Virtual US 2022

Organizational Coherence: Simplification and Slowification, Part 2

Organizational Coherence: Simplification and Slowification, Part 2


Gene Kim

Author, Researcher & Founder, IT Revolution


Dr. Steven Spear

S2S: Founder & MIT: Sr. Lecturer, See to Solve; MIT





Thank you, Doug and Raquel. So one of the main things I've been working on these days is a book with my mentor, Dr. Steven Spear, due to come out in a year. So he is famous for many things, but he's probably most famous for writing one of the most downloaded Harvard Business Review papers of all time in 1999 called Decoding the d n A of the Toyota Production System. This was based in part on his doctoral dissertation that he did at the Harvard Business School. And in support of that <laugh>, he worked on the manufacturing plant floor of a tier one Toyota supplier for six months. And so since then, he's extended his work beyond just a high repetition work of manufacturing to engine design at Pratt and Whitney, to the building of the safety culture at Alcoa, and how to make truly safe healthcare systems.


And he was part of a US Navy initiative to create a high velocity learning dynamic across all aspects of the enterprise. So in Las Vegas, we talked about what we've been working on and all the amazing discoveries that we made as we work towards getting our book out, that attempts to describe a theory of how organizations work, both in the ideal and not ideal, trying to synthesize what is in common between DevOps, agile, the Toyota production system, safety culture, and so much more. It has been so valuable to use these DevOps enterprise summits as a forcing function as we write our book to force ourselves to try to articulate as clearly and concise as you can, uh, certain core concepts of the book. And I think this presentation that we did two months ago was one of the most important ones because we're starting to zero in on the core concepts of the book. So I'm going to replay the majority of that presentation, and then we will present our most recent reflections and learnings since then, which I'm so excited to share. And I hope that this all strikes you as, as relevant and as illuminating as it has been for us. I,


So with that, Steve, over to you.


Minute back. Hey, good morning everybody. Um, before I get rolling on this, uh, gene and I are, uh, trying out some new material today, and we like your feedback, throw it into the Slack channel. And now here's the reason to do that. If you do not put comments into the Slack channel, when we finally publish what we're talking about today, you're gonna have to read the same crap again. On the other hand, on the other hand, if you put some good comments in the Slack channel, when you get the book, you can turn to your colleagues and say, Hey, Dana, I'm the one who bailed 'em out on that book. So anyway, lots of comments, please. Now, in terms of, uh, the work we've been doing, we start with a basic question, why are the winners, why are there losers? Lemme say that again. Why are there winners and why are there losers?


And we take for granted that there should be winners and losers. 'cause so often they, there are winners and losers. But lemme just say that we should actually be surprised by the winning and losing in commercial competition. Why is that level playing field, right? Everyone goes into a, a realm looking for opportunity, right? The opportunities are similar to whomever it happens to be. They draw upon similar resources to construct solutions. They're using similar delivery channels to deliver those solutions level playing field all across the board. And yet there are winners and there are losers. And so when you take away all the common resources, you have to find something uncommon to explain uncommon outcomes. And what we think the answer is, is that the reason we get uncommon outcomes from common resources is that the winners are much better at harnessing the brains in their enterprise.


What I mean more particular about this is that when they take, when they have access to the collective intellectual horsepower distributed throughout their organization, they're way better at harnessing that through collective action towards common purpose than anybody else. Now, here's how we're gonna build out our thesis. So Gina's going to introduce an example. I'm gonna throw in some color commentary and develops into a model. And like I said, as we go along, your comments into the Slack channel, please. So next year you can say, Hey, that book is good cost of me, gene, over to you.


Thank you, Steve. So, um, I must say this has been one of the most intellectually challenging thing I've ever worked on, and also one of the most rewarding. But there are times when, um, I think that I actually don't know anything. So, uh, I wanna explain something that you might've seen. I've been, we've been, um, I've been writing a series of blog posts on something that might be, uh, a little strange, uh, but it's because, uh, I'm finally getting some clarity. Um, after working with Steve, after, you know, this has been almost a decade, and so this is, uh, the, the aha moments came after Steve and I spent a week together in Boston. So, uh, what the first essay I posted was about how even moving a couch, uh, moving a couch is not just, uh, bra work. It actually, it records a lot of brain work.


So let's suppose you have two people. Let's call them Steve and Jean. There's actually a ton of communication and coordination that is required, uh, as they move the couch when they start their immediately in conversation, actively communicating in according where do you put your hands? How do you keep the couch balanced? Uh, what do you do to move the couch through a door sideways or lengthwise? Um, you know, uh, how do you get downstairs? So suppose the homeowner shows up and insists that Steve and Jean can no longer speak directly to each other. In fact, even certain actions shouldn't be taken without first getting approval from the homeowner. Suddenly, very urgent messages like the couch is slipping. Can we slow down? Or, the couch is pinching, my finger in the doorway are no longer making turns in time. Everything is getting worse. Uh, it takes longer.


Uh, things are getting damaged around them, and even more dangerous as well. So I thought this had a whole bunch of, uh, for me, profound insights. So the second essay was, uh, the extension of the scenario. So it turns out that Steve and Jean are moving the couch because they're helping Miriam and Marguerite paint the walls and ceilings of that room. And so they're moving the furniture out because they don't wanna get paint all over the furniture. So in the room now, there are two people moving the couch, and two painters and strewn across the room are open paint cans and four ladders. And now the work, the communication and coordination required is substantially higher. People have to signal what they're trying to do, maybe ask people to get outta the way, right? Uh, you know, Miriam and, uh, Marguerite, uh, have to ask Steve Jean, uh, to not, uh, move something.


There's dependencies. We can even imagine a scenario where they deadlock and no one can actually, uh, do anything. So this is where coordination dominates and no, uh, energy is spent on the value creation activity of the task at hand. So, uh, what we've been working on for two months is the extension of the story. So the story begins because it turns out that Miriam and Marguerite are so pleased with the way the painting turns out, that, and everyone is so impressed with the beauty of the rooms, they've now been hired to paint all the rooms of an entire building, you know, 50 rooms across 10 luxurious apartments. So Marguerite and Miriam take charge of the entire operation. They receive a list of the rooms from the apartment owners along with the needs size right and so forth. So, uh, their first decision, uh, is to appoint Steve as a chief moving officer responsible for managing the movers.


Jean is, uh, appointed as a chief paint officer responsible for hiring skilled painters, um, to paint the rooms as, uh, uh, desired by the customer. So Jean takes the list from, uh, Marguerite and Miriam, uh, and creates a draft schedule of the rooms to be painted, uh, in what order, uh, shows that Steve, who scrutinize the schedule and make whatever changes is necessary, uh, so that he can promise that when the painters are, uh, gonna start painting, the furniture will be gone. And when the furniture is done, I'm sorry, when the painting is done, the furniture, uh, will be restored. And so that the owners can take, uh, move back in. So they print out the schedule, put it in their clipboards, uh, give it all to the movers and painters on day one, but things quickly go wrong. Sometimes painters show up to paint the room, but all the furniture's still there.


Why's, because the movers are running late because there was more furniture than expected, or the furniture was harder to move than expected. Um, but something else that looks benign is happening too. Movers are removing the furniture sometimes long before the painters arrive. Why? Because sometimes a painting takes longer than expected. Sometimes, uh, they can't do the second coat of paint, uh, because it took longer to dry than expected. Um, so this leads to a spectacular moment where no one can start painting new rooms because the movers have run outta space to store the furniture. So, uh, and now we have to reset the whole system, um, and everyone potentially has to stop working. So at this point, uh, Jean is very frustrated with Steve. All of, uh, Jean's painters are complaining about how Steve's movers are never in the right place at the right time.


Furniture's not being moved in the right way. In fact, painters are starting to go directly to the movers, telling them what to do, and which leads to this weird situation where painters aren't painting because they're too busy telling the movers what to do. Similarly, Steve is, um, having to, uh, frustrated with Jean because the schedule is widely inaccurate, and it didn't take into account all the unexpected variety being encountered in both the moving and painting operations. Steve is having to firefight steaming movers from different teams with difficult jobs. Um, uh, and a jean does the same with painters, but this is actually causing something even stranger to happen. Problems are now rippling out. Problems are not isolated to a room anymore. It ripples out through the entire system. Steve and Jean, how are they spending their time? Uh, they are hopping from one problem to another, just trying to get teams what they need.


Uh, movers and painters are, uh, yelling at each other just because they don't, uh, have what they need. And it seems like everyone is talking to everyone else. Everyone in the system is spending time coordinating just to get what they need to get their job done and incident. The irony here is, despite the huge amount of coordination, it is nowhere near adequate to the task. Um, uh, and their bosses, oh, they all agree on one thing, though. They all think that Steve and Jean are not very good at their jobs, including, most importantly, their bosses. Marguerite and Miriam, they're both very unhappy with Steve and Jean because they have to explain to the customer why all their promises that they made have not been met. Uh, none of the apartments have been painted as promised, and the apartment owners can't even move back in. And when they do like the furniture's, half the furniture's missing.


So they, at this point, they think they might have to fire Steve and Jean, uh, before they're fired by the customer. And all these characteristics of good architectures and modularity that we've talked about over the last two days are absent. So that evening, they come up with a different structure, uh, they decide something else is needed. Uh, so they realize that having the world's, even if they had their world's best movers and painters, they would not solve the problem. Uh, instead they have to figure out how do they get their painters and movers work in anything that looks like harmonious coordination and collaboration. And so they get one more chance from Marguerite and Miriam. So instead of sending painters and movers to a room on schedule, they form room teams, uh, where they're assigned a group of apartments, which these room teams will complete one after another.


And in each room team will be a coordination lead responsible for all the internal moving sequencing coordinating within that activity. And now all the room teams can work independently. They will own all the sequential steps of removing the furniture, painting, restoring the furniture within the team. When something goes wrong, uh, they can be handled within the team, right? And if they really need help, they can go to Steve and Jean. Um, what does Steve Jobs Steven Jean's job now, it is to create great painters and great movers that they can give to the room team because they become the customer. And incidentally, if something goes wrong, if furniture takes longer to move, painters will wait patiently, or maybe they will even help. And the notion of what the team is at the edge changes in the previous, uh, scenario, the team was the movers, right?


Or the painters. And now it is all around the room. Instead of a transactional, uh, interaction, now they're in a co-creating activity of, uh, uh, giving rooms that are beautiful. Um, the result movers and painters are now happier. They feel like they are working towards a common purpose with a genuine sense that they're on the same team and they're actually creating knowledge, uh, which benefits every future room they paint. Uh, some teams decide not to return the furniture until the paint is dried. Uh, some teams close the windows when they notice pollen sticking to the walls, which we require painting. And in comparison to the previous system, the system is calm, quiet, and orderly. Communication and collaboration, communication is way, way down. Instead, they are, uh, collaborating around the value creation activity. So, uh, apartments are beautifully painted at a level of quality that Marguerite and Miriam are delighted by, which also delights the customer.


And so I'll just end with, uh, a couple observations for Steve and Jean. They had to jump from crisis to crisis in the first scenario. Instead, uh, they are now in a contemplative, uh, mode. Uh, Steve notices that the furniture, um, uh, some from the furniture team, they're using blankets to protect the furniture, and he decides to roll that out. And he creates blanket teams to help support the movers, um, elevate the quality. Um, um, and Steve also knows as many innovations and exemplary practices, um, that are now spread through the organization. And this way they're experimenting with new structures and configurations and architectures and improved performance, and they're pushing the frontiers of what is possible. So I learned from mc, Kirsten, I was reminded about, uh, cohesion. Coupling in the first scenario. We had high coupling and low cohesion in the second we have low coupling and high cohesion. The teams are able be able to work independently, but also in harmony towards a common purpose. Um, and we can also, this parsimony example gives us ways to say and team topologies the notion of enabling teams, productivity teams, all these things. So, uh, what is the difference between the first and second scenario? Nothing except for the management system that Steve and Jean used. Steve, how'd I do? Uh, is there anything I forgot?


Oh, I know. That's good. All right. So let, let's step back a moment to our thesis, which is there are winners and losers. And the reason the winners win is 'cause they're way better at harnessing the intellectual horsepower distributed throughout their organization, harnessing it through collective action towards common purpose. And losers do what Steve and Jeanne were doing, which is they're very, very poor at harnessing that intellectual horsepower. So just let's walk through the case. We start thinking through, and this is one, one part where we want some slack comments. We start thinking, if we're competing on our ability to harness the intellectual horsepower in our organization, what are conditions in which we're really, really bad at solving problems? And what are conditions in which we're really, really good? So conditions in which we're really bad is that we've got really hard problems. They're very complex.


The environment is moving very quickly. We get few iterations at, if any at all to get some learning loops going on. The hazards are very high, the risks are very high, et cetera, et cetera. Wicked hard to do, slow, deliberative, contemplative problem solving when the environment is so, um, aggressive against you. In contrast, what's the, if that's the danger zone, what would we call the trium zone or the characteristics? So the trium zone is all the opposites, right? Things are slow moving, they're decoupled, they're, um, less complex. You get iterations, shots on goal, low hazard, low risk, lots of safety. So when we start thinking about our own organizations, ideally what we're doing is helping people get out of the danger zone and more and more into the triumph zone to do their problem solving. And with this example, gene sort of illustrated this going on.


So the first setup was just total chaos. People showing up on a job, trying to figure out what to do while they have to, trying to figure it out while they're trying to actually do it. It's a mess. So then they do a little bit of simplification. They decouple, they partition time. So there's a moving element and a painting element. It helps a little bit, but it's not great. The movers are still arguing with each other, painters with each other. So then what do they do? They say, whoa, we gotta slow things down. We gotta calm the situation. We gotta move in the triam zone and slow moving. They say, Hey, why don't we stage in the basement a mock-up area so we can figure out where's the furniture starting? Where's the furniture going? Where do we have to put ladders? Where do we have put cans of paint?


And then create standards that we can bring back into the operating environment. So now we have an edge, again, over the fast moving environment. Now that, that works somewhat, but not perfectly. There's still some problems. And they say, well, you know, we've simplified things. We've created these standards in the slow moving environment to bring into the fast moving environment, but there's still stuff we just don't anticipate. So the first instinct, the first instinct is to grab Jeff off of another paint crew and have him help out in the apartment over here, which is having problems. But you know, what happens then? You've now coupled the systems, right? 'cause the problems in one apartment now ratchet right over to the other apartment. 'cause you pulled Jeff across. So now Gene and Steve have this third realization, one, to get people from that danger zone into that triumph zone.


Simplify things, that's good, stable, standardized things, which you can do in that offline preparation and stabilize things, but not like by pulling one resource from another project and now making the systems more tightly coupled and more prone to failure. Actually throw a little slack into the system. Maybe we'll have Steve Thomas, who, who's the, who's there on hand, just in case. All right. And then what's the third thing Gene talked about Is that, um, once we have much simpler flows and it's much easier to make sense of things, what's feeding me? What am I feeding? And we create standards. So now we've got into a situation, Ooh, what can I expect that I have to do? 'cause what can I expect about my surrounding environment? And then we stabilize that whole thing with Steve Thomas coming in. Now all of a sudden we can have this thing self synchronized.


We don't have to have the paint crews going up and down all the time trying to find Gene or Miriam or Marguerite trying to figure out what to paint and when to paint and where to paint and do all that. We don't have to have the movers going up and down to Marguerite and Anne and Jeff trying to figure out, do I move the couch first or the automo or the chair and where do I put it? This or that. Now that we have simple workflows with standards built in and stabilizing factors, the work itself becomes self synchronizing. And what does that mean? Is that what we've done is we've created a much simpler, less complex, calmer environment in which people can do their work in the operating environment. And now they have a huge advantage over it. 'cause they're carrying into it the product of their harnessed intellectual horsepower, harnessed through collective action for its common purpose.


And so what do we see Gene do here in managing this? He simplifies things, pull things out of performance, execution, back through practice into planning, and then takes what's developed and plans and brings it back into practice. So that's our story. And like I said, we wanted some feedback on this. And uh, I guess this is where we typically end with, uh, asks. So I've got two an ask and an admonition. Um, this whole issue of pulling people out of the aggressive, dominating, disadvantaging danger zone of trying to think during operations and allow people to do really good hard, deliberate problem solving, collaborative problem solving in planning, it turned out the type of work you do, we just type it into keyboards. There's no movement of physical material. Sometimes people lose track of the system into which they're embedded. So, uh, do me a favor.


You got your little phones and tablets and that kind of thing. Type in c2 My man Daniel is seeing how many click-throughs we got. So type in there and we've got a product protocol flow. Take a look at that, and we think it'll help you be more productive in the triumph zone. All right? So anyway, that's what they ask <laugh>. Um, as far as the admonition, as far as the admonition, if you have responsibility for other people, if you have responsibility for other people, and I think everyone in here has responsibility, at least for one or two, if not many, many other people, then you are in a position to shape the time and shape the space in which they operate. And if you have opportunity to shape the time and shape the space in which they operate, you can determine you have great authority and great power to determine whether they're operating in an environment which is the danger zone, where they're constantly trying to solve problems on the fly, real time with all the frustrations that ensue.


Or you can shape time and space so that people, when they're solving problems, they're solving problems in the triumph zone, not the danger zone in the triumph zone, where individually they can give fullest expression to their innate creativity. And collectively they can give fullest expression to the collective whole. And I gotta say, if you're in a position to shape time and shape space, that's kinda like God-like powers alright? But think about this. What it really means is that you have opportunity and consequently responsibility to shape other people's time and space. The finite space that we occupy, the finite time allowed to us in such a way that at the end of every day, someone can go home and say yes or no. Today, the person responsible for me today, the person responsible me for me shaped my time and my space. So when I went into my moment of test, I was prepared to succeed. Not only was I prepared to succeed, but what I did was appreciated by somebody else. And when I went into my moment of test, the person responsible for me shaped my time and my space in such a way that when I did my work, it added value to my life. And that is a choice. And so make the right one. And so


Steve, that was so fun. Uh, I mentioned in our Vegas presentation, um, how working on this project, uh, with you has been one of the most intellectually challenging and rewarding things I've ever done in my career. Uh, and I thought it would be useful for us to share some of our latest reflections and aha moments, uh, studying this parsimonious example, getting ready for the presentation even afterwards. Uh, does that work for you, Steve?


Oh, I, absolutely, I agree with you with the sentiment about the experience, and I think explaining it would be fantastic, <laugh>.


Awesome. Awesome. So, uh, four observations. One, it's amazing to me to see how thoroughly we can screw up a system that only has two functional silos, <laugh> and yeah, only three interdependent steps. So I think it's one of the simplest systems we can create in terms of all the dimensions of number of functional specialties, frequency, complexity, variety. Mm-hmm <affirmative>, consequence, speed, information, density. Uh, does that resonate with you? <laugh>


A a absolutely. I think it's a beautiful example because, uh, what happens with the, uh, the movers and the painters in that example gets replicated when you start adding the functions or the silos or whatever else the case may be. Uh,


Awesome. Secondly, um, I love this parsimonious example because it can explain so much of the observable phenomenon that we see independent of industry. Um, but I had mentioned briefly that one of the surprising insights that came to me after doing that, uh, example, was that in the software industry, we spent 30 years, uh, trying to figure out how to quote shift left. It's always been a mystery why, and I think it's because that those interdependencies or interdependent steps as we go from design to development to qa, to deployment to operations and information security, it is so costly. So by having QA build automated testing, uh, rather by having QA with development build automated testing that can run on every code commit, we actually eliminate that interdependency, which has high value, but also ameliorates so many of the downsides of that interdependency. Uh, does that resonate with you, Steve?


Yeah, absolutely. Jean, if you think about what we're trying to accomplish with this work, and I think part of the reason it's been so, um, both rewarding and hard to do is what we're recognizing is that no matter what sector you look in, there are these enormous differences in performance, uh, separating the very best from their peers. And we have a, a fairly simple hypothesis as to why that is all else equal. The only thing left is how the superior performers, um, manage the intellectual, uh, the human intellect within their enterprise relative to everybody else. And they manage that in human intellect in such a way that it's much easier for individuals and collective groups of people to solve really hard problem problems better, quicker, easier than in the peers, than the and in the near peers. And, um, what we're trying to do is explain what you can do in terms of partitioning the space in which you work and how you manage that space and the time and how you manage that time to make it much easier to solve hard problems than otherwise would be the case.


Awesome. And maybe just to briefly expound on that, it is to me a minor miracle that we can actually execute work in all those domains of aviation medicine, software engineering, r and d, right? Despite all these problems and make the mistake, I think we can, uh, we can do so much better, right? The last, um, observation I have is that this example shows how we can create the properties of modularity even in sequential processes that have interdependent steps. And so we talk about a lot in the software community about software architecture, the, which enables teams to work independently towards common purpose. Um, and one of the key properties of modularity is that not only can teams work independently, but when things go wrong in one area, it doesn't spread and cause global impact everywhere else. But one of the key, um, a dazzling aha moment was this explains, uh, how Toyota can do 3,500 Andon cord pulls per day. Uh, you shared with me very recently that, uh, in a Toyota plant, uh, this year, uh, an average worker in the Toyota plant will pull the Andon cord 40 every 40 minutes. And this can only happen if you have a modular system where local errors can be contained and solved, uh, before rippling out and spreading everywhere. Does that resonate with you?


Yeah, that's right Jean. So one of the things we've been talking about in terms of making it much easier to solve problems is that we have to make our problems simpler and give ourselves the opportunity to have more learning loops, more experimental cycles. And so, um, referring back to, um, Charles Perot, he talks in terms of systems being complex and the parts of the systems being tightly coupled. And the more complex the system and the more tightly coupled the things are. One, it's much harder to do sense making about any experimental cycle 'cause there's so many factors at play all at once. The other thing is the more, um, complex and tightly coupled the system is the more, um, laborious it is to coordinate testing a change in one locality with a changing a change in another locality. So not only is the sense making harder, the number of experimental cycles goes way, way down.


Now the um, reverse of that is to have simpler systems. So linear systems are very, very simple and ones where, um, what happens locally is decoupled with, with what happens more systemically. And in the case of Toyota, that's exactly what the Andon cord allows, is that if Gina is doing his work in the absence of an Andon cord, what ends up happening is the work he does becomes much more tightly coupled with everything before and after because the disturbance he's experiencing has to spread, whether it's spreads in the form of a defect, which gets passes along, or it spreads in terms of a cycle time corruption of the system. Because gene is either starving something downstream or blocking something upstream, it spreads and causes this greater and greater and greater coupling of the system. So what does the Endon cord do? If, if Gene is doing his work and runs into difficulty, he calls attention to the difficulty and immediately gets support, let's say from Aaron who comes over and says, gene, what's the problem?


How can I be helpful? Well, what that does is it keeps that local disturbance local and keeps gene's locality from coupling with the things that a foreign aft. And because of that, um, as we said, the, the beauty of having simple systems with low coupling is that now you get this opportunity to do this independent experimentation where you get more cycles where each cycle is a easier and easier sense making. So Gene, because he's got the Endon Quats, which provides, um, or triggers Aaron to provide support, means that everything foreign aft has this beauty of these great learning dynamics without being corrupted in terms of defects being passed or cycle times being expanded because of the local problem gene is having. Right?


And to map it to the, uh, moving, uh, painting scenario, uh, we could put an and on cord through all the room teams. If they pull the cord, then the floor supervisor would, uh, be escalated to hopefully solve all the problems, right? Potentially escalating one more level, but all the floors would still be able to work independently, decal from hundred


Percent the other teams. And, you know, and taking that to the, you know, the, the granularity, if the individual mover when, uh, he or she has difficulty 'cause the couch is harder to manage, et cetera, et cetera, can call attention to that, then moving the couch doesn't then couple with moving the dining room table or the chairs. And similarly, if the painters have this opportunity to draw attention to the problem and have it, um, supported and contained, it keeps the painting decoupled from the other parts of the painting keeps it de decoupled from other parts of the moving and keeps the experience in that apartment decoupled from the experiences in the other apartments.


Love it. So that answers a, a huge question I've had for, for a decade. So one of the topics that we've been spending a lot of time discussing, uh, and it showed up in one line, uh, in the Vegas presentation, it was the notion of coupling and coherence mm-hmm. <affirmative>, and we've been talking a lot about coherence. Why do you think this concept is so important?


Oh yeah, Gina, it's, it's great. So if you look at the definition of coherence, it has two and I think related meanings. One is, uh, something being a unified hole. The other is that it's, there's a, a logic and a consistency. And of course, if something is not in a unified hole, it's missing pieces or there are pieces in there which simply don't belong. One can imagine that thing is gonna start, start losing. Its, uh, its logic and its consistency. Now, one of the things we've, uh, we keep coming back to is our, our thesis, which is that the enterprises which, um, are far more successful, they are far more successful in delivering value into the marketplace because they're, they've created these conditions in which it's much easier to solve hard problems. And what we were describing with the, um, the painting moving example was taking, um, a unified whole of the work required to, uh, transformative apartment and creating what it was a coherent working unit and making it smaller and smaller.


So before it was the building and then it was the apartment, then it was the painting crew versus the moving crew in a particular apartment. So what we were doing is trying to, um, reduce the size of something that could be called coherent, that it had all the resources that need and nothing else to get the work done. And because of that, um, it could be logical and consistent in its behavior. And again, with this idea of being, uh, pushing towards smaller and smaller, um, units of coherence, what we're really saying is that the resources in that coherent unit are increasingly decoupled from the rest of the system. And it's through this decoupling into smaller and smaller coherent units, they'll get these beautiful advantages of simpler situations in which we're trying to make sense from our experience or our experiments and the opportunity to get more experiential cycles and more experimental cycles in which to make sense.


Love that. I love that notion that fully, um, enabling, you know, the unleashing of problem solving and creativity. Uh, but we were also talking about this kinda magnificent example of Steve and Jean moving a couch that, uh, we can actually make their, uh, working situation increasingly coherent. Can you talk about that?


Well, that's right. When we started off with that very simple example of, uh, two guys trying to hef the couch and navigate it out of a room into wherever it had to go, um, the complete unified unit is the two people trying to move the couch. And the way they remain a complete coherent, logical, consistent unit of work is through the communication that can occur. Um, hold on, slow down, speed up, lift higher, I'm moving to the left and moving to the right. And what we introduced into that example was this idea of distractions, either, um, and impediments to communication, whether it was the increase in background noise or the dimming of lights, whatever else that Steve and Jean no longer could communicate well enough to maintain their coherence, their unified wholeness through the bonds of communication and through the loss of the coherence in terms of being a unified whole, their behavior would become increasingly inconsistent and apparently illogical as one would do one thing based on the, his best understanding of the situation. Um, and the other would do something else in contradiction from the system perspective based on his understanding of the situation.


I love that. And to me that is such a startling insight. If you degrade the communication or the quality of the communication, uh, what is one team of two people can be come now two teams of one people <laugh>, and you get all the, uh, bad, uh, side effects of that. Uh, Steve, I also heard you recently share what you, uh, said to a bunch of leaders in a semiconductor manufacturer, and I thought it was so dazzling because it helped bridge the notion of modularity, uh, in say A C P U to the modularity that we need in our organization to increase coherence and decrease coupling. Can you repeat that story for us?


Yeah, absolutely. So the, the problem they were addressing was their, um, the effort required and the time required to get a, an idea from its initial, um, inspiration through its maturation, through delivery. And what was happening was that the management of ideas in their maturation was done in a very sort of top-down job shop approach where a little bit of work was being done over here, and a little bit of work was being done over here. And the people up here had to, uh, be monitoring what was being done, where its state of completion, the state of engagement by the different, uh, engineers and technicians and sciences, et cetera, et cetera. And in effect, their processes were very coupled and very complex, um, by the only awareness being at the very, very top of what was going on. And, um, to make the point that this is a very difficult situation, we started reflecting on how one would design a microprocessor.


And a microprocessor obviously has some, you know, huge number of functions, but what ends up happening is that the functions are broken up and divide and assigned to different parts, locations, they're architected onto different discrete parts of the overall chip. Then within that, um, discrete part, um, that location has given all the devices it needs in order to perform that function, whether those devices are, um, resistors or transistors, whatever else it happens to be. Um, that function gets everything it needs, and it, it's telling that by and large, um, when a microprocessor is designed, devices aren't shared even if they're not used to full capacity. So you wouldn't take a resistor and say, well, that resistor is only used half the time or to half its capacity <laugh> or that inductor or that transformer. So we'll share it across functions. And there's a reason for that is that by making that function, using the term, again, coherent, that it's a unified whole.


And then by making sure that that function is not sharing devices, so not only is it a unified whole and coherent, um, it's decoupled from some of the surrounding functions, it means those people responsible for the design, the improvement, um, the testing, the validation of that function can run one, they can run their experimental cycles quicker because they don't have to be coordinating with the folks responsible for other functions. And when they run their tests, they're running it on a simpler system for which it's much easier to make sense than if they had to coordinate all their experimentation across, um, the whole system. And so what we're trying to do is establish a parallel for them, is that just as they do with their microprocessors of trying to get this local coherence through, um, the decoupling of the pieces from the whole, they're still connected, right? They still speak through interfaces, but they're otherwise the decoupled that doing the same thing for their processes would be advantageous.


And we had this incredible conversation about how we can do things to enable or not enable coherence and say a, a house, right, that right, whether we share, um, dependencies or not. Can you talk about that? Yeah.


And, and again, bringing this back to sort of, uh, very familiar situations, um, you know, most homes, you know, most homes which have, you know, some measure of, of, um, material comfort. The, the chairs that are put into the kitchen are separate from the chairs that might be put into a dining room if there is one, and which is separate from the seating, which is in the living room. Now, if you were trying to do this in sort of a, you know, resource, mini minimization optimization way, you would hardly ever need to do that, right? Because it's very rare that all the seats in the kitchen and all the seats in the dining room and all the seats in the living room were all occupied at once. So you could argue saying having all those chairs is actually redundant and wasteful. But I would argue just the opposite, that actually having all those chairs creates a tremendous amount of, um, clarity for you.


Because when you're, um, doing whatever the interior design is for the kitchen, you can pick chairs that are appropriate for the kitchen, they're good for the casual dining. That is what we typically do in the kitchen. They're good comfortable chairs for sitting and doing homework or, um, whatever sort of bill paying and local correspond we have to do. How does that local coherence and the decoupling, um, help? It means that when you go to pick chairs for the kitchen, you don't have to worry about the chairs being picked for the, um, living room. You don't have to worry about the seating problem in the dining room. What you've done is you've made for a much simpler problem to solve, and one which we can give much more, um, thought to in terms of iterative cycles to converge on a good answer without worrying about the answer you're trying to achieve in the living room or the dining room.


And I think another important property is, uh, if you don't do that, you've now inadvertently coupled anything that happens in the dining room with activities in the living room, <laugh>. That's


Right, that's right. Yeah, that's right. The, the, the, the decisions made about what to do in terms of picking the kitchen chairs and using the kitchen chairs now directly impacts, um, what you can do in the dining room based on the chairs you've now picked, which have to, uh, be used. And also if for some reason you need to have people sitting in the kitchen, um, then they can't sit in the, the dining room or the in the living room. And again, that, that combination of coupling makes your problem solving much harder in design and it makes your, um, the fragility of your system much greater in operation because of the shared resources.


Oh, awesome. Uh, Steve, uh, this has been so fun. I mentioned it's, uh, been so challenging, it rewarding, uh, and all this is just, uh, happening at time as we're sort of marching towards, uh, hopefully manuscript complete, uh, mid next year for a book release, uh, late next year. So, uh, Steve, thank you so much and, uh, to be continued.


Absolutely, absolutely. Gene, thank you so much.