plant managers. This was an unprecedented thing to do, especially for a CEO of a Fortune 500 company. It seems like this was another mechanism he created to make sure that he was getting the feedback into the system. Dr. Steve Spear (00:28:40): Yeah. Jean, it's a really interesting point. What he did was he put an Andon Cord on top of his Andon Cord. Gene Kim (00:28:46): Yeah. Dr. Steve Spear (00:28:46): And what I mean by this is that, ideally everything just works the way you want it to work and you don't have problems, you don't have to call things out. But given that's not the reality of our experiences, that nothing works the way you want it to work. So you have to call it out and you have to call it out early and often. So, how do you do that? Is that you give authority responsibility to the person who might be the first one to experience the problem to call it out to the person who's supporting them. And that person has authority responsibility to call it out to the person supporting them with the idea that if you call it out and it can't get resolved, you call it out to the next level of responsibility authority till they get it resolved. And eventually if it has to, if no one else can pull the resources into solving the problem, then Paul O'Neill has now got the authority but the responsibility to make things better. Dr. Steve Spear (00:29:41): Now, what his concern was, and it was a concern informed by the experiences people reported to him, is that that series of sort of Andon Cords, Andon Cords, pulls for help, wasn't working all the time. And so he instituted an overlay on top of that which was, "Look, if that whole escalation mechanism is not working, so the problems that you recognize aren't being resolved, then call me directly. And when you call me directly what you're telling me is that you've got a problem that's affecting work on the shop floor, the carrying of a heavy ingot manually rather than on a conveyor, et cetera, et cetera. But what you're really telling me is that the escalation process for the purpose of recognizing and resolving problems, the escalation process has broken and since I as CEO have taken responsibility for the escalation process, I need an Andon Cord myself to know that that process broke down and not just the technical process, which is introducing risk and hazard." Gene Kim (00:30:43): Right. Let's go to maybe one other such a memorable example of behaviors that Paul O'Neill exhibited that also reinforced the dynamics that he wanted. I think this showed up in the unicorn project, it was taken from your book and Paul O'Neil's talks was flying down to the... I think it was the plant in Arizona after an 18 year old employee was killed by a boom inside of some machinery and it seems like it was so clear as he tells the story that he felt his job was to make sure that everyone felt responsible, that not just remorse, but that everyone had a role in the conditions that led to the accident. I think his words were, "We killed that 18 year old kid. I killed him, we all killed him. We trained him in the procedures he was going through that led to his death." Can you talk about that? Dr. Steve Spear (00:31:43): Yes. So, this gets back to what can the senior leader do? And we kind of agreed on is that the senior leader can't solve all the problems everywhere all the time. There're just too many problems with too much subtlety and nuance for anyone to solve more than a few problems on any given day or week. Now, in terms of setting the context, what Paul tried to do was establish that it was the responsibility of leaders to set the direction and the distance. We're headed in the direction of safety and the distance is perfect safety and the means of getting there, which is the regular recognition and resolution of problems when there're still risks and hazards and before there're injuries and deaths. Now, the account you're sharing was from very early in his tenure as CEO, where he found out that a young man when the machine jammed, he jumped a protective barrier to unjam the machine. Well, it's a spring problem which is once you removed the obstacle, the machine which had tension loaded into it uncoiled and hit and killed him. Dr. Steve Spear (00:33:06): Now, what Paul did in that instance is said, "Look, our job as leaders is to set the context in which everyone else operates. And so if this guy thought the right thing to do was volt that fence and remove that obstacle, then that's the context we created for him, that that was the right thing to do. And if that's the context that inspired him and motivated him to do that thing which led to his death, then we killed him. We didn't hit him with the boom, but we had at most second order causality, because we created the context in which he thought the right thing to do was volt that fence, so that he could get hit in the head." Gene Kim (00:33:58): And what was the effect of that? Dr. Steve Spear (00:34:01): Well, again, this was very, very early on in his tenure. And you start thinking about how a young man dies due to an industrial accident in a hazardous situation. So what are the knee jerk explanations for that? So one is, "Well, you got to understand with all due respect, we do some dangerous work here. The stuff is heavy and hot and fast moving, et cetera, is dangerous here." Gene Kim (00:34:32): A human error, it was a technology failure. Those are the classic- Dr. Steve Spear (00:34:39): Yeah. And then, well, human error. We're just kind of spitting on the grave because not only is the guy dead, but now you're saying he's dead by his own fault. It wasn't a suicide. And so Paul gets in there and he says, "Look, everyone is doing all that appropriate remorse, "Oh, I feel terrible for him, I feel terrible for the family." Blah, blah, blah. And they should, I don't want to diminish that part. They should feel bad. Well, there was a loss and then the consequence suffering by those left behind. But what Paul did is he disallowed the conventional explanations. He disallowed they blame the victim and say, "Well, it was human error. Had he been smarter about doing that or not or more skillful. Had he pulled out the obstacle and ducked quicker, he'd still be alive." Dr. Steve Spear (00:35:27): So, it's kind of his fault. Let's blame the victim or you just blame circumstance. You yell at the universe. "Well, this is high hazard and what could anyone do?" And Paul very quickly took away those safe pat answers and said, "No, no, no, we created the conditions in which that young man thought he was doing the right thing. If it was human error, you know whose human error It was? It was our human error to convince him that that was the right thing to do." Gene Kim (00:36:00): Right. "And in fact, I think it's my fault." And he flew to that plant with the Alcoa executives, not just the local plant managers casting that responsibility throughout the team. Dr. Steve Spear (00:36:13): Yeah, that's right. Gene, I think that's a very generous way to think about it because it could have been Paul who shows up at the plant and says, "You bunch of jokers, you created the environment, you created the context, so it was your human error." But then by showing up and saying, "We killed him," not you killed him, right? "We killed him." What he's really saying is, "The context you created reflected the context I created for you. And so that I own this as CEO, as the most authoritative, most responsible person in this system I own the context." Gene Kim (00:36:52): Gene here. I've been reading about the safety culture at Alcoa, going on 10 years now. And it is amazing to see Paul O'Neal speak about them. I'm going to include two links to his videos of his talks that he gave over the years in the show notes. It is amazing to see to what extent he cared about these issues throughout his tenure at Alcoa. And one of the videos he even says what he is most proud of in his entire career, including his tenure as the Secretary of the treasury is creating a safe environment for all Alcoans, where they got fulfillment out of their daily work. I highly encourage you to watch his videos. And in the meantime, let's go back to Steve, as he continues discussing this topic. Dr. Steve Spear (00:37:34): Actually Gene, I'm sorry, but going back to our story about Moses and all, is that throughout all the accounts of the 40 years wandering around, he keeps coming back to every time the Hebrew people, the sons of Jacob, err and they're about to get punished, Moses says, "Whoa, whoa, hold on it's my fault. I obviously didn't teach them well. I was the one who enjoyed not one but two revelations on the top of the mountain, I'm the one who's had this direct communication with the godly spirit day in day out and yet, if there's a failure, it's my failure." So, we have cultural examples to say that those who are more senior should have responsibility for the failure those younger and more junior. And... Gene Kim (00:38:23): That are not delegated away. Dr. Steve Spear (00:38:27): [crosstalk 00:38:27] over said that. Gene Kim (00:38:28): Correct. Dr. Steve Spear (00:38:29): [inaudible 00:38:29] over said that. He said, you can delegate authority, but you can never ever delegate responsibility. Gene Kim (00:38:37): Brilliant. So, I think you've made a very logical and persuasive argument that says the person at the top can't make all decisions, can't have visibility into all the problems, can't be the one solving the problem. I think you make a very persuasive case for that. Let's go to the topography of the problem. If you look at the topography, whether it's at Alcoa or within the Toyota plant or the supplier network, it is varied and deep that it is hard to imagine its expert being several, let alone all of them. I love your example from your book about the 1950s Ford. In that era home mechanics could fix most of the issues that you'd find in the car, but now because of computerization and software and probably many other issues, there are fewer and fewer parts that could actually be repaired at home these days and maybe even fixed by the Service specialist at the dealership. We talked about in the last podcast episode in the 1950s and maybe through the 70s, in the healthcare system, there are only a handful of specialties, the most obvious being the doctor who's laying hands on the patient. Gene Kim (00:39:42): And now there are scores of specialties more being created every day. And in technology, we see the same thing happening. It used to be developers and operations, but now we have container, security, data, not just data, data in Kafka, data in relational databases and SQL. There are so many more specialties. So my first question is, why is this happening? Completely, there must be some benefits that we're getting through specialization, whether it's through automotive or healthcare. Can you talk about why this is happening to us? Dr. Steve Spear (00:40:12): Yeah. Look, Jean. I live in New England. And back in the day, Henry David Thoreau went out to Walden Pond and lived a simple life. And he was able to live a simple life out there and he could probably cultivate whatever food he ate through the spring, summer, fall and winter and mend his own clothes. But I'll tell you what he couldn't do that I can do, he could never eat an orange in February. I can. I can get a delicious orange in February. I can get a delicious orange in July. And the reason I bring this up is that no one grows an orange in New England that I'm aware of and certainly not in February. Dr. Steve Spear (00:40:52): Now, the reason I can get a delicious orange in February is because since Henry David Thoreau went out and lived in a cabin in Walden Pond, people have learned a lot about horticulture and growing oranges and transportation to move an orange from wherever it's grown in Florida, California or south of the equator in February, to me here in Boston and refrigeration and preservation and advertising, et cetera, et cetera. The amount that Henry David Thoreau needed to know to sustain himself at Walden Pond was a ton. But he had to know a little about a lot in order to sustain himself. Now to get me that orange in February, there are a lot more things that have to be known. But today here in the 2020, we're still working with more or less the same kind of brain that Henry David Thoreau had back in the day. And so you asked the question. Well, now this is much, much greater set of sciences and expertise is that have to be known and mastered in order to get me that orange every February. Dr. Steve Spear (00:41:59): So, given that we can jam more and more stuff into same size brains, what do we have to do? We have to engage more brains in order to get me my orange. And so that's where the specialization comes in, right? Because there's just more and more pieces to the system to create these systems, which can do things which people back in the day, they couldn't imagine. They simply couldn't imagine systems doing these things. Gene Kim (00:42:26): Let's go to the car example. Let's go to healthcare next. But for the car example, [inaudible 00:42:30] actually very surprising to me is that for a given car they can get at a certain price. Inflation adjusted, the car you're getting now is so superior to the one that you're getting in the 1950s. Can you validate that claim? Dr. Steve Spear (00:42:43): Yeah, so you go back when the Ford Mustang was first introduced or the Chevy Corvette before that. Could the competent car owner do a good amount of the service maintenance and even repair on the car? Yeah, they could. And they could completely ignore the electronics, they could completely ignore the software. And why is that? Because that stuff wasn't on the car. It was made of iron steel, I don't even think was made of aluminum. Probably a limited amount of plastic, some wood, leather, fabric, not electronics, but electrical. It had wires and bulbs. So yeah, that was within the skill set of someone. But as those sciences and engineering technologies advanced and you started putting electronics on a car, the person who used to say, "Wow, I can repair my whole car." Well, even with the same skill set, they can now only repair half the car because they can repair the part of the car that's mechanical, but unless they add to their mechanical skill an understanding of electronics, they can... That's the other half of the value. Dr. Steve Spear (00:43:58): You start getting more advanced cars, where it's not just mechanical and electronic, but now it's a really sophisticated engineer software, now you're down to a third. You can fix the body and parts of the transmission, et cetera, et cetera, if someone else has to be responsible for the electronics and then you have to get someone else in to deal with the engineering software. And again, it's like the orange, which is a car today is a much higher performing product than a car of the 70s, 60s or 50s. In order to get that much, much higher level of performance, a whole lot more knowledge and expertise has to be synthesized into the product itself. In order to get that synthesis, you have to have many, many more experts who are making very big contributions. That's not to diminish them, but they're big contributions in an absolute sense, but in a relative sense, relative to the value of the product as a whole, the value is smaller and smaller. Dr. Steve Spear (00:45:02): That's just the nature of the beast. But in some regards, that's a good thing because now we can drive cars which last 20 years if you maintain them and they give you fantastic performance and if you have an accident, you're more likely to survive, certainly than compared to the past. You think about the functionality of a supermarket. If you went to the supermarket in February in 1964, 1968, no oranges, lots of potatoes, lots of onions, lots of canned goods. But the functionality of a supermarket now is that when I walk in in February, I can get fresh fish from anywhere in the world, I can get produce from anywhere in the world. The functionality is off the charts. And the other crazy thing is Gene, 1950, 1960, the supermarket not only underperformed in terms of variety, but it underperformed in terms of price and quantity. So, if I lived in 1950, 1960, I might actually worry even as an American about having a hungry night because I couldn't afford to get the nutrients that I might need. Dr. Steve Spear (00:46:05): Now, what's our problem in 2020? Is that we walk into these supermarkets with the platitude that's there. And not only is there so much of it in terms of variety, there's so much of it in terms of volume and quantity, is so freaking cheap that our concern is obesity, not starvation. Anyway look, in order to have a supermarket like that, the only way to have that is this huge explosion in the different disciplines that have to be known and the depths with which each of them has to be mastered. And that requires... That creates a huge opportunity and huge benefit, but it also creates this integration problem, which was Henry David Thoreau didn't have to worry about integration. Gene Kim (00:46:51): So let's go to one last example, which is healthcare and you had this amazing part of your book, where you're describing what happened to say one of the benefits of specialization, what it drove us towards was if you look at cancer survival rates in 1950s versus now. Can you talk about that and just the miracles that modern medicine is creating in the ideal and the integration challenges it also creates? Dr. Steve Spear (00:47:15): Look, Gene, that's exactly the same problem, which is... We're dealing with COVID right now. And if COVID had occurred 50 years ago, essentially 50 years ago, it wouldn't have been quite so severe because fewer places were connected with fewer places so quickly, so it might have appeared in Wuhan, China, it might never have gotten out of there because it takes two weeks for the infection to become symptoms, to become transmittable, but if you can only go 50 miles out of Wuhan in two weeks, how far can it go? But anyway, let's just assume that 50 years ago, we had all the advanced transportation technology we have now but not the medical science. So, what happens with COVID 50 years ago? We're all dead, because the disease gets out there, no one can diagnose it, no one can recognize it. There's no treatment protocols developed, on and on and on. Dr. Steve Spear (00:48:08): But what's happening now is that if we would just socially distance, wash our hands and wear a mask, we can buy time for all these fantastic advanced technologies to swing into play in terms of genetic sequencing of the virus, genetic sequencing of the human being, construction of rapid trial vaccine experiments. On and on and on. And so look, this thing eventually will be contained and it will be cured because of the vast expansion and the vast deepening of scientific knowledge. Now of course, to bring all that stuff to bear, you have to have integrative mechanisms for the people in each of these disciplines to make their contribution to the larger system of diagnosis, treatment and cure. But thank goodness for that, right? Because otherwise this thing would be a plague, but worldwide rather than limited to some town in Italy or France or England. Gene Kim (00:49:18): So talk about the orders of magnitude improvement in cancer survival rates, just to put into perspective of just the vast leaps made in medicine because of everything you've talked about. Dr. Steve Spear (00:49:30): Look, when I was a child, if someone you knew was diagnosed with breast cancer, that's when the hand wringing started, because it was a terminal illness. And the question was, "Oh, what will her experience be? Oh, what will her husband do with the children?" Da da da. And over the years people have come to recognize that while the symptom may be malignancy, the disease is in the dozens, if not the hundred. Each of which have their own genome typical, phenotypical presentation and cause and their own treatments. And with that much greater precision as to what the diseases are, which have similar presentation and what the treatments are for those individual diseases, the survival rates have gone... Exponential growth in survival and exponential collapse in mortality due to the disease. It's just been fantastic. And again, we can check the book for the exact numbers. Is it a 90% reduction and some crazy number like that? Gene Kim (00:50:36): And so speaking of Walden, for those who wish for a simpler time, that we want the topography of the problem to be a little more homogeneous or more homomorphic, a little more generic. Does wishful thinking make the case that complexity is going to keep increasing as the years and decades go by or is complexity being reduced? Dr. Steve Spear (00:51:02): That's a good question. So, is the human species going to plateau on what it knows and plateau on the necessary divisions in intellectual labor, so that we can have some expertise, meaningful expertise in our field? I think it's hard to predict the plateauing in discovery. And for the simple reason or two simple reasons. One, is our understanding of the world around us is minuscule. You just pick up the science section of any decent newspaper and they start telling you about these things have been discovered in the universe that no one knew existed. The universe has had a 15 billion year opportunity to create stuff, it's not new. And we're barely scratching the surface about what's just barely beyond the perimeter of the Milky Way. So, there's a lot of stuff we simply don't know are there. And since the universe has had more or less, mankind's been around for a few 10s of thousands of years and the universe a dozen billion. It's got a quite a head start in terms of building up a repository of things we don't yet know. And it takes a long time to catch up on the universe. Gene Kim (00:52:16): And that also seems that our goals don't get any bigger, right? Dr. Steve Spear (00:52:19): Yeah, yeah, that's right. That's right. We decide that there's nothing left to learn and we don't care anyway. So, no I don't see this plateauing telling and I certainly don't see it going backwards. Gene Kim (00:52:32): Gene, here. Okay. That may have seemed like an odd question to ask Steve, but I find his point to be very, very important. The world is getting more complex and the need for specialization is going to continue to grow. And therefore the need to integrate efforts from an even wider base of skills is needed. Like it or not, this means that the problems that all leaders will face going forward, will keep getting larger and just staying in place is actually falling behind. That's a big, big problem that demands better answers, which is why I'm so excited about this quest I'm on to better understand why organizations behave the way they do and why structure and dynamics help us better understand how we can change them to get the outcomes that we actually want. Okay, back to the interview where we talked about the MIT beer game. Gene Kim (00:53:23): See, if another thing that was just a genuine pleasure, but also an eye opener beyond words, was looking at the famous MIT beer game through the lens of structure and dynamics. So the MIT beer game of course, was created in the 60s by Jay Forrester and the Sloan system dynamics group. It's a simple game. Every table is one supply chain, there are four positions each played by a player, retail, wholesale, distributor and factory. In every turn, a customer puts in an order and that flows through the supply chain, everyone orders from their downstream step and fulfillments come together. And the observations of the MIT beer game, [inaudible 00:54:03] 50 executives perform no better than high school students and so the anger that their fellow players don't actually seem to actually understand the rules, this bullwhip effect where inventory builds up in the system and soon is swamping the entire supply chain. Gene Kim (00:54:19): And so, so much of that is embedded in the structure of the game, the rules of the game, the way that players are able to interface with each other and the nature of how quickly inventory ordered from the retailer can actually come back. I asked you this question and I found your answer to be so startling, which is, if you could change the structure in just one way that you think would lead to far better outcomes, what would it be? Dr. Steve Spear (00:54:48): Acknowledgement. As you're describing the beer game, it's a simple setup, it's very linear. A customer goes to a retailer and says, "I'd like the case of beer." The retailer goes to a wholesaler and says, "I'd like to replace that case a beer." The wholesaler goes to the distributor, the distributor, the brewery and says that. And the only change in the entire game is at some point, the customer walks up and says, "Instead of one case of beer, I'd like two." That's it. Now, were everything to work instantaneously, this game would be a simple one, right? "I want one, I want one, I want one. I want two, I want two." And you get that back. The problem is the system has lags. So the first time the customer shows up and asked for two, the retailer doesn't have enough so the customer is disappointed. And then when the retailer asks the wholesaler, it's not an instantaneous response. It takes a few cycles and so on. And there's disappointment. And what ends up happening is that when people don't get what they asked for the first time, they start escalating their requests. Dr. Steve Spear (00:55:54): So, the customer rather than asking for two, then two, then two, they ask for two but didn't get one of those two. So when they come back, they don't ask for two, they ask for three and then they ask for four, this thing just starts to amplify and explode. I tried thinking about why all that happens, because when the customer shows up to the retailer and says, "I'd like two," the retailer doesn't say, "Here's one, if you just wait and be patient, the other one will follow on a lag of three cycles." And so they say, "Here's one," and the customer walks away frustrated. Dr. Steve Spear (00:56:28): And when the retailer shows up at the wholesaler and says, "I'd like two." If the wholesaler simply said, "Here's one for now, hold on the other one that you need it will catch up and it'll catch up on a cycle of two or three or whatever it is." The retailer should be similarly patient. But they don't get the acknowledgement. Someone doesn't say to them, "Hey, Gene, I heard what you asked for, I'll get you what you asked for, just wait please." Gene Kim (00:56:56): You'll get it in four weeks. Dr. Steve Spear (00:56:58): Yeah, four weeks or three, whatever the number is. But no need to ask again, I'm telling you absolutely, I'm giving you a guarantee that I heard you, I'm working on it as fast as I possibly can and if you ask me again, it's not going to make it any faster, you're just going to get aggravated. Look, we're having this now with COVID. So, people are used to have the same day, next day delivery from Amazon or some others, right? And now they have to wait two days or a week even or even till next month to get something. It's like the end of civilization. But the reason we're not falling apart on that, is that when you order something from a retailer, they tell you when it's reasonable to expect your order. They say, "Hey, Mr. Kim, we'll get it to you may 16th, is that okay?" Dr. Steve Spear (00:57:47): If you say it's okay, it's fine, then the order is there and you know may 16, the thing will show up and maybe you'll get it on the 15th, but all right, you've been heard, you've been acknowledged. On the other hand, if you place an order and they don't say they accepted the order, one or two things, one is you start going there or there is you keep calling them. "Hey, did you get my order? Hey, did you get my order? Hey, did you get my order?" And that's basically the beer game. It's this person is desperate for their case of beer, they get no acknowledgement, they have no other place to go. So they just keep calling over and over and over again. And the system goes out of control. Again, because it doesn't have this acknowledgement, this feedback to say, "Everything's okay, just hold on." Gene Kim (00:58:31): It seems like a profound point here. So, one of the papers about the MIT beer game they say the lesson to leaders is to have a calm, detached, unemotional view to trust your fellow players, but it seems like a more direct better solution is... I think the key learning really is to change the structure so that it doesn't cause this terrible dynamic. Dr. Steve Spear (00:58:55): So, that's an interesting one, which is supposed to have calmness. Well, that's ridiculous with human beings. So, who the hell is calm when they're under duress? So, it really is a causality here, which is you should be calm because you have trust. Where does trust come from? Trust comes from experience. The only reason I trust you is because we spent time together and you and I know that in certain circumstances, I know how you're going to behave, what's to be expected, what's to not be expected, et cetera. That's where the trust comes from and that's why it can be calm on a new experience, because we've built up this reservoir of trust from past ones. And again, if you start thinking about that experience, so what was the experience? The experience was one of feedback, that I acted in a certain way and you reacted or you acted or I reacted in such a way that I recognized that there were certain patterns which I could trust, right? Dr. Steve Spear (00:59:45): So, it seems to me somewhat, superficial, cursory to say, "Oh, Gene, don't panic, just trust." It's like, "Yeah. Well, I got no data." I got no evidence, no experience on which to build that trust and the absence of experience to build that trust when that happens, that is a very comfortable natural panic. So I think underneath that advocacy for calmness built on trust, which implies built on experience, what we're really saying is you need feedback. And if you want to get to trust quicker, the way to get trust quicker is to give that reinforcing feedback faster and more frequently, Build those sets and reps early on. Gene Kim (01:00:26): That seems to be the other universal truth in the nature or the structure and dynamics you want to create, is that fast feedback, ideally in small batches that allows for local decision making that best enables the achievement of the global goal. Dr. Steve Spear (01:00:44): Right, right. Look, we know from Bob Merton and Black Scholes and all of that, all those people who innovated on the auction theory that there's a huge price to uncertainty and a huge value to certainty. Their whole Nobel winning research an explanation of how one price is an option, it's what price do you pay to relieve the uncertainty of a situation? And we see that I think in the behavioral sciences, things that people will do which seem self-defeating or suboptimal, just to relieve the uncertainty. Is it the end of the Dirty Harry movie, when he... "Hey, punk, did I fire five times or six?" And the right thing for the villain in that moment to do is nothing because he doesn't know how many times Dirty Harry has fired, right? And he should just do nothing, just in case. But he's, "I got to know, I got to know, I got to know." It turns out it was five times and he gets shot by the sixth one. Dr. Steve Spear (01:01:48): And in some ways, I think there's a very empathetic response the audience has. On one hand, they're kind of glad that Dirty Harry shot this horrible character and there's that kind of tapping into that part. But also I think [inaudible 01:02:02] is like, "If I were down on the ground after this whole prolonged conflict with Dirty Harry, I too would want to know if it was five or six." I don't think anyone looks at that as ridiculous, ""Oh, well, the killer should have just been patient and calm and trusted." No, anyone in the villain role. It's a funny scene because thinking about it as a movie viewer, we end up identifying with both the hero and the villain, which is an unusual thing because we identify with Dirty Harry is like, "Yeah, I want to get this guy and with the villain also I just want to know." Gene Kim (01:02:37): I never thought about that way, sort of that final scene is really about he didn't have to call that option. Dr. Steve Spear (01:02:46): He did it, he did it. He could have just said, "You know what? Let's assume five because the cost of assuming five and being wrong is way different than the cost of assuming six and being wrong." But he can't wait. He can't live with the uncertainty, he needs to know. Gene Kim (01:03:09): Oh my gosh, I love that last part at the end. I debated on cutting it because Dirty Harry and Back Scholes option theory seemed a little bit off topic from the MIT beer game, but wow, these topics actually came up in the discussion from NPR Planet Money about how markets can be an incredibly efficient and accurate signal about supply and demand and how it can actually increase the number of people benefiting from a system. Okay, on the next episode of the Idealcast, I am so delighted that I have on Mike Nygaard, currently VP of enterprise architecture and platform development at Sabre, the software and technology company at the heart of the business of travel, who pioneered travel reservation and booking in the 1970s. It is difficult for me to overstate how much Mike Nygaard has influenced my life, I believe his seminal book, Release It, is one of the most important books in our field. Gene Kim (01:04:06): His work has pioneered patterns that we're all now so familiar with, such as circuit breakers, bulkhead, gas engineering and so much more. And I would be remiss if I didn't mention that Mike Nygaard at the Velocity Conference in 2013, he introduced me to Clojure, the programming language which reintroduced the joy of programming back into my life some years later. He is one of the best people I can think of to talk about software architecture. And I learned so much from him about the structure and dynamics of bad software architectures and great architectures, as well as how to create them. See you then.