Specific formulation in terms of how you can compute it.
I totally agree with that. It's
like you read, I don't know count or Descartes, and its philosophy is great. But so what then? How do you turn lots of books into concrete, actionable things that you can maybe measure, can maybe compare. So I think this is the big challenge you have, is first, maybe at some point, stop the heuristic of going to explore all the theories in the world, yeah, yeah, and then pick some and then translate them into something actionable. I think this is the challenge. Yeah,
so that's why I think the rational analysis would be the good place to start, because this is a feasible space. You remember that we had the example of you have this Q and depending on whether it used to be a cube in my YouTube video I shared, do you remember that there is a desirability factor and visibility factor and desirability, if desirability changes, That means that your objective, this way changes to kind of this way, and with the same feasibility space. For instance, here the optimal value was this, but now it became this, whereas if you change the visibility, this is desirability, then this cube, what you do, cube is half, and now this is not visible anymore, so it gets changed to this. So what I feel about the expectation is you are in advance, expanding this feasible space so that later you have more room to pivot to
but this space expansion is not infinite, though, yeah, still obeys some it's still constrained to it's basically derivative to what you already have. This is what I think is interesting in government, this idea that you, you basically, you, you just, yeah, you have to reduce all what is possible, or what is feasible, let's say to already some, some kind of contingencies, and they could be physical, historically, an expectation, But maybe it's not physical. In your case, it's like some kind of skills that you have, resources and so on, and so, yeah,
so what I feel about entrepreneurship, if we are to represent an optimization sense, this is the basic template. And when we write this, a, x equals b, it's like a matrix. And this rows, rows and columns. This column stands for how many variables that you have, decision variables, like who to hire. For instance, there are hierarchies of choices, like for Airbnb, um, there was actually one relevant talk.
Um, uh,
so each of them,
each of this, can become a one variable on the decision variables like this is fitness of the strategy, and that is determined by supply of housing, product and demand for housing. I generally agree that things can be decomposed into supply side and a demand side. And there is a product, and with this product, there are platform to search for roommates, peer to peer payment. This are the factors that are not relevant to supply, either supply or demand right? And for demand is conference in town or need for cheap accommodation is purely for demand side that is relevantly independent from supply of housing. So everything should be met in order for we to say that, yes, Airbnb business model works. So my point is this is more on a desirability side, meaning when we say desirability is more like here, so elicitation of utility given environment and organizing of visibility. So
wait a second, because you say a lot of things, I want to be sure I follow first. How much time do you have today? Just to be I have a lot of so we can, we can work now and then, at some point, we can have lunch. Yeah, would that be okay with you? Okay, good. So you first you talked about rational analysis. So what is rational analysis? So the definition is the context. What is the context of that?
So the definition is, there exists a system where their behavior, observed behavior, can be analyzed in an optimal way through
model. Yeah. Okay. So there is, there is a model that is that is capable of describing what is happening in the system, yeah, even
though things that people call irrational behavior is because the physical space is limited to the cognitive limitation. So that's why we act.
Okay, so there is this assumption that you can model the system at hand. There
exists some systems where observed behavior is optimal.
Okay, so this is already we have, this prerequisite, this kind of first thing we have. So it means that here, this space, here, you assume you can describe all of it, yes. So you can describe everything that is feasible, so that the thing that is desirable. So, what would be something desirable? For example, it's like
just to avoid for future confusion. So this is like, there is equals one, and kind of everything in here is t so your next step. It's temporal, okay, I'm trying to say that this can be synthesized with government theory of adjacent possibilities, because based on what your state and action is currently, your perception of it, you can develop what your next environment would be. So
if we would use this formula here, we would apply it to protein evolution, we might be able to predict what are the future proteins? Because, I understand, Kaufman says, says that it's not possible to do that. He says that actually, for proteins, you, if you do the math, this is not how they evolve. They they like there is not enough time and space and energy in the universe to do that. Basically,
yeah, I don't think impossibility is a reason why we shouldn't do that. And I think finding a way to at most, because one of the tempo valiant talk was about heavier than air flight, like, how this can became a possibility. And
so you are, you are in the, in the, in the conference, yeah, just back right yesterday, yeah. How was it
that was great. I got feedback from 13 different scholars on what I'm trying
to do. It's, it's funny. I mean, you were supposed to go to the conference anyways, right? Or you decided to go because he was there. Ah,
yeah, I was it's 10,000 people come to the conference. So it's another coincidence, good,
but it's good that we converge a bit in before, so at least you could read some of the thing, yeah, read Kaufman, and then meet him and then maybe help you to ask more targeted questions and so on. So
he works with Todd zanger, and he came to Bayesian entrepreneurship. So
that's good is a related field, because he's not biologist like Coffman is like thinker. I know, I know this guy is.
He resonated a lot with what I'm trying to do because he was in similar situation. He has, like, a broad interest, and Todd said he is one of the best broad reader. And what he said was Andrew Gellman and Josh turnbound, if I could help them in my comedy like that can be a very great cross pollination, because there are problems that he and Todd cannot solve
yet. Some time to spend with this guy. Or yeah, I had lunch with him. He was welcoming. It was nice with you, yeah? Well, that's good, yeah, I'm happy to hear thanks. Yes, good. Okay, so let's go back to this. So rationale, you said this thing is Russian, Russia, and it comes from the 50s. Ah,
I don't think so. This is that. There was a slide by Tom Tom Griffiths, and he was explaining that.
Okay, good. So then explain to me this again, the cubes
this thing? Yeah,
I'm sorry if I ask the same question many times, but I want to be sure to to understand what you what you mean, because I think it's part of the challenge is to be clear on what we are doing here.
I'm just trying to see whether, like, we're delving into two, whether we're doing a bottleneck breaking operation.
Because, to be honest, I don't have a good prior on expectation yet, and I think introducing biology at this point too much might be a little confusing, because what they are trying to do is, the question I ask is Todd and zangie, Todd and tepo is trying to do the model based Thinking that entrepreneurs should be more scientific, and we should have causal logic that entrepreneurs should be able to build, and very good. And that's what kind of formal modeling is trying to do, assistive dynamics people have been trying to do. So I think bridging the two field is the top priority.
The two field would be which field, the Bayesian and the entrepreneurship. The two fields you refer to, yes, is that your slides? Yeah,
this entrepreneurship field and this field
and and and this field is the people interested also, it kind of of predictive entrepreneurship, like
they're just burgeoning, like they don't have a formal model yet. They it's kind of imagine if, kind of like undergrad, first year students go out in the entrepreneurship world and do some random experiments without week enough prior, they just think they're right without design very noisy experiments, meaning, like there are some, we had this fitness of strategy, and if you design the experiment very scientifically. You should make some like by interacting with you, you are kind of the customer of my theory, right? For instance,
totally. And
by having a conversation with you, I learned many different things, like you're the customer in like biology field, and what are the need from that field that needs like, what are the theories they need in order to understand the startup biology, for instance? And there is a product field like, if I develop this theory supply side, all this theory is wrapped into a pivot game. And what should that feature be? And I kind of have discussion with you on that and that would inform this. So designing an experiment, including persuasion or any conversation, should be aware of there are different factors that determine the final measure of how like, willingness to pay something like that. Like, for instance, how much are you willing to help me do this after our talk, is a output of what are some visible space out there? And how can this wrap this product, and how much need there is in biology space of this product. So to understand this is not a, I think, straightforward thing to do without a formal education or a modeling. And what I think is needed in the entrepreneurship and what they're beginning to discuss is how to design a causal logic for that. And one example is Oconee University, has
you said that rational analysis is Who?
Who? May I come back to you? Yes,
yeah, they are developing this causal logic, for instance, for Toyota, what are some multi tech powertrain is affected by environmental impact and lithium storage and hybrid technology, different components need to be mapped into their own customized logic. Actually, this line of people have just written a scientific entrepreneurship, science scientific entrepreneurship,
I see
tapo. Was very excited about this research scientific approach to entrepreneurship.
Yeah, I think we should start from this paper, scientific method for startups. I'll take because it's collaboration between the Bocconi University people who's developing this tool and using this for for the past eight years to do the randomized control trial and tempo and Todd they're in use.
It's a tempo article, okay? Todd. Zinger, so DoD Zenger is in London. No where is it? In Utah.
So let me try to explain. Tepo and Todd are very close together. And our phone, so it's rubber covid university.
You know these people? I know all of them. Yeah, yeah. So
there's three professors. There
is a bias business school. Yeah, it's so cool. But I think it's a it's a coincidence. No,
no, I think it is from London. They. I even visited there too. So, long story short, there are some converging efforts within entrepreneurship world, and I think just letting Scott know, because Scott is leading that convergence, and Scott went to visit Utah, like two weeks before, to have a conversation with Scott. Still,
yeah, Scott, yeah. You know his face because you met him, but I never Yeah,
I know, sorry to just, it's,
it's fine. I usually, I have, I have a good enough bandwidth, but
just because my network right, and
it's also part of it, I can, I can reassemble the parts I don't, I don't know, especially all of this. I'm not a specialist of this, but because I have PhD in computer science, I have a ground foundation for all of this. But then, because I use some of this model as libraries when I code and stuff, so I know the history of AI, so I can recompose this pretty much. I'm not specialist of probability, or like Bayesian kind of because, if I understand correctly, it's not probabilistic Bayesian, it's more like correction mechanism. It's not probability per se, because in bias, you don't read it. Theoretically, you don't predict. In Bayesian, you cannot know the future, but you can, you can error correct, basically some models from what I understand, maybe I'm wrong, but and then evolutionary, I start to understand from the point of view of startup, and from the point of view also what you mean by evolutionary, which I think is very related With time, based also from what I understand. So one thing I wanted to say, because you said at some at some point, that you have already so much work to do to bring scientific approach coming from your bison community to the entrepreneurship community, that bringing evolutionary from biology of man, exactation is already a lot, and you don't know if you do that so, but if you're still interested by pivot, which was the first thing that also connected why we are collaborating, I think there are still value, though, to take pivot from biology, from exaptation, like, and maybe all kind of other pivot, like A Greek collage, like, you know, tinkering, like the like Fab Labs. Because if there is one thing that is not very clear, the design thinking model, which I must say, is a heuristic that, to my knowledge, never have been very much demonstrated by research. And I, I know part of this research because I was working with HPI Germany. They are in Stanford and in Germany, they are the top of design thinking research. It's still at some at some point, David Kelly that said, this is what is startup and and design thinking, this, this and this. It's as much as a marketing claim than something that it's not like you did 30 years of research and you're like, Okay, this is what our startups. It's coming from practitioner in the field. So, so yeah, so what I can help you with this for pivot, is that if you do a game, I can help you tune the context of the game so it's realistic and it if you need some random aspect in the game or some, let's say, perturbation, I can help you have it so it's similar to what I know from the theory. Yeah, so it will make the game meaningful. And I think this idea to adapt a theory that is out of soil meaningful to a concrete place is very important, because this is the biggest challenge, I would even say, drawback of this formal method. And I know it also from computer science, because there is the same divide in computer science. You have the people from formal methods, and they do this beautiful algorithms it's great. It works in their computer, but then you put it in the field and it's it's not working, and they're like, yes, but it's because you haven't adapted good our model to people in the field. They don't care if we didn't have adapted the model correctly or not. It's because sometimes, in reality, the model has some limitation that usually comes from some priors or some first assumption, and then you say, Yeah, but it's normal, because in our model, we don't have this and that. But the people in the field startup, they don't give a shit for that, yeah, of your thing. So sorry to say, but I could help to ground your pivot game, I think, without having you trying to bind evolutionary entrepreneurship to what is already there, I think
so first of all, I think, yeah, this was, I would just start from here. This was the supply chain I drew and introduced to Abdullah and Charlie. And it takes, like, 10 years to even launch this suppression of what, just to have the context supply chain of tool based research, okay, and so there is much, like, simpler version than this. But please bear with me.
I bear with you, yeah, because I see all the people you want on the 10th of September, yeah, I would be here, by the way. Oh, cool. Just so, you know,
um, so Abdullah is interested in solution based science.
Tell me where, what is this field, Abdullah?
Is that amazing?
Yes, I know. Sandy pants. Okay, so network science, okay, got it.
Um, but he's interested in integrative experiment design. Got it. In other words, prediction accuracy in yet unseen situation under intervention. That's the definition. That's
interesting. That's the unseen. Yeah, that's not a pure we can compute everything, and we know what's going to happen. So there is some kind of uncertainty. And okay,
and that can be supplied. So this template comes from system dynamics, top management model, and this is applied line, and this is inventory. And so what this is trying to say is input output for their solution based science can go into the Scott's Bayesian entrepreneurship academic field, and this can be in the short term, like benchmarking, simulation tool is what we can build. And I think what you said about pivoting game is in this bucket, because the cash has some tool to write the probabilistic green, and Abdullah has some experience in assembling this in a tool. And with that, this is the supply side, and this is more demand side. So Charlie has Charlie. Charlie has some experience in like, he has more than 10 case studies.
Where is Charlie here?
Charlie's here, and Charlie's also here. He's both concerned supplier and Scott is in here. So, so,
so then I would be helping here, though, if I would have my picture, I could help to tune this game. That's
what I was thinking pivoting game, and that is to say that I don't think evolutionary should be like that was related to what Charlie asked the other day when you shared your comments.
Yeah, I prefer to talk with you first, because I don't want to overload Charlie with information. Because, yeah, you already supply him, yeah, with lots of things to think, yeah, that's and I want to be very mindful of his own. I don't want to confuse him. I would like to, if we contribute to something from my field, he would know exactly what it is. You will see the value and it's, it's not all the time, basically,
yeah, because that is needed, because what Charlie has been doing is evolutionary part, and in order to bridge like, it happens to be that Scott and Charlie are in the very opposite of the three by three. So I need, I would, see some evolutionary concepts in order to bridge them? Good.
So I can help you to to compute the literature, and we can discuss together so how to identify them.
Yeah. Have you seen this table? Yes, I
saw. I know how you made it also, oh yeah, got it, and we discussed it last meeting together. So I like it. Very good. Table, thanks.
Yeah, professors in a UN conference was confused why they were kind of classified in that cell when I persuaded them. Like, for instance, there are a paper where defines one firm's entry and exit, and within that many ideas, how they compete with each other. But that's kind of hybrid, because it's the idea and the individual bio and bead and bio based combined conditioner on one bio, how the beat interact with each other. But long story short, I identified that I got an agreement that this and this and this is all m to n to k relationship. It's not very fully nested in each other,
which is fine, yeah.
And Charlie asked with this example, how the Bayesian and evolutionary differ, and I think that's relevant to your comment on the model based thinking, because the gap between the theory and the practice what you just mentioned like it's not about the product themselves, but how people use them, meaning the model is useful. But when theorists are making this model, they should be aware of whether some limitation, and should kind of share that practitioners that this is not the end model that can tell you everything. You should build it up. And that is the movement that Bayesian people have, trans transition from Bayesian inference tool to Bayesian workflow, meaning, how do you start from a initial model and gradually build them up?
Yes, and how also the model is capable of acquiring new informations, yeah, not only to change itself as a model, to better the model, but also in operation, to be able to be refined by actual data, maybe sensors, or whatever like to be situated, which is usually where there is a problem, because it's very hard to to feed this into the into The model. You can generate random like, you know that data sets, but to a real world model, real world information, it's a bit more complex. Yeah. So okay,
so this is
Andrew Gelman is writing. So that was the scientist thing. Maybe it might be helpful if I just go over the people so that I can introduce her face.
So this is Andrew, Andrew Galman, yeah. Man, yeah, and
your Galman is one of the most famous painting statistics.
Why is he famous? He wrote a
book Bayesian data analysis, and I translated his book in Korean.
And is he involved in Stan?
Yeah, he's a developer. He's a father of Stan. He's the
father, yeah. He created Yeah. And Stan is one of the most famous, popular framework for Bayesian Yeah. Okay,
good. And he is tom tom Federman, and he Yeah,
and f, f,
i, D, D, D, A, D, D, maybe, right. Vensim, CTO, vencima.
No, it's in French. How do
you say Vincent? Um, v, e, n, s, I m, CTO, Tom fiddler. I am
Vincent, the system software,
Vincent. I see diamond. Okay,
so he and Andrew's goal is to,
yeah, he has a nice blog
meet. Okay,
I'm sorry to overload all this, because I think this would be, yeah, I think in order to make a pivot game, well, this is our final product, their help, and both their help is needed. And I think evolutionary would be the last piece to be assembled in
here. So this equation again, where does it come from? It's a rational analysis. This
is from Josh lecture. Josh, Josh Tenenbaum, Josh, there was a How To Grow brain from mine, I recommend that how to
equal this equation, rational analysis equation, yeah,
it's hmm.
Is MIT, yeah, I'm
taking his class next semester.
So that's good, yeah, so it's very good. You can start from something that, did he publish this?
He said he is inspired from the AI textbook Russell and Warburg is one of the most excited.
Yeah, I know Peter Novi, and, of course, Russell, I mean, so
I looked up there, and this is not explicitly written here, so maybe we can just cite here. He doesn't have
a citation. So this book, yeah, yeah, you read it, yeah,
but not fully.
No, you have been trained in computer science, right? Yeah, Python, yeah. Um, yes, my point. Okay, yeah, data analysis.
I wanted to read this book that is most relevant to what I'm trying to do. Because, like, Do you have any recommendation on where I should kind of the chapter so I can start from
for this book? Yeah, because
Josh is also writing a book on, like, Bayesian models of cognition, reverse engineering the mind with Thomas Griffith, is the one who you asked previously. And I think this should be where I kind of grow my work on in defining entrepreneurship mindset
at Princeton. This guy, yeah,
and he has collaborated with Abdullah,
so there's connection.
Okay? So many people? No, I don't, I cannot answer this question now of where you have to start from this. I think you have to skim through all the books and and then look at what is good for you. I mean, obviously you see
talk about games. For me, planning, reasoning and acting were most interesting thing, not
machine learning all the Yeah, I know them already. So okay, I Okay. So what you don't know, or you know what?
Yeah, what I know less as multi agent, because you see
here, this is very interesting. There is nothing evolutionary here.
I think multi agent decision making is, includes evolutionary,
yeah, but evolutionary as temporal, like I know, for example, just to give you another point of view. So Rob Brooks, you know this guy? You heard about him. He was the head of CSAIL. It was, it was like Daniela Rose is now, okay? So it was sorry, it's French. So is this? Yes, this paper called elephants don't play chess. So basically, what he argued, he was the first critic of planning in robotics, meaning that you can anticipate with a model exactly what you tell me here. You can know in advance what's going to happen in the future based on correct premises. He said, All this model they failed. This is why NASA is sending bots to space all the time. They are stuck, because they have this predictive model, and then there is a rock never seen in their life that is blocking the million dollar, that's a thing. So he has a theory of he created a company you might know, called iRobot. They do the Roomba, for example, you know the computer so that are based inversely. It's the context that tells to the robot what to do. So there is very little planning. The robots are very stupid. So he's not making it intelligent machine is making very stupid machine. So it's an alternative to trying to model and to have the sense that there is rationality everywhere and causality,
basically. So he's intentionally designed this stupid in order to Yes, just say that nothing is predictable. No,
no to say that, because it's so hard to predict, it's better to and it's hard. It's, it's, it's harder to predict as time goes by. So it's, I can predict when this thing will fall into my hand, because it's few microseconds, but where this will be in 10,000 year, I cannot predict, because I don't even know if the Earth is still going to be here. So, so because of that, he said, It's better to have very, very short loop, very situated prediction system, very simple. And then the robots, they learn how to adapt very, very quickly, and they do lots of iteration of adaptation, maybe millions of them. Yeah, compared to pivot, where, every six months my company will pivot, his theory will be like every minute the company is pivoting and is changing his course of action. So just to give you, in robotics, there is also alternative to that. So
I don't necessarily think is alternative. I think they're trying to converge. So this diagram was from the master of algorithm, and I feel that there are say
it again from what is this master algorithms book from Master
Algorithm by Pedro Domingos. Do you remember the PDF I shared with you last time? Dominguez
commingos, Pedro Domingos. And this is based on on studies and research. Yeah,
is Washington University,
okay? Professor, okay, okay,
so my point is, this is just to this evolutionary and Bayesian, maybe two different representation of what's out there, and the
families, these are their families, how they're trying
to converge or interact
or find that place in the family, okay?
And also in system dynamics, there is a saying that you need to make both agent based modeling and compartment based modeling in order to verify which model
work, to put them in competition to
No, not competition if they're consistent with each other. Okay, that is that somehow to make them external valid, it's
more robust. Okay, got it.
So my point is, are ultimately the entrepreneur learning that I think we should do is that's part of the reason why I said which had which tool as the presentation title is there are different tools that we can use, but maybe simulate them all. And simulating them all forcefully can give us some information on how much modeling that we are doing is correct. It's like, I feel the internal isn't there inertia navigator, like without seeing the outside, just by comparing using different methods, there is some more information to be derived from.
You know what? I say it again? No,
just say what you just said. So
instead of given that, we can always get more data, but we want to know whether the different tools that we have just by using different tools in order to solve this problem and observing the different results or similar results that this tool is giving, that gives some more information on what's out there. And I feel if like inertia based navigator can be something I would
as a metaphor, no
like the mechanism can actually help, is what I think. But do you know about them? I've read about them at some point, but
you're thinking about
Imus, right? I don't know the abbreviation, but like
inertial measurement units, which it's a device that help you to navigate and that give you different kind of sensor information, let's say acceleration, orientation, like, different kind of rotation, this kind of thing.
Yeah, because I heard there was a story in like 18th century that Hamilton had developed this measuring the latitude.
They know that Hamiltonian
there was a
but basically what this story is about is England, the government has made a competition of, how can we measure the longitude, not latitude? Longitude? Longitude? Is this one? One of them? Yeah, this one. And He came up with a device that does not use the polar system, the star system, and I think this may be somewhat like starting off the inertia. I'm not sure, but my point is, there are different ways to know where latitude and longitude you are, and just having different tools in your toolbox and comparing them can give you more information on how well I'm my model and my tools are,
I can, I can show you one thing, yeah, so, you know I'm a diplomat. Okay, you don't know that. You have
a lot of hats. Yes, you
don't know that.
I think I yeah, I told
you. I told you. So I was, last week, I was in Israel, there is a war there. Okay, you follow the news sometimes that there is a war in the Middle East. Now, okay, so you know about this. So it's called GPS jam. So these are GPS jammers, a jamming device. You might know what it is. It's a box that prevents GPS to work. So when you have a GPS, normally, you know where you are. But now today, it's every day. Today in Europe, this is where the GPS is not working. So you can observe something very interesting in terms of security, from Finland to Egypt, there is a huge line now between Russia and Europe, where GPs are not working.
So this green is working, green
is working, and then red is not working at all, and yellow is half working. Is like the interference, because it's statistics. So what it means? It means now that planes, when they, let's zoom to so this is, this is Israel. This is Lebanon. This is Egypt. This is Cyprus. So when you take a plane from Europe and you go to Israel, you fly like you fly. You don't go to you don't this is Turkey. Just you see the zone. This is Turkey and Greece. So you don't go to Turkey. So you go here, you fly, and then when you arrive here, suddenly you're playing pilot. You want to go here, but you know that your GPS will not work. So now playing pilots, and it's the same. Actually, I was, I was talking to zafa husband, because he's a plane pilot. Oh yes, in Iceland from the Icelandic company. And basically Iceland is over here. So they go from Reykjavik, and they have to go, for example, to Finland, to Tampere, to Helsinki. They want to go to Helsinki here jammed by the Russian. The Russian are jamming European country. So that's pretty serious. Nobody's talking about it. So now the plane pilot, when they arrive here, they have to change. They cannot trust their GPS, because the GSA said that they are in St Petersburg. So you can where in St Petersburg? Yeah. So this
is all by Russia. Yes, I thought this countries were no, no trying to do this in order. So, so for the enemies not to know their location,
this is Ukraine. So all of this, it's not even on the map, because it's war. So, but this is Russia jamming. And here, for example, this is Israel, Lebanon, and this is Iran. This is Tehran. And it's jam for many reasons. It's jammed because now missiles, when you send them, they have GPS, so the missile know where to go because of GPS. So if you jam the GPS, you protect yourself, not to have the thing. But back to my example of the plane pilots. When they go they want to go to Helsinki, when they arrive here, then they cannot fly with GPS, so they have to go back to map and to prediction, to integrate the next position based on the map and the compass. So your company is very good in this in this context, because new plane pilots, they have been trained to rely on GPS, but all the plane pilots, they know how to drive their plane with other tool. Yeah. So what you argue for is multiplicity of tool. Yellow Line mean that this is so I think it's an example of what is happening now for that, if it's if it's relevant, if
we because, since Scott is also using the compass metaphor, and I said, in order for campus to compass work, well, you should at least have a magnetic field that is unified across different people. And the reason we can discuss this at least, is we have a map. We have a unified reference representation, reference map of how the work looks like,
latitude and the longitude. Yeah, and
I feel with so the whole reason behind that we need a formal modeling is because with some representation that everyone, everyone, including startups and investors and like suppliers for the startups and mentors and he students working in entrepreneurship for to start with some model. So we were like comparing health policy versus entrepreneurship the other day with system dynamics, modelers and the different major differences here we need, we have at least SR model for people to start from,
yeah, it's, it's like in the 80s, the critic of Wonder books, of traditional AI, the planning AI, in order to be in a planning process, you need a plan, yeah, and this, it's true, and the plan usually is a sequence of action. And this could be a special map. It's like, you can navigate only if you know some kind of a route, but the route is on some kind of space to do the route. So this is what you argue. You want to create a common reference where different kind of algorithm, or let's say process, can help you go on it, but then and then people can can interact with that. Okay, makes sense? I
think we discussed at the end of our meeting last time, like toolboxes and simulate what tool to use based on the test, quantity based choice. Do you remember swiss army knife? Yeah, so that's the ultimate thing that I can I want to build. So maybe this is the first one we can start from, a pivoting game. And I think my PhD friend who's taking both Josh's class and entrepreneur strategy, and he is precautious. Students. Precautious is Josh Steven, another professor. So list
students.
These slides are for what this is for. In general,
no. So Charlie suggested to give a presentation to this
for the 10th. So these are the slides for the 10. Yeah. So you already have, like, one month in advance. You already have your slides, almost three weeks. Three weeks, yeah, sorry, because this is very important to me, yeah. So it's also a conductor of your line of thought, yeah, it's a way to linearize your thought. Okay, good.
But um, so now I'm on this corner and like the cash and so they are. They want to end so this is kind of what their interest is and what their what the tool they have. So Charlie has, and Charlie and Scott has a goal of making operations and strategies of entrepreneurs using nail scale sale and entrepreneur compass and Abdullah and Matt wants to make learning and social science more cumulative by design experience integratively and by constructing theory programmatically and Andrew and Tom system dynamics. People and Bayesian workflow, people is interested in making simulation based experiments in Bayesian workflow more automatable and more customizable. And I think that customizable piece is what I what led to me here,
because you could, you could see exaptation as customization, just just, this is my, this is a, this is a, yes, I can show you my PhD advisor. So I told you about wonderlikowski. She's a teacher here. That what's his name, Eric, I think he knows. He knows Wanda, very, very well. She's, she's faculty here in E, 62, 1418, so on the other side, yeah, right, if I'm correct. And so she was the advisor of my PhD, my PhD advisor, Wendy, it totally does exactly, but I tell you again so we re situate the context. So this is Wendy, my PhD advisor, and her PhD was
a PhD. I like your kind of research.
Yeah, it's important to to be to be sure of the context, because I might have told you this. You have a big bottle. I'm
very lazy to go.
Let's have some water. So a PhD, which, which I was very influenced by. So she started by working on behaviorism in psychology. Okay, she was working on pigeons, of what of the pigeon agents? And pigeons is very interesting, and it's also remark I wanted to do. There is this military guy from the US called Gibson. Is a very important guy, and I will tell you about about him later, that I want to mention him now, James, Jake Gibson, is an American psychologist from the US Army that worked on visual perception. He coined the term affordances. He heard about this concept of affordances, which is very important in design. And he worked, he worked on visual perception for plane pilots, and this is why they worked. Also lots of how birds fly and how they view the world while flying. It was very researched by them. So the book The perception of the visual world, which is a very, very good book, it's about visual perception and processing by the brain, basically. So why is important? Because he was working with the USAF, US Air Force when they created the first virtual reality headset. And they were in this world, the people from USA, they were engineers. They were like, okay, the world is probability. We would put a headset, and we would create a Air Force pilot driving thing where you will not see reality now. You will only see a reconstructed reality through all our models and all the plane they crashed, and they were forced to bring back the context and the physical world, because there are things in the physical world that the pilot they need that escapes the model
and physical thing.
Viewing the real world, they couldn't explain what it is, but without seeing the real world overlay with the simulation by the plane, they had to to have also peripheral awareness, other kind of information that if cut from the world, it was not possible for them to fly. It was, it was a big trauma for the engineers of yourself, because they thought that they would have mastered the question and predict everything I'm
reading listening to this book, upper history of intelligence. And one comment from there was, when we open our eyes, we condition our thoughts based on what we see. So there is some conditional distribution going on. So
maybe it's linked to that. But basically, you see a particular interest to him was the effect flying an aircraft had on visual perception with the Army Air Force, basically because he was also military guy. So he's a very important person, also as an example of and his theory, it's called City no ecological perception,
by the way, I'm gonna share our like transcript, so don't put too much pressure on writing down everything.
Oh yeah,
I'll share it with you.
Okay? With otter? Yeah, otter.ai. Whisper, so this is a theory that at some point it could be interesting for you as an alternative also and also for the game. Maybe it could be one of the Yeah, that's interesting. How, how could we put this theory people in the game? Maybe they could be character. No, you know, like,
What do you mean, like,
scottstown, like Charlie, it's a game. We can do what we want, right? Yeah? So we could have like, different kind of perspective, yeah? And then you could, you could basically, like, lens or hat. You're talking about hats, yeah, lens, because hat you put on your head, yeah, it's great, yeah, but, but lens you can, yeah,
that's maybe you can that was a separate comment in Bayesian workflow, the Charlie's lens, yeah. So when we were building this, um, there were different versions of Aki virgin, Andrew virgin and Paul version, because they like AKI is more computational based. Person looks romantic. So, yeah, we were talking about so Andrew's greatest interest is how to make this more incremental and gradual, so that, because this is too overwhelming for people who are just developing, I have two data, 1x and Y is like two, and you can just write like of 2x then you discover sometimes it can be Three, and you develop this into y, follows normal of like x and sigma one, something like this. You are gradually, as you see, more and more data, like two became four, then I would say, oh, maybe this 2x sorry, 2x I'm lowering this to 0.1 because got it so. How do you gradually help people design this? When the system designer has this in mind, but they don't want to show this
at first? No, it is the same that, let's say you play a video game Exactly. Sometimes you have some kind of a map, but it's, it's very rough, and it's once you have finished to play all the game that maybe you can reconstruct this, yeah, a posteriori, not a priori.
So could you tell me more about, like, acceptation as a customization? Yes.
So, because you asked about pigeons, this is why you wanted to take a picture. This is why I dig into that. So the the the customization is here with a PhD, by
the way, like, since you mentioned vision, this was the main picture for my PhD, entrepreneurship purpose. Nice.
And you see, there is still the real world. Would you? Would you consider driving a car where the wind totally black, and you would only rely on digital representation that the computer is doing.
To be honest, I'm relatively okay with that, but Okay, interesting, because
this is, this is the to be totally cut from only trust the mediation. So
I feel there's a difference between whether you see driving as a mean or kind of ends, like some of my friends who really enjoy driving would like, have you, you
would find to press a button and to go to school. Yeah. So you're not driving. You are basically hopping in a car and waiting to be a destination. So it's not driving.
So that's what I mean by like, are you saying driving as a process or outcome?
Yeah, I mean driving, I think infer that you are doing some kind of action and you have some kind of agency in it. So a PhD, where is a PhD? PhD, PhD, when D map, PhD,
interesting, yeah, there was a similar comment discussion about whether scaling is a process versus outcome. And I think driving, when we say driving, I thought it was okay to just see this. So, so
this is our this is our PhD. This is a PhD users and customizable software, the CO adaptive phenomenon. So basically she, and this is why I went to exaptation, because she studied the CO adaptation between users and the way they would tweak their software. And she studied something you might know, it's project Athena. Do you have maybe an account on Athena?
Yeah, no, no, no. But you know what it is. I know the Greek god.
You don't know, you don't know MIT
project Athena, this, this thing, basically, it's the abit undergrad here. Yeah, yeah, I'm not. So the undergrad, when they arrive to MIT, they have an account on a system called Athena, which is basically the you have Kerberos, for example. It's part of it for the password. So basically, you you have Athena terminals all across the campus, and it's one of the first Do you know Linux? Yeah, so X Window. I've heard of it. X server, the one that does the window windowing system in Linux or Unix. So this is a part of Athena. It was how to have to go from command line to a visual kind of, Oh, I see, yeah, yeah. So when it was developed in the 80s, she studied how the developer of the system were customizing their email client, because at the time, the email client was mutable. You could, you could program your own email client. Now, when I go to my email clients, the most I can program is if I do, if I go to settings in French, and I go, for example, to rules, and then maybe I can add a rule. So with this, I can program a little bit. If this email does this, it will go in this and that. But this is very limited. In the first time the email client was a program that's in C and you could have the source code of the email, and you could, you could reprogram your own email client, and people would do that all the time, and then they would share also part of their code through email, because then they could send email. So she studied how these patterns of customization were slowly evolving across MIT, across the campus. So why I say that there is a link between customization and an excitation is that when users of a technology have the possibility to to modify the products they are using. And here we are actually in the office of
user based innovation, yes, of Eric
Von Hippel, which is the master of that, the lead user, you know the fact that the user, they have agency and power, they can actually not just being consumers or passive receptors, but they could be actors, actively transforming the product they are using. They are capable of in sometimes going beyond the functions. Software, the products were initially destinated to like intending to so they can create new functions, new usage. And this is what happened with the email and Athena, they started to create other kind of software based on that and practices that she studied. And she studied how it was a co adaptation. It was like both the system and the users were co evolving. It's a co evolution. And based on this CO evolution, I started to research other kind of CO evolution in technology, and I had lots of different terrains. And this is why I said, Actually what happened in when this case was exaptation, it was similar to what gold and people in biology call adaptation. But My PhD was in starting 2004 and I did, I didn't publish a lot, especially in English, about this after, but my knowledge people using exaptation in business was on maybe, like six, seven years after my PhD, it was not very well known outside of biology to use these things. So this is, I did lots of research, and I, I converge to that. I said, this is very apt, because it's co evolution. We should, we should think about about this. Yeah,
that's kind of like what I was trying to Oh, that's a cube, because
it's an hypercube. Of course, I forgot about it. This is why you were, you were sketching a cube with more than four Okay,
so the comment I want to make is there are very valuable supplies out there like and your thoughts, your thoughts on acceptation and the cautious probable asset program. And what I'm suggesting here is we can assemble this and supply to the entrepreneurship world. And we need a very like plan to do that, because, including me, if I'm wrong, like Charlie's cockpit, I think is a really gold idea in your expectation discovery, with connection with the management, is a really like gold discovery. But I feel it could have been diffused more if we design how we're going to diffuse this idea in entrepreneurship and management. So that's why I'm trying to develop not only the supply, not just push, but also the pull as well, together. And that is related to and doing the research on pivoting is about because, after defining the pivot as the change of optimal action in response to the updated word model of an agent, I define three levels of pivot. Which is
so this is a paper. This is your your paper with everything in it.
No, I have lots of more to say. But which
paper is it? This one? It's the one with Bayesian and, yeah, there is no evolutionary in this.
This is pre three by three model, but this is mostly on Bayesian and entrepreneurship exactly, and mathematical model behind it, because, if you see the contents, it's rational agency. You can take a picture if you want, but entrepreneur decision making and Bayesian pivoting with computer and minimal viable product of computational cognitive support for pivoting decisions. And what you previously mentioned as we need some mathematical model, and this might be included in here.
So, um,
so Scott also like this part, which is
using Overleaf
to create this. Yeah, I think it will be in one of the text, but I believe is
nice. No, that's huge. You prefer what
I don't know. I have many different tools, latex, I am evolving. Sorry. So what this is trying to say is you should choose your level of pivot, meaning, Hoffman has a book on defining different pivot levels, shift, Swift, switch, swerve, rebound and reboot and rebound. And I'm classifying that in three levels of whether you view demand as fixed versus your core idea as fixed. So if you're just given the demand, you don't have ambition to change the demand, and you just want to accept what's out there, and you just want your idea, or an idea to implement that you're, you know, kind of,
yeah, you're looking one of the dimension,
yeah, it's kind of the theory ideas in here, or the basic idea of, let's say you and I want to develop pivot game, yes, and this is like using acceptation to entrepreneurs, and this is the idea, and this is how to implement that, and this is the demand out there. So if the first is, I don't have any intention to choose the demand out there. I would just accept entrepreneurs behavior as there is. And I don't have any intention to choose to change the idea. I am all into expectation to entrepreneurs as currently is, and I just want to change the way we would design the game. So if that's the case, we should just based on this our observe distribution of demand and idea distribution. We should optimize here so we would the discussion we're gonna have when we are searching the space versus if we are optimizing over ideas well, would be very different, because here we're gonna just find what is the best specification of the game should look like, given This and this. But if we now be willing to change our idea a little more, then we can just have an idea of acceptation is actually like related to customization. So for those, we can connect more with test quantity and Bayesian workflow, and we can be able to pivot in a larger which
so if excitation is linked with customization, it means basically that the customer, the customer is able to pivot.
No, I'm not including customer as a pivoting now i The perspective here is just entrepreneur, okay, but entrepreneur can educate customers so that their needs can change. That's basically what I'm doing. The reason I went to the conference was I was part of the Bayesian statistics where we intend to diffuse the idea of Bayesian statistics and management and the other organization organizers are doing that for 12 years and So, long story short, I think education, the reason I'm doing education on Bayesian is to change the needs, increase the needs, so that when I make a tool that has Bayesian idea in it, people have an absorptive capability to use
that. What absorptive capability, absorb absorptive capability you can absorb,
yeah, that's the word I made absorptive, yeah, kind of, if we give system dynamic model to any people, there are difference in how people are willing to accept the people who has more kind of physical and modeling experience can absorb it well. To give you a
piece of something to think, we are working with NASA with RPM, basically
trying to help through design to help diffuse some simulation tool inside NASA better, and why they don't diffuse good, because there are assumptions by the tool makers, the Engineers and the designers of a specific mental model of the users, exactly that when the users receive the tool, actually the mental model assumption were wrong. Yeah. So we are now helping them to think about tools where the mental model will be able to change, and we tried to help them to represent these different kind of mental models. So the tools would reflect the mental model of the users. So then the users could see them, and they could say, oh, actually, this is not at all what I need or want, or it's not what was happening. And we're talking about predictive tools, interesting bias and tools, Monte Carlo, MCMC, what is the name? It's basically, you might be interested by, by by this lab. By the way, it's a Artificial Intelligence Lab called of JPL at NASA. It's the data science that JPL and they are, they are super nice while working with with them. Do
you think I should enlarge my thing right now?
No, no, no. It's just to give you an example about, about about what you're to try to,
sorry, don't get don't get me wrong, like I, I have a bad habit of kind of trying to expand infinitely.
Yeah, it's not it's not good, yeah, so,
so, so I need your kind of, kind of intentional help and trying to limit myself.
Let's try to help you feel the appropriate limits to find and no, of course, with pleasure.
Thank you. Thank you. I want to know everything you're doing, like, in my, like, deepest heart, but I'm trying to retain myself from doing that. You know what I mean? Yeah.
So, so, okay, so back what we're saying. So we're talking about a game where different kind of approach, including Bayesian approaches, or like Charlie's approaches, would be useful to play the game, and you could play the game with different kind of either hats, or when you play, will have to have all these models at your disposal to play the game. So and like this, you can compare, then you can align to see the drawbacks and advantages of all of them, basically. So it's kind of an educational game. Use the bayensian culture into, for example, the entrepreneur community.
It's one of the goal. Yeah, that's one of the goal. And, like, I think toolbox idea comes after, like, depending on which lenses that we are using, and the core model should be how we'll design the agent state and environment state in the agent's head.
And by Agent, Agent, you are the one who's playing the game, yeah, the person the player,
yeah. Like, it should have its word model. Like, let's say, if we are making a pivot game, then we are assuming that people entrepreneurs, even though they don't have enough time, they have some time to play a game, right? That's kind of we have a series of assumptions that we're making. So it
should be, the game I should be playable offline, because like this, when they are in the plane and there is no Wi Fi, oh, I have one hour I can play. Why not? Yeah, to think about how much time they think about it. Let's be serious one second about that. Look
about let's discuss what is the product level and what is, yeah, demand level, and what is the supply level and demand level of pivot key, exactly.
So let's think about the context of this game, you know, Duolingo, the game to learn the new language. Yeah. So let's say, let's it's an hypothesis. Let's say that Bayesian approach is like a new language. You you talk, you're fluent by Asian and you met Master of Bayesian language, like these people, okay? Andrew, so, so in the game, we are able to be more fluent to this Bayesian language, but we need to practice. Need to practice how Bayesian concepts are thought of and in Duolingo, what is nice is that it's a casual game. They have very little chunk of knowledge you can practice to make the game could be done by having this playing rounds that are very sharp. Maybe you can five minutes you're not, you know, like one hour playing. It could be it could be sliced. And then when you have a bit of a moment, you could say, Oh, what about if I put my product or my market in the pivot game? What would it say now, and especially with this and this element of context, and if I put the lens of Charles is fine, what do they tell me? And maybe they are. They are actors of the character in the game. And they will tell you which are fine. I will tell you that this is you fail. And I think what what is resonant for a founder. Every morning, when he wakes up, they think about their company, how much money they are making, how much risk they're having, and they think about their product. And some of them their customer, all the time they will. They look at the world through their company. Also, this is great. I should maybe for my product. I could do that. When they go to have a piece they can say, Oh, that's interesting. The toilet. And so that they are 24 hours around their company centric. So if every morning, when they have a bit of time, they're like, oh, yeah, in this game, if I put the context for my product or market of today, what would it say? A bit like, what, what people they use with AI, given this context and new information I had today that I didn't know yesterday, what can the game give me as new information or a way to handle this new information in a better way than just with my brain? It's my brain assisted with Bayesian conceptualization level. Basically. I don't know if it makes sense, but that's what I know. The way to look at the game, maybe a game. It's a series of micro, micro games, not just one game.
I agree, I agree. So that's what Vikash and Josh was trying to make so incremental, so they have a generative work model and like this can in put as
a competition. Already, I see what's
a competition? No,
no, no, against claims, victory against Alex. This could be beats Josh beating and victory. And this is a bit competition. It's one of the four kind of game we can do. You know, the agonistic, I told you the classification. There are only four games structure. It's one of the four. So this is one of them. It's the competitive, competitive.
Yeah. So they are calling the spays and tug of war. And what they're trying to illustrate here is Bayesian with probabilistic reasoning. Can give you visuals of the belief update based on new information. And that's what like for me. I was making a demand forecasting and inventory management algorithms when I was doing a startup, and I learned in warm morning, Amazon is trying to enter this market, and I needed to understand what is the meaning that entry in new information, yeah, yeah. So we were actually discussing, our teammates were discussing, how should we interpret this new information as like with with this people, and so this the meaning of it. And another example is, can I one manager was telling me our sales, they have been increased. EV, sales has increased for the last, like, one year, but last month it decreased. And the managers are wants to know, like, how we should interpret this as and how we should pivot. Is
it? Is it our fault? Is it from the market? Yes, yes. Is it structural? Yeah, and coronavirus.
So the meaning, I think, is what this tool is trying to get with the its work model based on its work model. Yeah, I think for our future meeting, if we discuss like, let's say we are entrepreneurs developing a pivot game, and what's our causal assumption that we're making and how we pivot that if we record that thoughts, I think that will be a great contribution. Doesn't make
sense. And how my own perspective, coming from exaptation, can bring a value to this game. And there are other approach, and the Bayesian is important also. Charlie is what is bringing for the game. Yeah. If
you
are willing to accept that, like what we are developing, a pivoting game can be in the future, a way to diffuse your acceptation idea, meaning, if you are willing to be more flexible on this axis of idea that we'll develop next year. Very, very flexible. And I think we can have a better optimization algorithms, meaning we are not stuck in this hyper plane. And actually, as we introduce this pivoting game idea to the audience, to the entrepreneurs you meet, like your diplomat and like I think that can also change. So that was the idea that it could
be also, not just for startups, could be used in some context after
including like sabufa, like if we have a prototype and show this to her, and also HPI people. Do
you know zafa? What are the theoretical foundation? They are busy themselves for what they do. Because you talked with her
design. No, no. I have only met with her once, and she seems to be very busy. I've sent him mail several because I
think they are doing some kind of a tool, yeah, for
prediction, prediction. So the so I in
this space. Where are they? For example, in this family, what are the families they would they would be
learning the behavior of the your investors and like they use some data driven optimization of how we should pitch. Okay, so observing your collaborator and how to navigate. So
it's refining the feedback from the investors. Yeah, it's optimization of that. Yeah,
okay, do you generally agree with my classification here? So based on the Tesla so where taboo fast approach would include is behavior like cash flow management, VC roundings and R and D costs.
Yeah, I read this. I read this. This table already. I will reread it. Thanks. Use Cloud or chat, GPT or both.
I usually use quote, you have an account, yeah? $20
$20 Yeah.
So, why is it bayezian Here? Because, and not for example.
So behavior. First of all, this was classification. And the reasoning was,
Oh yeah, okay, that's the reasoning, okay, I It's where it's most appropriate to use. Okay, got it for a toolbox,
remember? Okay, yeah.
This is Claude thoughts, and I will share my thoughts. Yeah, your thoughts might not be that, but I all developed the prompt to do this. So maybe because I attached my paper the rational analysis and classification. So basically, you
try to result in the same way. Okay, got it. Okay, let's go back to this then.
So what is the reasoning for the two evolutionary Yeah,
so that was also what Charlie asked. But I think if there are situations like, regardless of how they classified, if there are generally a lot of uncertain situations and so that your work model is not valuable enough, that's when I would use so,
so that's that's an interesting Can I write on the ball? Yeah, sure. So there might be something to consider related to
on one side, like,
like, what kind of certainty we have of The evolution of something, and because of that,
maybe, maybe it's something more like this. So like, very sorry
to interrupt, but
this was, I think, an easy way to interpret this, like, go, see, plan, do reflect. I think plan reflects more on being you're using your brain. Do is more evolution? You're using your hand and go see, is more behavior. So I made a brain and I brain and then
go, see is that you don't act? Yeah,
usually you're observed. Like it's like, compared to investors versus you, like you less act, but you observe like their investing thesis, right?
That's why, but so,
just as a reminder. Okay, so here, let's say it's here, it's certainty. So here, like, let's say this is very certain and this is very uncertain. So certainty, 100% this is 0% and here, I don't know yet, but basically, if it's very certain, maybe you would use some kind of bias. And if it's not certain, you would use ever, right? I
thought the otherwise, if you have large uncertainty, I would use pains because you need Provost reason, interesting. So evaluation measure, the first median uses posterior and evolution uses fitness for evaluation measure.
So it's a new causal or not, if you have high causality or if you don't have causality. And this is something, I think, a bit a bit so this guy just to to think about your thing. There is his diagram in design thinking that you might find relevant. His name is a collaborator of David Kelly. His name is Bill blank, and is this diagram he did. The thing here that make me think about what you what you're what you're thinking of, is basically saying that you can do your when you interact, when you you create a design, it's, it's four things that you are doing. Is the question How, how something is working. You can act in the world. You can do you can look at the world, or you can feel. So this is maybe more like the the how you you receive, or, yeah, maybe I found a better one of this one, yeah, this one,
yeah, that looks very much like reinforcement learning
diagram. It,
it does. And this as well. Basically, it's, that's from Charlie's, yeah. And this, this looks like,
so how he designed the Asian School of Business program like he made people in iterate between learning and classroom and going to the site and apply their learning. So
this is, this is from the classic cybernetics like Norbert. Norbert winner,
uh, feedback, feedback loop, basically.
So it's classic system control theory, basically, by
the way, if you're interested, I think you your line of thought, and also submarine is more on design, right? Yes, there was this. I think you may want to take pictures so how effectual causal and scientific and design approaches differ. So this session was invited tap po and arnardo camufo, who is scientific, and
actually there is design as well. So this is where. So what was the name of the conference? Again,
animals Academy of Management. Show you the link of the section.
Aum,
this one, yeah, do you use any like messengers at the mail? No?
Oh, okay, sorry,
it's okay.
I mean, I also use the slack of the Media Lab that you are not on it, I think. This slack server, yeah, project, whatever, but
I'll follow up with a
snippet that's in the GitHub that summarizes the next meeting, including all the links. So let's make it through there. So
it was a good slack and people were messaging me. I tried to minimize messaging things like this. I have less thing to answer. Got it. I try to minimize, yeah.
So if I call you in Jason, like, can you get the message? Yeah, okay,
I have lots of message for you.
Get notification from the GitHub, right? That's
right, if I go here, for example,
here, for example, got
it, yeah, I would prefer if you could leave this not through mail, but in GitHub. If you first press like view in GitHub, then you can
why, when I reply all, why isn't there a GitHub email that would fit it directly into the GitHub. That would be a good idea.
Yeah, maybe you know what I mean. Yeah. I do what you mean. It might
be possible, because then I have to go on GitHub, and then I have to sign into comment, yeah.
I think that prevents you from non GitHub user leaving the message, yeah, they want you to log in, yeah. But
they could vet my sister if it's sent from an email. They know, yeah, yeah,
got it. Um, what were I tried? Oh, yeah. So this session was about, so,
what is this game? So Bucha need are on top of things this
was the session, and I think you would be interested in this person. Demo,
demo, P, demo, yeah. Demo, yeah.
He is from ice Iceland. Iceland, I think, yeah, he has Iceland in his sport. And I think you might be interested in like related to our pivoting level. I think he showed this diagram where you have a problem and you have levels of agency. So this regional rational agent can be viewed in different levels. So if you change operations in, if you fix idea and demand, then you'll be in operation level. And I think if you are willing to change your idea in reasoning level, and if you're envisioning, you're a thought leader, and you're willing to change the demand, and you're in this level, and how to switch between this level is important, is what he said.
