Alexandros Marinos: Systems Thinking, Organizational Structures, Startups | Transcript
Full episode transcript below. Beware of typos!
Alexandros Marinos. How are you?
Alexandros Marinos 4:01
Hi. How's it going?
Nick Jikomes 4:03
Good. Good. It's good to see you. We've talked briefly once before. We're both based out of Seattle. Can you start off by just telling everyone what your background is? Yeah,
Alexandros Marinos 4:16
so I'm the CEO and founder of bilena, which does edge computing. I've been running this company for about, you know, almost giving us a decade now. We sort of, you know, trying to do what Amazon did for cloud computing for, you know, devices that are on the world were not necessarily your mobile phone. Um, and I guess of late I've been using some of the reasoning tools we've been developing inside the Laina to understand it. So there's the very fractured space of edge computing, to something that's kind of come in and affected us and affect the supply chain sports as well, which is a pandemic. So I guess that's how as we can.
Nick Jikomes 5:02
So, we're gonna do a lot of discussion, I think about how you think about organizational structure and, and how your company is structured, maybe not so much about the actual technology and what you guys do per se. But can you give people some kind of anchor there? So So what is edge computing? And what what kind of products or services do you guys actually provide your client?
Alexandros Marinos 5:25
Yeah, so basically, if you look around you, there's, you know, more and more computers, and not all of them are, you know, your phone or your laptop or something that has a user on them all the time, right, they are a plant, they are, you know, anything from a drone to a self driving car to a thermostat to, you know, some building monitoring solution. You know, smart cameras, all sorts of things that we put in our houses, but also in an industry in commercial settings, we've just did deploying all this technology. And we did don't have very good tools on how to make that easy to do for one and for another to keep them up to date. And then, you know, do all the things that we do for laptops and phones, cloud computing. So that's what we have a lot of customers at having from 100 to 10s of 1000s of devices. And they need to manage them as a fleet, right, not just like as one device, but like as a swarm. And we just make all of the infrastructure, all the plumbing, we do all the you know, unsexy work to to make our customers able to sort of do the thing that they specifically what to do on top of that. Technology.
Nick Jikomes 6:33
I see. So you guys are a startup that writes software to help to help people manage fleets of machines. Yep. Interesting. So in your Twitter bio, it says, systems thinking is the way to a positive sum civilization. So I'd like you to unpack that for us. What is systems thinking, in contrast to other types of thinking? And what what does it mean when you say positive? Some civilization?
Alexandros Marinos 7:00
Yeah, so I'm, I can explain systems thinking, the best way I have to explain something in my head is Tesla, it's a company, I've been investigating a lot. And you can see the sort of the conflict of points of view in the conversation around, right. So there's some people that really get it. And some people that really, more so a few years back before the numbers that come in. So what's the what's the interesting thing about Tesla is that if you model it on a spreadsheet, right, you're like, Okay, well, you know, the revenues, and how fast are they growing? And you know, what's the market share, and you come up with once. But if you look at something else, which is like all the feedback was sort of built into that company, you come up with a different set of conclusions, for instance, um, they gather data on their motors, right? That they have on their cars, about all sorts of like deep physics performance, because they're the first sort of electric motor that's been deployed at scale. They gather that data, they feed it into algorithms, those algorithms optimize the operation of those monitors, which they put back down into their devices onto their power, sorry, which are devices, they call it fiber woven. And then that improves the performance of the vehicle now, improving, improving how well American use energy means you need fewer batteries to account to go the same, just having fewer batteries means you have less weight, which means you have even fewer batteries, right. So there's all of these reinforcing feedback loops. And then the better you make your car, the cheaper you make your car, the more cars you sell, the more data you have, the better training you do are your motors, and you feed that loop back in there again. And so I've got a thread somewhere where it's like, listed out 10 or 15 of these feedback loops that are feeding into each other. Right. Now, if you put on a spreadsheet, again, anything that is cumulative, will look pretty weak in the early phase, right? If you have a nice linear growth, you're like, wow, that looks amazing. If you look at something that goes like this, the early phases, you could confuse yourself for thinking it's going nowhere. While it's building up all this internal strength, that product did not sort of appear have nowhere to be built first or to deliver its value. So it's really about systems thinking is really thinking about things in their interaction with each other. And thinking really about the composition and interactions of things rather than just taking in numerical targets in how you use it, you know, that could help right so much that is essentially a feedback loop that looks at temperature and what the desired temperature is and adjusts to hit that continuously. Um, but you know, when you go down that linear path of modeling things and saying, Okay, well, it's sold this much. How much next year and you like, Okay, well now 10 years in the future. It's like, Whoa, it's bad. paradigms, this. So, I'm, again, I think of systems as you know, this sort of holistic things, and I don't sort of I know that word, you know, kind of leads you somewhere else. But I think of, when you make something, you think of it as a system, it means you can build a lot of strength and, and, and a lot of resilience as well. So by resilience, I mean, something like antifragility actually is a better word, um, where it responds to challenge by getting better, right, that's kind of what you want, you want something that won't crumble under pressure, but it will actually respond creatively and become even better. So that's kind of a systems thinking way of building a company, which sort of applies decently well, to how volcanoes getting on as well. Um, now let's talk about positive stuff. Right? So these are the games that we play, right. So when you build a company, you design it up that, like, there's one designer, of course, larger organizations, like a spotter down. Um, but you know, when people cooperate, that there's no sort of higher authority to tell them, you know, what to do. So game theory sort of comes to play. And sort of in terms of interactions with people, you can have a zero sum games, like, you know, there's the pie how much everybody gets out of the pie, right, a lot of that thinking, but because when there is zero sum thinking, people tend to get into a very competitive mindset, you get, also get negative sum games, right? Which are games that, you know, like, well, if I destroy half the pie, I get to get three quarters on what's left, and you get only one quarter, right. And that's better. For me, it's worse for everybody else, right. Because if there was like four more people in the in the game, I get to like, get two of them out of the game, destroy half the pie gets three quarters of a pirate, and that's right, three eighths, let's say the pie that's left.
And somebody else gets one quarter, so I'm better off at the expense of everybody else, right, that's kind of a zero sum game, sorry, negative sum game. But zero sum games tend to become negative just because of the tension. Whereas problems and games is the opposite. It's like where our collaboration makes something makes both of us richer, makes both of us better off.
Nick Jikomes 12:15
I see. So zero sum is, when I win, you lose. So for every winner, there's a loser. negative sum is everyone loses. It's just a matter of who loses most, but we all lose something. And positive sum is the opposite where, at the end of the game, everyone has come out ahead who's playing?
Alexandros Marinos 12:33
That's even even better is it even better kind of a positive sum game, it's only when, strictly speaking for positive sum, it just has to be more value at the end and the beginning. But it tends to correlate very well with AMI, which is like everybody participates wins.
Nick Jikomes 12:49
And so when you were describing things like systems, and what systems thinking is, it seems to me that the important important things to think about here to use your Tesla example are sensors, and a negative feedback loop. So when you're describing the Tesla example, what you're really saying is they have in this case, internal sensors, sensors that are detecting and measuring something about the the insides of this product. And of course, there are external sensors as well, the company has to sense what's going out in the market through various mechanisms. And there are different kinds of feedback loops that one can build that are connected to these sensors that allow you to that allow you to do things to say it abstractly. So when you're thinking about systems in this way, feedback loops, censors, and when you're thinking about constructing an organization that is going to result in a positive some dynamics. How, how do you actually do that when you're constructing a company and sort of designing how people are going to interact with each other?
Alexandros Marinos 13:54
Yeah. Yeah. Great question. So this is actually at the, at the bottom of how Bolinas is constructed. So is design, at least a principle that we started with, we've been iterating on that. But I think the real inspiration came from, from my wife. So she is, was working full time as a journalist. And she was like a pretty traditional media company. And one thing that was really interesting to me from the story she was telling me is that she was on the beat, right? She was talking to the people with stories, but it was a train publication, which is important because the people you're writing the stories about are the people that are reading your stories, right? You're talking to your audience, basically picking out you know, which companies have good, good stories, etc. And so she had a very good sense of the company and the market and the product and how it needed to be she was saying, you know, we need to do this. We need to do that, right? Because you just when you talk, they're like, oh, yeah, I like this. I didn't like that, but you get a sense And, and weirdly enough, that was not getting to the hertz, right? So she was like on her own coding some stuff and doing visualizations with data for her for her stories, but like nobody else was paying attention. Um, and it just baffled me that you have this massive company, right? Where the frontline people, the people who actually have access to the outside world, there was no process by which, you know, the sensors, like you say, right, they have no way of actually being heard systemically, organizationally from the top it might happen that, you know, she might tell her manager, the manager might tell the manager and whatever, but that's like almost an accident, right? Because you have a chain of where anyone by cutting the signal means the signal doesn't get up, you have like five people, they all have to agree for something. Usually it gets beautified, right, usually, everybody has their own layer of polish. And at the end is nothing to do. What goes on top has nothing to do with what actually started a farm. Even the fact that we think of the tree, right and talk about the bottom, the people have all of the actual, like real world experience is really problematic. And that's basically baked into how companies are structured, there's no way around that problem. If you if you stay with a hierarchy, you just add abstractions and attractive attractions until the management basically has no idea what's happening at the front. And, you know, there's this countermeasures, you know, the Japanese have, like a school of management, around concept, which is like actually get to the frontlines. But, you know, these are counter metrics to to the systemic issue. Okay, so, seeing that I basically was like, Yeah, well, that's, that's completely stupid, and we shouldn't. So we did a few things, actually. So So one of the things that we did is we we structured the information flow the company before we structured the people, right? So instead of saying, like, we're gonna have people, like, a tree of people, it's gonna look roughly like the Prussian army looked like in the century. Because that's exactly what we're doing. Right? Creative, you know, work in the 21st is just about, like, you know, building an army unit and then 19. And so we've said first, like, how does information flow, right, so we have what we call a surface, which is sort of what our customers like everything that we put out there, everything that we need to support everything that we're responsible for. And from that we get to, right, we get all these sensors, like you described, it could be support, it could be from a sales conversation, it could be from social media, it could be from our machines, or servers themselves, like complaining about something going wrong. And even better, when you get the same signal from multiple places. Right, that's kind of now you're starting to get you know, not only you can see something, you can smell it, you can hear it as well, like, that's definitely real. If you're getting it from multiple, multiple sites, and that's where we get patterns, right, like something is happening, which is, like, robust. And that's what drives our brainstorm. So so we, we think of what we call improvements, which we then sort of implement in a somewhat much more standards or continuous, continuous integration, deployment. Agile kind of way, which is the other side of them, right? It's the it's the actuator, right? It's you can you can have a sensor, and you can have like the reasoning, but it has to also be able to affect reality, right? So that's kind of your your implementation deployment arm. And when we're done, we basically change the surface, we're back where we started. Um, which means a couple of things actually means we can actually then walk backwards and go where the original signal started and say, Hey, remember this thing you told us about two months ago? We fixed it. Do you want to see if it works for you? It's there, right? This just blows people's mind when we do that. Like, it's the sort of thing you just can't do, unless you build things exactly.
The other thing is that we can iterate right now we've learned we've made a step so we can see what the next step of feedback is. And that doesn't mean necessarily that we are always acting as a greedy algorithm just like doing one step at a time based on what our customers tell us we can always inject vision we can always just have an idea based on like, a lot of input over you know, a depth of time just something clicks. What it means though, is that when you actually deliver that it's still subject to the same feedback that will make it make sense also for a customer because you're not going to nail it the first time right Steve Jobs did not merely iPhone this copy paste it for they didn't have an app store, right? If you learn these things on on the fly, so by starting with information flow, then we build like a human sort of hierarchy on top of that, which is very, very likely because actually turns out when you have a lot of information structure, you need the less hierarchy, which is counterintuitive, but it's you know, Bitcoin is kind of the extreme like a ton of information structure, no higher or almost, right. So we're not that far away in that in that continuum, but we're in that direction.
Nick Jikomes 19:59
So So you said something I want to try and read reiterate some of that, to help people wrap their heads around it. So you said something interesting. A moment ago, you were talking about how, if you've got multiple different kinds of sensors, that are collecting information through different modalities, that this can help you identify real patterns more robustly, and that immediate. So my background is in neuroscience that immediately just reminded me of something that most neuroscientists are very well aware of, which is the idea of multimodal integration. So one of the key things that our brains do to help us figure out what's actually going on outside of our bodies, and then to build some kind of adaptive response to it. And our behavior is multimodal integration, right? So I have information that comes into my eyeballs, I have information that comes into my ears, I have information that comes in through my fingertips, right, I can touch, I can smell, I can taste, etc, etc. And when all of these things are in agreement, that tells you okay, there's very clearly something out there when all those senses, you know, are congruent. If there's a mismatch, right? If I see something on the corner of my eye, but I turn and it's not there, and I didn't hear anything, and I don't smell anything, so and so forth, it tells you that signal is wrong. So in what sense? Are you building sort of these multimodal sensors into your organization? It sounded like maybe you have ways of doing this with respect to things like customer service. Yep. Yeah, that's,
Alexandros Marinos 21:25
that's exactly the model. And really, it's kind of fascinating, because I spend a lot of time working around ideas on good hearts law, which is like when you get a metric, and then you try to make it a target, it stops being a good metric. So that's kind of the traditional way organizations work, like the Commission's or whatever, that's how they try to steer the their, their, their, their teams in that direction. But these things are fragile. Right? So so I've been always asking myself this question of how does organization solve, you know, difficult sensemaking questions in a in such a polluted environment? And what I came up with is exactly what you say, this concept exists in neuroscience, it exists in astronomy, there's multi messenger astronomy, right? If you see it in the gravity waves, and you see it in life, and you see it in the, in the, just like this, the four fundamental powers basically, forces are telling you the same thing. It's a real thing. But it's so many places that, you know, there's triangulation, in social science, there is I'm forgetting now, there's, I had found so many examples that was like, okay, you know, or hedge funds, right? They use what they call alternative data, right? Like, which is like drones flying over parking lots or like, whatever. And they combine that with other areas. So again, you get this exactly the same pattern. It's like, okay, this isn't by chance, right? This is something that is a deep pattern that exists. In has been discovered over and over again, from nature to various fields, in the real world. So that's exactly the time we're trying to copy, um, to be inspired from and, and yeah, so. So basically, every interaction we have with the outside world is interaction itself, like we try to help a customer support problem. But we also do what we call tear down afterwards, which is like, we try to find what this is an instance of, right? You know, if somebody says, like, how do I secure my application, right? This customer is confused about security. Right? That could happen on Twitter, too, that could happen in our forums, it could happen and whatever. Or it could say, like, your your, your, your pipe there, your your bill, pipeline, whatever, it doesn't really matter. This is slow, right? We could see that from our server logs, we could hear from support something some big customer could call up, you know, their their customer success person to discuss this, like we could, again, it could be popping up all over the place. And the more he pops up, the more we know, we should pay attention, but it's the same same pattern.
Nick Jikomes 23:56
I see. So one of the things that's interesting. So So we've sort of been starting to discuss how you've organized your company. And I want to talk more about that. But there's also a, like a very basic and pervasive thing that you see seemingly across all companies, which is, in the beginning, when they are small. You know, people will describe a surface being agile or nimble, they can move quick, they can do things very quickly and creatively. And as companies grow, in almost all cases, they basically get slower, and they get worse at decision making, at least in some ways. What exactly, so eight, do you think that's true? And B, what is the reason for that in terms of organization? And what does it have to do with things like Dunbar's number that I know you've talked about?
Alexandros Marinos 24:43
So I think it is true, I don't think it's inevitable, or at least I hope it's on note. And the point is, as companies grow you know what when a team is small, they do what we describe, okay? Right, you just hear a bunch of things. Like, there's five people, like, you know, it's a startup in the garage, right? Like that's, you see things, you connect things that you wouldn't have connected. It's not your department, but there is no department. So it's not your department just saw that thing, you might start somebody to talk about it, you may get an idea. That's how it works, right. But as you grow, and this is where the problem is, we, we nobody can know everybody, right? And that's 150. People Dunbar's number, which is like the theoretical, the hypothesized sort of limit for a the, the the early human tribes, like that's was about as big as we hypothesize that could get because that's about how well our brain does in tracking, you know, distinct people and their mental states. So once you once you get beyond that, you start to lose track now, like, there's people who come and go into the company, you don't, you've never really talked to, this thing's happening, you do a thing. And then, you know, you go halfway through and somebody says, No, that other person's working on that don't work on this, you know, there's like information transfer issues, there's coordination issues, and then it starts to feel more like politics, because now you're actually starting to negotiate internally, rather than actually trying to build a thing. Um, and, you know, the organizational structures, we have our basically, our attempts to resolve this issue, I don't think we're doing very well, which is counterintuitive. I think a lot of people will say, you know, like, we have quite advanced sort of management, you know, we've got all the MBAs and all that stuff. Um, but where's the control? You know, how do we know this is working? If we don't know what else there is, we can very well could be that our best companies are working in spite of the structure, not because of it. So so that definitely is a thing. And I think number numbers are coming into it. And I think the escape from that is to take this core dynamic that occurs naturally list small groups, and build infrastructure that build tooling, basically, for the team to be able to continue to operate in that mode. As the team grows without needing to track everybody like to institutionalize that loop. You know, before it kind of fades
Nick Jikomes 27:10
away, I see. So let me see if I, if I've got this, basically, what you're saying is, tribes, small tribes lower than Dunbar's number, which is just this sort of abstract number that's supposed to capture the idea that there's only so many people you can really know or be reasonably good acquaintances with, right? We all have memory limitations, I can't no more than 50 or 100, or 150, or 200, people, whatever it is, as soon as your organization gets bigger than that, you kind of get these silos. And it just fundamentally changes the architecture by which information flows from person to person. So what you're saying is that you hypothesize at least or maybe this is what you do inside your company. You can use tech to build communication tools, such that when the group and the organization gets bigger than Dunbar's number, you can continue to behave like that small tribe in the startup, anyway.
Alexandros Marinos 28:06
Yeah, I mean, something will get lost. Right? The, to me the Yeah, it's a great time, right? The we faced with this dilemma at some point a couple years ago, because I, I kind of saw that some things were going in the traditional direction, right? And, and I kind of said, well, look, I basically show them a diagram, which is exactly the bigger you can be nimble, or you can be big, and it's kind of like a trade off. And I was saying, you know, why can't we use our you know, our size? I think we had just raised around we had, you know, had money in the bank? And we had some some optionality? Why can't we use our newfound funding to become faster, rather than slow? You know, because I even think saying that you want to maintain the original, you know, speed is defeated, you're kind of saying, like, We're either gonna, like, stay as we are, we're gonna get worse. Like, that's kind of what we are playing with, was not good, faster, you know, like, and then I found out some fun, fun factor. It's like, I still need to confirm this, but I choose to believe it was Henry Ford, apparently, in the height of his power. Ford had 10,000 toolmakers, inside Ford, right. Like, these are not people who are making cars, there was people who were making tools for the people who were making car. Right. So this investment in the meta, this investment in how we do things, not just accepting, you know, Google's management playbook in a startup, which is like what everybody does, it makes no sense. But actually just saying, you know, why do we do we do this way? And what is the right way to do it? And investing in improving that is, again, very counterintuitive. It hurts your metrics in the short term, you're not going to look good because that money is not going to, you know, sales or whatever. Um, but in the long run, and this is maybe where the space we're in is helpful because edge computing is so brutal You either, you know, many, or most of the companies really that have tried to do what we do have died. And or if they're large companies that step back, it's kind of very interesting because it requires a lot of integration and just a lot of patience to build the technology just right. So I think that has worked, that's that's where our DNA from what we are working on has helped us adopt this long term mindset. And that has helped us really invest in the actual way that we, that we
Nick Jikomes 30:27
work, I mean, some of what you're saying. So it sort of reminds me the way that many I've met many people, I live in Seattle, after all who have worked at Amazon. And you know, a big thing you always hear about Amazon, when people talk about why it became what it is, is that there's internally this this very big emphasis on internal tooling. So sort of like your Henry Ford example, there's a lot of people building tools for the carmaker so to speak. And a lot of you know, treating internal products used internal tools used by employees as if they are customer facing products and sort of taking the same amount of diligence and, and care that you would take for the customer product for building these products. Is that something you're doing? So maybe just a question that leads into this is, how do you guys like actually just informally communicate in the company are using slack and other tools? Or are you building your own software for doing this in a new kind of way?
Alexandros Marinos 31:30
Yeah, so So that's definitely, to be honest, internally now. But in terms of all your air, your Amazon example, I've taken a very hardline position of if you're not building a product, you're not building anything, right? Like because there's so many people, like I'm gonna build a process, and it's gonna be a wiki. And you know, just gonna ask me and, like, yeah, we've seen this 100 times you build that thing? And then it becomes this haunted mass, but but it's not No, no, no, look over there. You know, like, it's tribal knowledge, it's kind of diffuse, and kind of sort of works, but but we'll rock, right. So if you don't get it to the plateau of being actual product, like with a front end with documentation with whatever it is that it takes to hit a steady state, where you know, it's not gonna start rotting. That's the the level we get a hit. So that's, that's definitely a thing that I have high conviction on. And Amazon is doing a good Java. Beyond beyond that, so it was a second. There was a well, to what
Nick Jikomes 32:33
extent are you guys building your own indication tools?
Alexandros Marinos 32:37
Right, so So in order to do that, what was really interesting to me was that so to build this way of working, right, when I said we structure information, first, initially, we were just integrating tools, right? So we had, you know, something like Slack, something like Salesforce, something like Zendo, something like whatever. And I have two big meetings at 1.1 with the with a customer success sales side, and one with the support side, and I said, Okay, draw me out a diagram with all the that they're using and how they're integrated, I just want to understand because my backgrounds from information systems, so I was getting a, you know, spider sense was tingling, that things were not, things were not amazing. So I saw this sort of spider's nest on both of those conversations. And knowing what I know, from from my background, I knew that this is only going to go one way and we're going to add more stuff, it's gonna get more complicated, get more integrated, it's gonna keep keep breaking. And there's no you know, security implications of this whole thing, or, you know, a nightmare. So, as a result, it was like, Okay, we definitely have to sort of do something here. We try to build our own sort of synchronization infrastructure, keep it minimal, that didn't work, we really try to avoid having to make our own thing. For the longest time. At some point, I was like, Okay, I'm just, I'm just fighting the inevitable here, I might as well just give up and start building. So we we built, we started building something that we call jellyfish, we've actually recently open sourced it. And jellyfish is basically the idea that most of these products have some fundamental functionality to it, they allow you to take your data and sort them, filter them, you know, commute, collaborate over them, permission them, share them, you know, interact with customers about them as etc. But the data items are priceless in this Catholic Sports thread. Salesforce has like an opportunity. Slack has a chat like whatever, but once you like, take that out, what they do around that is roughly very, very similar. And they also don't like you to share data between them right, because that makes them irrelevant, right. So the their business model is seats and seats means you got to be in the product right, which I found very interestingly, was enforcing in bilena a departure structure without us actually wanting to happen, right? Because if you have the high priesthood of Salesforce, they're the people that have access to the data. They have the high priesthood of Zendesk. They're the High Priestess that has that data. So now they talk on behalf of the data, right? Like you have an implicit department, even if you never
Nick Jikomes 35:20
I've seen the same thing. And we're describing it's not. I mean, this is a very typical problem, right, you get this sort of explosion of different tools, you get, I mean, I like your term for it, you get a priesthood, which is the subset of people in the company that knows how to use the tool. And then very quickly, you're in a situation where, okay, I'm working on some problem building some product. But now I've got, you know, five different kinds of information that live inside five different tools. And I am incapable of just going and getting what I already know, I need, I have to like, go and consult these different priestly classes of people, one of which knows the Salesforce database, the other one knows this tool, and it just becomes slow. I mean, it's like quicksand.
Alexandros Marinos 35:58
Yeah, yeah. So so the thing I kind of I was like, Okay, what's Is there any hope in the horizon? Like, it's kind of kind of look forward to something, maybe the situation is bad now, but it'll get better later. And I was like, No, it's just, we're just going to get worse and worse than this. So I kind of just pulled the plug, or maybe plug the project in, and said, Okay, we're gonna build jellyfish. And the idea is, again, that it can accept all sorts of different kinds of data items that we can find it can synchronize with those products as well, it doesn't mean that we're not using GitHub, it doesn't mean we're not using a tool that has some value, it means that we build a one to one synchronization with jellyfish. And then jellyfish is our internal sort of Clearinghouse where we operate as much as possible, we might skip outside, but at least you got a hub and spoke model, right? You got 10 programs, 10 connections to your center, you have 10 programs, and you need to connect to each other, that's 90 connections, right? Like, that's a lot, that's a lot more complicated without a center. So So jellyfish is becoming quickly our, our communications sort of nursing ever Center, or bilena. And I kind of call it, you know, across between Excel and slack. So, something like that.
Nick Jikomes 37:12
And you said, you open sourced it, what would be the motivation for doing that.
Alexandros Marinos 37:18
So for longest time, we're keeping it, we're keeping it to ourselves. And so you know, it could be a lot of potential etc. But that's where the positive something that comes in, and we said, Well, look, we're gonna start talking about how we operate up Atlanta, we consider this to be a magnet, right? But we started to talk being vocal about how we work because we want to attract people that want to work this way. And the magnet has to two poles, right? Like, we also want to tell people who might not be very comfortable working that way that you know, that you might not, you might not want to waste your time applying to find out because if you're looking for a more traditional structure that works for you, that's great. We don't do that. So we wanted to clarify on position, I wanted to transmit that to the world. And then we hurt people with at least some level of interest, it's still very early days. And we said, Well, okay, why? Why are we you know, we kind of challenge our assumption, why are we keeping the other fish closer, right? And said, Well, there's no, there's some, like theoretical benefit, if, like, we make it one day, but like, we could have just put it out there, and maybe somebody else has value from that. And we can create this positive some direction, right? Like, they make it better, we benefit, we make it better than maybe other people can step in. And, you know, we all operate better. So that was kind of the very simple way in which we thought about it, we've done a lot of work is, you know, years and years of work have gone into it, it's still, you know, probably it's not the sort of thing today that we would just take a stand up and, you know, be on your way into two hours. It's not as easy to get started with that fact. But, you know, that's kind of how you get started in these, these feedback loops, right? You put it out there, you hear complaints, you get it, you get you improve it, it'll, it'll get there eventually. But um, we just thought that, you know, this was the kind of binary emotion that we could we could perform that sort of sets all those things in motion, rather than, you know, saying, oh, you know, we'll perfect it internally and when they will release it or whatever, just, let's let's just open it up. It is a you know, there's a effectively alpha right now for for people who are not us, right, we're not running it and currently, we don't have the team sort of to just hand it up. But you know, there's only one way to to get where we want which is to allow people who have the desire to work with I don't know that's not a lot more. You know, like the right thing to do, I think for for for that technology, then Holding on to some sort of a coat but like eventually building some big business on it, who there still might be a business owner, but, you know, keeping the source close not, that just didn't feel like the right way forward.
Nick Jikomes 40:12
So when you're doing a lot of the stuff that's new and takes a lot of time to bake, how do you think about actually tracking success there? So So you also mentioned something earlier than called good arts law, which is that when a measure becomes a target, it ceases to be a good measure. So help us unpack that. So I doubt you're saying that you guys don't use metrics. But in what, how do you actually think about something like this in terms of how you measure and motivate success for something that takes more than just a couple of Sprint's of work to actually get to a product?
Alexandros Marinos 40:50
Yeah, the way we think about metrics is the metrics are good when we're sitting on the same side of the table, right? If we're having a database, let's say, right, we're gonna pull up the metrics, and we're gonna sit there and just look at them and say, What's wrong with the database? What good hearts law sort of forbids you from doing is using metrics as a weapon, especially against people who, you know, it's upon them to do a good job in measurable and non measurable ways. So if if I go to my sales team, and I say, like, You got to get those sales up, you know, or else sort of your your job is in line, or you can just, you know, commissioners or whatever, which we don't, that gives them a very clear message, right. But of all the ways in which they do their work well, there's measurable ways and non measurable ways, right? The measurable ways are what counts, right? So they should do as much as they can there and the non measurables don't care. So they should take energy from the one apply to the other. And what's worse, you might say, okay, not everybody's like that. Right, then everybody's motivated that way. Even if the permission system in place, people, you know, some people will still do their job properly, you have set yourself up to fire those people by the weight, right, you literally will punish them for not doing that. So eventually, you'll get to the point where you only have people that are following that loss. So that's the problem with with with good hearts law, and that's fine. I think metrics use as an argument as a, you know, no context way to make decisions about things are toxic to in, you know, they directly lead to a lot of the ways in which sort of, you know, modern society is breaking down. There's this book called The Tyranny of metrics that I wholeheartedly recommend to people. It has, you know, examples of this same pattern everywhere, you know, from somebody reading somebody going to university state, right, like, what if you've ever chosen a class because he thought it would be an ECA, rather than because you've learned the most, you participated? And you have now optimized the metric of your GPA over the non measurable like, how much did I learn? Right? So we do it all the time. We don't even think about it.
Nick Jikomes 43:15
I see. No, yeah, that's, that's a great example. I think probably everyone has done something like that. Where Yeah, the metric becomes the thing you want, rather than the thing that you started out wanting to get. So you have this interesting parable, I guess, in this essay you have on substack, about something called the anti corruption agency, which I guess is about how incentives can how incentives can drive good people to do bad things. And I think it's related to what you were just describing. So can you unpack that for us? And explain, explain that story?
Alexandros Marinos 43:48
Yeah, yeah. So that's Sorry, I was really happy when I wrote because I think it really captures. It's like a thought experiment, right? That really puts you in a position where you, you kind of see how everything starts to go sideways. So the idea is like, imagine you're this all powerful head of the anti corruption agency or some nation with the fullback, everything is set up for you to succeed, you have a full backing of the politicians from both parties, they really love you, you've got all the budget you want. You've got incredible people in your organization, and somehow, you know, this ever happened, but let's just like set everything up right in the exact perfect way. You've got many you've got a lot of resources, a lot of people in your, in your in your group, they've all they're all, you know, set up, you know, they're really committed to their career and anticorruption that you've got unions pension plans in the holding. Um, and you know, you've got a long string of successes like weeding out various factions, etc. And one day, a scientist shows up in your your office and says, you know, Commander, how to know what the title says, I've got this plan that we can implement. I think in the in the written forms like this button, you can press but it's not really about magic, it's like, we got this one, you know, we're tricks the pilot with you, if we implement it, corruption will be eliminated fully for everybody, forever in our country. Now, it's really interesting that if, and let's just say that the person sees this, we kind of read through the plan, like, you know, this is gonna work, right? They're convinced that this is actually the question is, do they actually press the button, do they actually put in an operation, knowing that at that very moment, their whole agency and all of their, you know, incredible people and all the pension plans, and everything else, is going out the window. Because essentially, what happens when you set up an organization is you have this implication of permanence, right? You assume that this problem will be there forever, that's a core assumption of hiring people and building up their careers and building up with it, you know, your get your promotions, etc. Nobody is trying to get themselves out of a job, right? You actually hear this a lot in organizations that are like really like settled, you hear this a lot. It's like, what are you gonna, like, get us out of a job, like, you know, slow down. Um, but even if, you know, again, everything is certainly they know that the moment they say, Okay, let's do this thing, their whole reason for being goes away. Or even, let's say the budget is how, right like, that's the hook like that. Um, so not only should, like you would think right like that, they would actively be looking for this. So let's say I have an idea. Or you have an idea. And you go to your friend who's like a member of some high ranking position, you're like, oh, you know, I got the plan, like, go to the anti corruption people, they'd love to see this, right, like, everybody directs ideas in that space to those people. But as we discussed, internally, there is a lot of pressure to have a plan. But the plan is like 20 years, right? And then you start to get into this weird thing, like, let's say, nobody hears you, and you go to social media, and you're like, I've got the plan. And we've got to, you know, advocate for the spine, and like, you know, bottom up sort of gets, it gets traction. Now, you are pressured from internal different organizations to stand out and say, Hey, your pseudoscientific, blah, blah, blah, opinions are getting in the way of the hard work we have to do here at the anti corruption agency. You know, we've had a 20 year plan. And you know, we can't hear all these like snake oil sales people on social media. In fact, we got a sensor. Right? So all of that, it's, you can sort of see it and all of the incentives, are you even if you come from the best place, just a little bit of stuff inception, like that's all you need. And we all have a little bit of that. Right? If things are set up that way, you will be forced to that thing. And just the cherry on top. If you don't, you'll be fired. But Mexico
Nick Jikomes 47:59
will? Yeah, I mean, this seems like I mean, this is all fairly intuitive in terms of what the problem is. And it comes from something very fundamental that we all are intimately familiar with, which is, as you said, no one wants to be out of a job. Not only do you want to not be out of a job, you want to, most people don't want to just maintain their job, they want to they want to move up the ranks somehow. So there's this built in Assumption of permanence just sort of baked into everyone's natural tendency for how they want to behave inside of any organization. So how on earth do you get around this problem?
Alexandros Marinos 48:37
Right, so So, to me, and this comes actually, interestingly enough, comes back to what we're talking about Amazon, right before you, to me, specialization is the real enemy, right? Because when you are married, the specific problem you are married to, it's meant to spread, you actually kind of sorta, maybe he won't say it, but you want it to continue. So, or these you act as it right, that's all that matters, it doesn't, you know, it doesn't have to be like conscious, but like, your actions are such that you will keep the problem check, maybe, but you will not actively, desperately look for ways to eliminate it. So the solution to me is to stop at a certain level of specialization where you are still quite a generalist. So what you describe the Amazon, everybody is service builder, thing, Amazon, they want to have that either unit is called a service level, and we call it a product. But it's roughly the same idea, like you are an expert product, when you succeed is when you've brought this problem to a steady state where it's self managed. All the loops are operating within homeostasis. From biology, everything is actually like cells. Now you can move on to your next problem. Your career is not tied to you being needed in that problem. In fact, that's a bad thing. If you constantly are getting pulled in to solve this problem. Maybe you're not doing it very well.
Nick Jikomes 50:00
I love, I love how you brought in homeostasis there. So so the idea is the job description is not, you know, build the blank product, it's to build products, which are does, it's to build products that reach some kind of homeostatic equilibrium such that there's their self managing to as much to the highest extent possible. And that you're saying that solves this problem, because it gets around the the curse of specialization, instead of being focused on building, you know, the blank product to do blank, you're not as focused on building products that become self sustaining. And at the point that one does, you can move on to doing a different deck a different product for a different problem.
Alexandros Marinos 50:41
Solving the problem should be success, right? Obvious, like you know, you succeeding at eliminating an issue. And sometimes it could just be something you might not be a mistake, you might just realize that if we just call something something different, you know, sometimes it's just goes away. Sometimes we've had some obvious realizations that there's something out there to just use and stop working on this massive project, we were, you know, that realization should be should be encouraged, like it should be worth it shouldn't be like, Oh, well, you know, we've got you got a job. Yeah, so so that's how I, I we're not, I can't say we're there yet. But I know that like this is where we're trying to get to where, think of it more like a dojo or like, you know, people who are excited about the craft of building, you know, creating these stable patterns, who are mentoring each other, right, instead of supervising that we're mentoring, we are learning from each other, we want to get better do. And that's a whole different way of interacting, then, you know, the Prussian army system that we were discussing at the beginning.
Nick Jikomes 51:41
So I think I read something in this essay, where you were describing some of the structure of your organization, and you had something in there about like, separate mission statements for different parts of the company. And that struck me is very interesting, because most companies have sort of one, they might have, like, multiple, you know, company values, or whatever. But there's usually one overarching mission statement that everyone is meant to organize around sounded like you guys do something a little bit different. So so how does what is that? And then, you know, I guess the the related question would be, how do you think about something like company culture, generally speaking?
Alexandros Marinos 52:20
Yeah. So, um, yeah. So So basically, what that essay describes is that, you know, polio was struggled with departments, right? Again, like how do we focus people. And if you look, in the literature, there's kind of two ways of separating company, you can do it functionally, like, you know, this is HR, this is marketing, there's engineering, this is manufacturing. Or you can do it by product, right? Like PNG, for instance, has, you know, the, I don't know, the dish show division and the, you know, detergent, and the beauty products, whatever. And they individually, these are very tongue, right. They're almost like separate companies, except for like, some overarching Financial Management and stuff like that. So both of these problems, right on the one, you don't get your your scale benefits, because each one is doing it differently. So learnings are not transferring very well. But you have the autonomy, right? On the other one, you have coherence as in one HR policy, one marketing approach, etc, etc. But you don't have, right when you want to get anything done, you got to get all the people together from all different groups and managers, managers. So this is this is really the fundamental problem like coherence autonomy, right? Like, how do we build something that operate, you know, that makes the most of what we all know, while letting everybody kind of do their thing. So what we decided to do is define this concept of a product. But internal functions would also be refactored as product. So it's kind of like a hybrid. And I think Amazon says, system is not too different in a time where you're like, Okay, well, if we're gonna have to make, you know, tools for our team, we're not just gonna make tools for our team, we're gonna make all we call people unless it's like a little another product we're working on. And in there, there's gonna be more products like how we scheduled support when we use like ministerial algorithms, to sort of find the perfect support schedule, because we have people all over the world, we have all sorts of different things we've built. And again, the idea is, it's a product, it could be that we'd have one customer today. But we always say like, tomorrow, we could open this up, there should be a landing page that you can go and sign up and use it for your company. And that I think, gets people out of the mindset of building things in this sort of parable kind of way, right? Like in this way where all the causal arrows are just kind of intermixed, you don't know where something starts begins with roles, something plays, you're like, Okay, we are, you know, we have to build a proper if you were gonna have an interface, we're gonna have a proper enter, right? If we're not gonna have interface, Fine, let's all work together, but to the degree that we're gonna have one, it's gonna be a decent interface, so that we kind of know, on both sides of that what the expectations are. And I think that's where the separate mission statements come in, because you want to set those individual mission statements or the individual loops to be such that when they add up, you do the thing that you ultimately want to accomplish the company. But you don't want somebody who's working on, you know, tooling for our team to be thinking about edge computing, or it's just not going to be very motivating for them. Right? If they think that the only thing that matters, like the liners, that's computing, and they're working on, you know, scheduling algorithms for support shifts. You know, that's not as sort of doesn't feel like you're making as big contribution as if you're thinking, if your mindset is I'm working on making things more efficient in general. And my first customer is willing, I think that context shift helps a lot. And again, it gives you that sort of product builder mindset, rather than out of like, second class citizen, internal tooling, you know, whatever.
Nick Jikomes 56:20
I see, you know, I think that's interesting. So so the reason you would have different mission statements is fundamentally because it allows you to keep everyone at a similar level level of motivation, because you have that that sort of one client facing mission statement, as you said, you're gonna have a bunch of people that don't really work on the client facing side stuff directly feeling like they're not actually contributing to to the whole. So this is going to reiterate a lot of what you've been talking about, but I pulled out this quote from your essay, and I don't remember if this is from you, or you were quoting someone else. But it said, large organizations of all kinds, from corporations to governments lose their resilience, simply because the feedback mechanisms by which they sense and respond to their environment have to travel through too many layers of distortion delay. So again, this concept of their sensors that are feedback, loops, leaking, linking these things together, and you lose resilience when this breaks down, which is often simply by putting in too many layers in between all the different nodes. And right naturally, as a company grows, you put more and more layers and more and more hierarchy into the organization. And what's interesting about this, to me is, as far as I know, you don't really have a background in biology. But this is very much a lot of this stuff very much reminds me of, of biology, you know, when an organism breaks down, it's for the same kinds of reasons, either the sensor stop working, or the feedback loops that help regulate and coordinate them break down. And I'm wondering, I don't even know how to ask this question quite. But, like, do you think of your organization almost like an organism that's, that's growing and evolving? Is it something that's sort of like you're cultivating like a garden? Or is it something that you're engineering like? Like a device?
Alexandros Marinos 58:10
Hmm, um, look, my intent, like the dream scenario, here is where I, you know, as we do better and better at what we do, I'm less and less relevant. Right, that I've handed over autonomy to more and more people on the team, while we don't lose our parents, that's the plan. So I guess it doesn't look like engineering in the early stages. But once we hit, you know, because we don't have enough time to wait for evolution to make the first self replicator, by chance, right, we have to get there by design. But the idea is to make it so that it can then grow organically and experience sort of evolution can have experimentation. Even this process I just described as kind of like mitosis, right? Like, we have like this one big hairball, and then you like creating like smaller clusters that look like the whole, but in a smaller way, and maybe they themselves will later trigger the same procedure. So I'm very inspired by biology, in fact, that, you know, I don't know if I'm gonna say this, right. But predictive coding, have seen in sort of in neurosciences has very similar things about you know, nested feedback loops as a way of explaining how the brain works. Again, I'm putting that on my desk that that stuff, but I read it and I and I get sort of ideas. And I think it's the reason why you're seeing these analogs is because these are deeper patterns, right? Like these are just on a systems level. It's systems all the way down anywhere no matter what you do. And I think it's kind of like with with with mathematics, category theory is like this universal way of modeling all the other modes of science of mathematics, right. So through category theory being passed through out of a typology and end up in algebra or whatever, because it's like supposed to be the pattern or patterns, right? So I think it's, I check across different fields when some of these ideas start to play. And that's my way of confirming that I've not lost the plot, right? Like, it's kind of a way of performing, concealing some of the higher level of like, do these other fields, which we can even see a sensor we have, they seem the same thing. Right? If these ideas occur, you know, from economics to do biology to, you know, finance, or, I don't know, there's like, all sorts of engineering, like even like mechanical and physical thing, you see the same patterns emerge, you're like, Okay, this is this is a thing, right? I've hit on a real on a real back.
Nick Jikomes 1:00:47
So one thing that I think is interesting to think about is, you know, you can almost think of, you know, an entire nation state, as, you know, a really big startup or a really big company or something. And, of course, it's going to have its own internal structure, it's going to vary from country to country. And I think you, you have to start thinking about some of the same things that one would think about, if you were if you're growing a startup like you are. And you know, one of the things that we've been talking about is how internal structure within an organization either does or does not lead to a breakdown in coherence in, you know, the organization actually achieving what it's supposed to be able to achieve. And to do that by responding as quickly and effectively as possible to what sensing both in its external environment, its internal environment, just like an organism. So, you know, with that in mind, you know, COVID, for a lot of people has been just an interesting experiment, in many ways. And one of the things that we're all watching happen, and more or less real time is very large organizations, government institutions, for example, and how they're behaving and responding and adapting to something that's, you know, evolving in on a day to day basis, which is, you know, just everything to do with the pandemic and the virus. And, you know, there's lots of different viewpoints on this. But there's also I think, a lot of, there's a lot of agreement that we certainly haven't probably responded to things in the best way possible. And, of course, you know, one way to look at this as perhaps that has to do with the internal structure of the institutions that are meant to be responding to everything. So for example, the Center for Disease Control, you know, it's been two years, and we still don't have this thing under control. And I'm just wondering, as sort of this systems thinker and the startup guy who's thought a lot about organizational structure, what's been your view for the last two years on how say, the CDC has been handling things in the US? And how has that view? Do you have any thoughts on how that might be tied to the organizational structure of that institution? Or or?
Alexandros Marinos 1:02:51
Oh, for sure, boy. So one very interesting thing, it's a very good example, because the CDC because it's actually got a dual mission. Right. It's both the kind of the data scientists here, both the people in control the data, but they're also advocates for vaccination. Right? They so their mission is like both to encourage a certain thing. And to tell us what's happening. Can you see where perhaps this could lead to certain distortions and and, and that's why you see, like, these horribly contorted studies coming out about like, how well if you look at like these three people in Kentucky, you know, in the summer of, you know, 2020, like a specific day, like natural immunity is five times worse than vaccination, which is very, very important. Like, just just on the background, like what's, like Extraordinary claims require extraordinary evidence, and what they're bringing up is like, something that looks like, you know, like gerrymander, right, like, and like you have 100 studies on the one side to say one thing, and the CDC is like, publishing into, like, a not a journal or like, an internal thing.
Nick Jikomes 1:03:59
What's tricky is very sort of clearly state for people what you're talking about specifically, yes. Okay. So,
Alexandros Marinos 1:04:06
yeah, so So there is one of the big disputes, in terms of pandemic and sort of the CDC is on one side of that is to say that immunity acquired through infection, right, like, well, what is called natural immunity, and people inject that term and I get why they objected, but it's the easiest term and got that immunity is, is it better or worse than vaccine derived immunity? Right. So if you get vaccinated or if you get infected, like you have two people, right one is, which one's better at training, which one's better off for for the future, like in the next like next time they come in contact with with the virus? Um, and there's very, very large body of literature, a lot of data coming out of Israel, like study the cover millions of people that very clearly state, you know, natural immunity is just quite robust and long lasting and does really, really well. As a community is what it is, right? I'm not saying it doesn't exist. It's not it's not a thing. It is. But it's not the same thing as as, as community, it's becoming quite clear there's, there's numbers we can throw out there, I don't know that it's worth getting specific. The point is that there's one body in the world that's like, respected and putting out studies that say the exact opposite. And that's the CDC, which weirdly has this conflict in its mission of telling us what's going on, but also encouraging explanation. So you can sort of put two and two together and say, like, okay, maybe they're not telling us, you know, everything, maybe they're trying to spear, you know, they're giving us a noble light is going to steer the public, maybe coming from a good place, steer the public towards what they think is, the better, the better behavior, maybe they're worried, for instance, that people, if they believe that they would try to go out and get infected, and they don't want that to happen. So they're trying to sort of distort the information field.
Nick Jikomes 1:05:56
And it was, I mean, we, I think it's quite clear that this does happen. And, you know, it's, um, I would say, it's, it's not obviously right or wrong. Well, you know, I think that's up to the individual decide, but like, you know, remember the beginning of the pandemic, we were told initially, don't buy masks, because they're not going to help you. And, you know, they, they knew that wasn't probably accurate. But the reason they said that it was like a noble lie, right, we wanted to bring supplies for healthcare workers, which is, you know, a perfectly reasonable thought to have, but it meant that the people reporting sort of what's true, were telling us wasn't actually true, because they were also the people reporting to do. Yeah, and it sounds like, that's what you're saying that conflict sort of puts you in the situation where you get to this, the situation where you're being told a noble lie, something it's not actually true? Yeah.
Alexandros Marinos 1:06:50
Exactly. They're incentivized basically, to do that, to the degree that they do, or they don't is, you know, it's hard to know, without discovery. But, um, you know, definitely seemed like that, and, and the fact that they have this sort of admission is ambiguous, to say the least, is definitely like, makes you predict that that's what would happen, it's not surprising. So that's one one kind of thing that definitely happens. And that has to do with lack of clarity of admission, but then once you also see, let's say, the head of the CDC, in the news, saying certain things that's so on our on our board is the is a former CTO of Microsoft. And he told me the best way I have of communicating to have communicating to the rank and file at Microsoft, this talk of the press, right, they don't read the numbers, though, they'll hear it when I talked. So you know, when when the head of the CDC goes on the mass media and says certain things, and knowing that there's like a little bit of an oval live, essentially, culture internally, you might be, you know, led to understand as a rank and file, you know, CDC data analysts that that's what we're looking for, right. And, and your hair, your, your boss might or might not say that explicitly. And their boss might not might or might not say explicitly, but people also gather from the Gestalt, or like, who gets treated in what way for doing what. And before, you know, you just have an organization where I'm not I'm not saying the CDC is like that, but like, deeply suspected. But just abstractly, you will have an organization where data becomes a tool, right? It becomes a, like, an instrumental sort of object to be used to accomplish certain objectives, your initial mission of like, you know, just tell us what's going on. So we can orient becomes, I'm going to tell you what I think you need to know, in order to do the thing I already know you want to do, which becomes really, really troublesome when the situation changes, right? Because then you get caught people remember, right, and then you keep changing your tune, like, Hey, what's going on? I thought you said this other thing. And what's the biggest tragedy, of course, is that you waste the one resource that is even more valuable than PPE, which is institutional trust. Right? And that kind of gets us to where we are today.
Nick Jikomes 1:09:20
So, you know, previously, when we were talking about, you know, little startups, disrupting the big incumbents, you know, it's this, this story that you hear so many times about, you know, the, the little organization that can move fast versus the big organization that has resources and like can have this long term vision, but like, they're just sort of sluggish and slow to adapt. So it's like this long standing issue that that we see over and over again, it's very hard to overcome this. You know, the story of your company is really, you were telling us a story of how you're trying to sort of overcome that tension. But you know, there's no obvious quick solution. So you know, for something at the system societal level like, That's right. How about an organization like the CDC or anything, anything at that level? And you're talking about a problem that, you know, it's not just Countrywide, it's it's globally, you need to, you know, you need to have organizations that are as big and resourceful as governments, but at the same time, because the the virus is mutating so quickly, and things are changing so rapidly, you want to have that nimbleness? Is there just kind of like this hopeless trade off here? Or is there actually a way you can have an organization the size of a government institution respond with the the nimbleness that we actually need it to?
Alexandros Marinos 1:10:37
Yeah, I mean, there's, there's solutions of different timescales, right, like, if you told me, you know, what would you do tomorrow? Where's where would you like to be in 10 years? There's different ways to approach this. But I think the immediate thing is that just encourage feedback loops of information as much as possible, right. So for instance, there is this very disturbing trend of the Empower disempowering the the doctors or nurses or the people who are face to face with the patients, right, they are being told how to do what they need to do and exactly what doing what not to do, being threatened or licensed or being threatened if they prescribe certain drugs. I'm to give you an anon AirMech. Example, Homeric, who is the one of the founders of the frontline, COVID, Caroline's has been banned by his hospital, and is there not a lawsuit, and I think he's been put on administrative leave for the same reason, from prescribing certain drugs to his patients in the ER that he runs, he's the head of the, you know, nothing. So the ICU, right is the director of the ICU, and he's being told you cannot prescribe these drugs. These drugs include fluoxetine, which has fantastic sort of RCTs. But the people who doubt let's say something like I reckon do not doubt, but it's actually like pretty well understood that either neutral, or it helps, but he's been banned from prescribing it, because top down, right, and I dug into the regulations a little bit, there might be some some additions, in terms of how it goes even higher up, like how the authors are being driven through various bonuses to prescribe certain drugs. By, you know, the whoever is running on this DNA surely appears to see, but somewhere from the higher levels of the government, their incentive structure is in such a way that they really don't want people to be using generics. And, you know, you get this thing where you disempower the leaves of your tree, right, you disempower the people who are face to face with trying stuff, just to figure things out, this is how medicine has traditionally worked. The doctors are free to prescribe off off label, right, and then things bubble up. Now we have the inverse, especially in emergencies, this has to happen a lot in these organizations, right, where they just make these likes tight groups, because they need to make make decisions fast, right, and then pose down, but that can continue for a long time. And that's what we're seeing. So we're seeing the opposite, basically, what we would need, which to me is like actually open up and allow a bottom up two types of things. Because we don't have a lot of certainty on desires, like we just don't know, a ton, but we behave like we know. So that's actually that's kind of a lack of humility, that you can discuss it in like this, like, like, like a character trait, but really comes out of this
Nick Jikomes 1:13:34
structure. Yeah, and I think I think there's two things are tied together very closely, this this reliance on top down authority, as opposed to listening, what's coming up from the bottom up, in this case from like, doctors on the ground treating patients, and this conflict that you identified in the CDC, but I mean, this exists everywhere, I see it in my own startup like between those meant to be reporting to everyone like what reality is, versus those that are meant to direct resources based on that information as to what we should be doing. And just, I mean, this is a pervasive, pervasive problem. And it results, as you mentioned earlier in people using data as sort of a tool for whatever ends that are after. So like, in a startup, you know, just to give you some abstract examples, to those listening, you know, I see this all the time in the startup that that I'm involved with. So there's some internal group, they're doing product design and engineering work, they've got their metrics that they want to hit, they've got their thing they want to accomplish. And what you often see is if that team that has to hit those metrics and build the thing is also the team that's meant to be sort of collecting the data to support what they're doing, then what you naturally get is they just fish around for whatever data makes them look better doing the best. So prize. I mean, I just I love this as a very general principle for organizing people whether it's a government organization, a start up your own life in many ways. You know, separation of church and state between, you know, how are you collecting data to define like what's actually out in the world? And then separately? How are you deciding what to do based on that? If you have the people making the second decision, the same people as the ones like collecting and finding the data, then you just get into the self serving forever loop and it breaks down eventually.
Alexandros Marinos 1:15:20
Yeah, for sure. I mean, this is in the rationalist community, they have two different words. For this, you have Instrumental Instrumental rationality, which is what do right what do we do? Like? How do we solve so wrong, we have epistemic rationality, which is what's true. And these are two different disciplines. Because, you know, deception, for instance, pays, it plays very differently into this field. So I definitely have developed this intuition about pandemic that we may apply sort of consequentialist logics to what we do. But we absolutely must be religious, like deontic, sort of, you know, thou shalt moss about information about what is true and sharing, because we don't really know. And by assuming that we know, and therefore starting to distort each other's reality picture of reality, we are almost certainly making it worse, like, we should assume we're making things worse by doing that. Even if there's some short term game, somebody's metric of bumping, it's the you know, the losses are, you know, we're burning down trillions the game.
Nick Jikomes 1:16:22
So, I mean, we've talked a lot about, obviously, organizations, literal organizations, in our conversation, your startup and startups, generally, bigger organizations, like we were just discussing. But a lot of this information, I think, is also useful to people, which is going to be the majority of people who aren't interested in starting their own company, or building some big organization or something like that. So I'm wondering if you could sort of comment on, you know, as a systems thinker, as someone who's thought about organizations and things like this, how has your thinking style how how the strategies you've taken for your own, like personal prevention strategies for things like COVID change over the course of the last two years? Like what what are the things that you've learned, that have caused you to either start doing or stop doing different things? What is your what is your approachment?
Alexandros Marinos 1:17:16
So I think one very interesting feature is viruses by spread, like we were sort of controversial early on, is it gonna mutate around certain, you know, measures that we're taking, it's gonna cetera? Is it gonna there's gonna be immune escape and whatnot. And that seems to be happening. So my mind has been going towards what can we do that is so broad spectrum, that it doesn't matter what the virus is doing? In fact, how can we attack the problem at its root? Right, so we have so many respiratory viruses actually, right now that we think about it, you know, there's the flu, there's norovirus, there's, there's V, there's, you know, COVID now and you you sort of get on the block. We don't even know, apparently, this is again, I learned this reason, there's like three other coronaviruses and we don't even know what we call the cold is like, the bucket.
Nick Jikomes 1:18:09
Yeah, it's actually several different viruses that just give similar symptoms, but the right thing Yeah.
Alexandros Marinos 1:18:15
So in my mind has been a lot like, can we do for air what we've done for water? Like this is a phrase that like echoes in my head a lot, right? Because when we fixed our water supply, we just dealt with a lot of diseases in Wall fell swoop, right? We didn't need to know what they are, but we knew they come through the water, right? Can we fix that? Right? We got good quality water, we're in a good place. Um, but with the air somehow, um, you know, this hasn't logged, you know, like, early COVID. I don't know if you remember where a lot of skiers are gone to the Alps. And we're coming back like deathly ill. And you know, if you think about it, they go to the chalets right like warm air humid or whatever. airtight seal just breathing into it could not design a better environment to get you know, Max viral load. So I've been thinking a lot about you know, there's this variable that we're not been thinking about at all and can we get our air better in our houses and businesses you know, at the Lana we did a summit we made sure there was like, we made a little device which were bite released about sort of shown a very clear in the room, like with the air quality was so that they believe belief of mine basically is in the neighborhood, right? Like people know, you will have information like what's the quality, they will do what they need to do to improve, right? So can we get personal or quality monitors, there's startup there the name of Athina that have a little device if people start pushing you know, in businesses and public spaces where the air quality is bad who improve that air stale, right UV filters as well, stuff like that. I think that stuff the government could do right like they just They're putting out mandates might as well. Yeah. I mean, they're,
Nick Jikomes 1:20:03
I mean, it's actually quite simple when I think about it now, but I never occurred to me. So like we have, right I can log on to the internet and see, like, what the UV index or whatever it is, and anywhere I go on the planet, I mean, here in the Pacific Northwest, right, I can see what the air quality is because, you know, if there's a big fire relatively nearby, but not so big that I noticed it, you know, I can still go find a number that tells me like, oh, wow, the air quality outside is like, not that good. It's filled with smoke. So basically, what you're saying is, Could we somehow bake this into our everyday lives such that, you know, you could look at an app or something that would tell you is the airflow good or bad in this building? is, you know, there are a lot of particles in the air or not, is that basically what you're saying?
Alexandros Marinos 1:20:45
Yeah, yeah, I mean, you're gonna it's just a fairly simple co2 meter something to tell me if the air is moving, we can do particles, you can go much deeper in that. But honestly, the one proxy you want, like, is there a mood, right? Like, people have been breathing in air a lot or not? If that was in our phones, like, if you had a light at the corner, your phone is green, or red, right? Like, or something in between? I think that would drive a lot of behavioral change people and you don't need everybody, you got to have the 5% though, that's going around, like, you know, coming in there, and I do think you'd get a lot of improvement and a lot of droppings, he's like, I have a young young child and like, they bring back whatever, from their, from their, you know, they activities. And I don't think that any of the, you know, preschools or daycares, or whatever, think about equality basically, at all.
Nick Jikomes 1:21:39
Interesting. Yeah. And I mean, I feel I mean, I'm not an expert in this stuff, but I feel like a lot of the tech would be there. Like, I don't think there's a fundamental reason why like phones couldn't start to have co2 detectors that literally, like make a light go on or something?
Alexandros Marinos 1:21:51
No, we just haven't we just haven't thought in that direction. Um, yeah, we haven't been aware of the of the, you know, that the variable for this purple arrow, I don't want to say we're completely unaware. But it's like, it's a Nisha Lee thing, like some people know about it. It's not like a daily concern.
Nick Jikomes 1:22:10
Yeah. But But I also think it's powerful to just alert people to the idea that if you have accurate rate, ways of measuring reality, if you have sensors that will tell you the way things like actually are around you, and you simply provide that information. Even if you do nothing else, people will organically start to experiment with strategies to alter their behavior, whether or not we're talking about SARS, cov, to epidemic stuff, or anything else, if you just sort of display, display the right information, people will will organically come to solutions.
Alexandros Marinos 1:22:41
I think so I think that's yeah, and that's really a big thing for me, instead of like, telling people what to do, I've found that things that work quite well are people that you enable them to do the right thing, you show them reality, and they're like, Hey, I don't like that's not good. These things tend to work quite well. But somehow, again, we are much more about a top down position.
Nick Jikomes 1:23:06
Yeah, I mean, this actually reminds me of something else I know that you've talked about, so you just mentioned enablement versus top down control. And I know that you've said previously things like, you know, there's two basic mindsets with respect to the future of humanity, one where we rely on control systems to keep things stable, and the other where we rely on enablement systems to do what life does, which is grow and grow. So can you unpack that for us and sort of explain what these two mindsets are and who embodies them?
Alexandros Marinos 1:23:35
Yeah, okay, I can I can, I can, I can do my best. This is sort of this. It's an idea I've been playing with. And when I do that, I usually put it out there in big bold letters, so that, you know, any pushback comes out me so I can I can I kind of just don't want the feedback, right? So what if we thought basically, the idea is like, what if we thought about good versus evil as enablement versus control? So there's one type of mindset and I mean, I'm a big Elon Musk, Elon Musk, and so you know, just you know, full disclosure, but like Elon does this a lot, right? He doesn't say like, you have to use electric car because like, here's an awesome like, don't you want to have it even if you don't care about the environment? It's like the fastest car in the world. Don't you want to have this car, you know, like, and somehow that creates momentum and now all the other car companies are again, they are incentivized right like to to move and I saw actually a tweet on the CEO of Ford saying, Yeah, Elon, you did it like you've gotten us all on your on your frequency, which is weird, right? Like five years ago, nobody was being so exclusive. But okay, so that's really be enabled in my mind says basically give people the ability to do the right thing, and they they will. The control mindset is the opposite. And I think if we blow it up to large scale humanity, levels, we come back to sort of positive sum versus zero. So there are, there's one mindset, and a lot of people would have heard of it that, you know, there's only so many resources on the planet, there's only so much we can we can do. There's only so you know, so many people that can be on the planet, we should, you know, maybe, you know, I've heard people, like slightly more extreme, and thing and I won't have children because, you know, the earth can take it or whatever, which is kind of the most Lucien philosophy, right? It goes all the way back to the 19th century, the guy who did like, some estimation about how, you know, in 100 years, the roads will be filled with horse manure, because we're getting more and more horses, therefore, you know, what's gonna happen with turns out we are so so so. And that's actually a very important, dynamic, because what do humans do when we have a problem? Like, we don't just take a linear mouse, like, oh, I guess the poop is like, you know, three floors up? Or like, we do we invent things, right. And that does not conform to linear models, right, we come up with like step changes. So so that's kind of enablement, titled, like, give people the tools, and they will find ways that you can't even imagine like, so it requires you to trust humanity, trust, serendipity, trust the bottom up, which is like, hugely psychologically uncomfortable, right? For a regulator to say, like somebody will come and make something he can't, he can't tell that to Senate. Exactly. Right, you have to be like the man with the plan. Um, and so that's the tension a lot of the time you see with these very, very large government bureaucracies and NGOs, and like this, like this whole complex, right, that has this, this mentality of we got to tell people what to do for their own good. And we're just gonna set things up so that they do. You don't you're not quite, we're not quite at one child policy that China had implemented, but you know, it's things are starting to, you know, look, as if, you know, people right now might not be so full throated and condemning, you know, that kind of policy, they're like, Well, you know, you got to do what you got to do for the earth or whatever. But that, again, that leaves out this whole other pieces, like, invention, innovation, you know, problems created problems resolved, like there is a feedback, there's a homeostasis in commodity itself, right, when something becomes a comfortable enough, enough attention goes to it to make it less uncomfortable. And that's my whole point is that life is about growth, right? No species has decided to self regulate its population, and manage to do that for an extended period of time. So in my mind, it's very, very clear, you're either growing or you're not growing, if you're not going, you're probably dying. And so that's why again, looking at Elon Musk, when he says, you know, we've got to, first of all, he's going out there saying, we don't have enough children, which is completely counter narrative, right? It's like, whoa, where's the, that's what the data says, Forget what the narratives that data actually says. Human population is like, is on the set to decrease. But also, he says, we need more planets, like, let's just go out and get more land. We don't have enough land. Like, that's arguable. I mean, but let's let's, you know, go terraforming the planet. Oh, that's very difficult.
You know, like, it's, it's this try, you know, let's, let's be open to the potential you never know, some people might might like to do that, you know, you might not, but some might. So, this idea of just, you know, living, let live and give others the tools to do things. And that that takes our energy, and instead of directing it at each other to fight to over, you know, limited goods, well, there's only so much we can use to enable each other invent, and to just go out there and take from like, the universe, the solar system and VR. You know, so you know, humanity in 1000 years, will we still be on Earth? And if not, then why not start? You know, going farther out already?
Nick Jikomes 1:29:07
Well, one of the last things I want to ask you, you've mentioned a couple resources, a couple interesting books and things if people are interested in thinking about systems and how to think about metrics and how to think about organizations, either in the abstract or if any of these resources you might point us to are specifically geared towards people interested in startups. Are there any any books are other resources that you want to point people to that might be related to some of the ideas we've discussed?
Alexandros Marinos 1:29:36
Yeah, so one of them is definitely this book. I mentioned the tyranny of ethics, which is not it's a critique, it's not a solution, right? It was just kind of a bit of a problem, but it definitely like opens your eyes to how things are. Okay. Um it's like I can't tell
Nick Jikomes 1:29:55
you what kind of person Oh, it's
Alexandros Marinos 1:29:59
like a It's like a coming from a business perspective, I think actually, it's like a sort of business school, but it wasn't like some sort of spiritual thing or something. It was more like, you know, here's all the examples from like, from, from finance, from the health, from education from whatever, all these things going wrong in this predictable ways.
So that's the, that's definitely a book that. Yeah, that comes up. There's this, I don't really want to call him like,
He is credited with inventing lean manufacturing in Japan actually isn't American called Hope at Washington, W. Edwards Deming. Deming era is systemic, who actually was trying to explain things to people doing manufacturing in the US didn't listen to him. So he went to Japan. And then he found a collaborator who was Japanese on the Gulf, his ideas there, and they lean manufacturing Toyota Production System, all that stuff now is sort of, but then got re exported back to the US. And so he came back and tried to then teach them again, here, but a lot of the thinking of Deming and Russ a caf, as well, a lot of these systems thinkers of like middle 20th century, I think, from from second world war on to the 80s. They've talked a lot about these things. Deming, I don't think has like a definitive book, I'm sure there's a lot of books out there on him. But just looking at some quotes from Deming, just as a golf as well. It's just finding, you know, almost anything that they've written is probably that and that same vein just tells you what their critiques were of the ways that we run our companies. And they were pointing to ways to do things better, I think, where they failed, and I feel really stupid saying that because they made just incredible gobs of progress is in pulling it all together in a sort of coherent way, like how to run a company. But I think that's partly because that had not been we don't we didn't have a complete answer. And part maybe because the technology that was available then maybe did not enable that answer. But um, I guess that's what I'm trying to resolve. And if we share the link to the substack, about Belinda's methods, we will continue to write in that subset called Rural forests.com. We're going to continue to write things in that vein as we, as we continue to either publish things that we're already doing or, you know, grapple with new new problems we haven't gotten into yet and sort of discuss that.
Nick Jikomes 1:32:51
Yeah, I'll be sure to link to that in the episode description in the show notes. But anyways, Alex, thank you for your time. This is a really interesting conversation. I look forward to following you on the internet.
Alexandros Marinos 1:33:02