 So now it's COVID-19, this new disease that's marching its way around the planet. Nobody really knows how bad it's going to get. It's already not very good. And it might get much worse. I mean, so the question to comes is, you know, is it going to turn out that modern society is just another name for that time between the plagues? Well, I hope not. I think probably not. But, you know, it's really never clear. And, you know, I understand things. I don't feel like I understand something unless I can actually push it around and see what it's doing and see this encourages it, this discourages it. So I make models. I make models of living systems in computers, artificial life, living systems to explore stuff, as well as to do engineering. So I thought I'd see if I could just throw together a model that captured some of the elements of what people have been talking about, in particular, this idea of flattening the curve. This has been going around in the last couple of days, what our health experts mean by flattening the curve. And it's the idea that, you know, if it's all, if people just think it's all completely hopeless, we're all going to get infected anyway. So why bother? Might as well just eat, drink, and be merry. That, you know, the goal is maybe that's not such a good thing to do. And that's what the model tries to get at. There's a particular graph showing that, you know, if you don't actually try to slow things down, you get this big surge. Whereas if you try to, you know, keep people wash your hands and, you know, don't go coughing in each other's faces and so forth, you can slow down the infection rate so that you stay within the capacity of the system to take care of it. So this is what I want to understand. This is what I want to explore. So I threw together a model just in the last day. And I thought I would share it with you for whatever it might be worth for folks. So, you know, here's how it works. I've got this language called splat. It's a spatial programming language. So the way that we're going to implement our model, we're going to have little atoms, little individual things that represent uninfected people, infected people, people in the hospital, represent hospitals themselves, and so forth. And we're going to lay them out in a two dimensional space. And then we're going to write little rules saying, you know, an uninfected guy next to an infected guy has a chance of making the guy infected and so on. And that's what this is. So this is splat we're looking at. We're not really going to look at it much. But the important point is that I define three of these parameters that you can switch when you do experiments. One that's going to affect how much you stay away from, if people stay away from each other. One that's going to affect whether you're going to sort of jump all over the world. Well, you know, you can't really jump all over the world, but jump around a lot or stay at home. And one about whether the hospitals are willing to transfer patients if there are spare beds elsewhere. Now it's really crazy the way my mom does it. So again, none of this is real. None of this is actually COVID-19. Just trying to make an incredibly simplified model to see the dynamics to see how the surge happens and what unexpected things we might be able to see just to get a feel for it ourselves. It's an intuition pump. It's not making predictions. All right. So here's a rule. This is the way splat rules look. There's a pattern on one side, which represents a little 2D view onto some place in the 2D grid. And then there's a pattern on the other side. And the idea is if the left hand side matches, then the right hand side can be done to rewrite it to change the world in some way. So this particular rule using various stuff says, you know, if there's empty spots far away from me, you know, at the edges of the diamond is my horizon. And there's empty spots, then pick one at random and go there. You know, this is the, you know, get in the fast car, get hit the highway, you know, the American dream and so forth. Now we've put it under a constraint, though, that says, whoops, well, it's okay. We have the, you can see it up at the top here, that if the we stay local, then that rule is not going to apply. And that's the trick. So we can turn this thing on and off to see whether that rule and it's going to jump long distances or not. And here's the next one, the reduce contact rule. This is a tiny little rule. So P at sign underscore N at sign is always me, the guy that's doing the thing. And we say given P is a person. Okay, so if there's a person next to me, and then there's me and underscore means empty. So if there's an empty spot behind me, and then behind the empty spot, there's something that's not a person, then step back, just move away a little bit. And these rules, by default, they can, they're set up so that it doesn't actually necessarily mean it's only in one direction. The rule might apply vertically or might apply rotated and so forth. So this reduce, if reduce contact is true, then we're just going to try to back away from other people and so forth. And there's, you know, there's a whole bunch of other rules. And I'm not going to go through them all. But so here's the infectious rule. So if I'm a person, I'm any kind of person. And so I'm the ad sign. And if you is an uninfected guy, and they're right next to me, and this rule happens to go off, now it might not because other rules might happen instead. But if this rule, we get to it, and we have the odds of base infection out of 100 is true. So it's a throw a dice and maybe you're infected, maybe you're not. Now the base infection percent for my model that I settled on is the correct one, of course, 42 42%. If we get to this rule, and there's an uninfected guy, we see next to us, we might not we see it, then we become infected. I'm sorry, I got that backwards, look at that. If we are an infected person, we're the typhoid here guy. If we're infected, there's an uninfected guy next to us. We have a 42% chance of infecting him. But there's another rule. Do I have it down here? Yeah. That if we don't have the saying, you know, reduce social contacts, we also have the thing where people are basically like coughing in each other's faces and doing all sorts of horrible things. That's modeled by this rule, that if I'm an infected guy, and there's anybody on north, south, east or west of me, then pick one of them and infect them. Okay. That's what you know, we're saying happens in normal normal business as usual, that we would like to go away. So that's it. So how does it actually work? There's a whole bunch of other rules, but let's just take a look. All right. So here's our guy. We have a we have this 2d grid. And we have room to put things down. So let's get some uninfected guys to start things up. Did I get him? No, I didn't. I missed them. There we go. Okay. So we'll get some uninfected folks. And they are just bopping around and doing the thing that they enjoy doing. And that's great. But then, you know, someone has to come along. And so here's an infected guy. And we're going to plop him down and see what happens. So actually, why don't we zoom in? So these, you know, UN are uninfected guys. The IN are infected guys. So we already have a whole bunch of infections. This thing is spreading pretty rapidly, because that's what we're saying happens in the default case, if we don't try a little harder. This yellow one here is CR. He's a critical one. So he's actually close to dying. He infects other people. If he can get to a hospital and stay next to a hospital long enough to get cured, he can recover. But again, lots of there's another critical guy. This blue guy RE is somebody who recovered. And a lot of these infected folks will just recover spontaneously. Only a fraction of them will move on from infect to critical. And from critical to dead, if they don't get enough help or they're unlucky. And so that's just going on. It's the infection is spreading pretty rapidly. And we're starting to see these white guys. These are the dead guys. Dead guys in this simulation, they head south for some reason, and they end up turning themselves into a bar chart showing where on the east-west axis, they died more or less. They spread out a little bit. And very soon, we have lots of blue guys. Those are the recovered ones. You can still see a few critical ones. And those are gradually dying out. And this is it. We're pretty well done. And now we're done. There it is. Okay. So we've got a whole bunch of recovered people. There's no uninfected people left. They all got hit in this particular case. And 20% 800 guys are dead 20% of the entire sites in the world, not even just the ones that were started up full of it. Okay. So let's, you know, so that's pretty bad. But suppose we start again, and let's put some uh, uninfected guys down again. But now we've got some hospitals. If I can just grab it here somehow. There we go. And we'll use the airbrush. All right. So now we blow some hospitals in there. And we let this go. Now the way a hospital works is you actually have to be right next to the hospital to the west of the hospital. It's like it's got a ventilator hole that you just got to go up to and get helped out. Oh, yeah. Hey, everything's fine. We don't need these hospitals. Well, perhaps we should have an infection. Let's put them up here just for change. All right. It's infection going like wildfire. But now if we look at some of these guys, let's see here. Right. So this guy is a patient. He's next to a hospital. This guy also a patient next to a hospital. This guy is critical. He'd like to be next to a hospital, but there's no hospital spots. This is taken. This is taken. This is taken. This is taken, and so forth. You can only be the one side of the hospital in the way this model works. And so there, and you know, eventually, maybe we'll see some of these guys will actually recover and go off about their business and we'll free up the resources, and then it'll be fine. But other than that, there's still an awful lot of dying going on in the model. And that's, you know, what we were expecting. The model was designed to be one of a particularly infectious sort of thing. All right, that's it. Now, how is this going to actually work if we did it in a slightly bigger world and let it run longer and so forth? Okay. So, all right. So I have some runs that I did ahead of time and I've got them on fast motion so we can take a look at it. So this is a much bigger world. And this is, you know, no restrictions at all. Just jump around, cough in everybody's face, do everything that you want to do. Hospitals don't transfer patients to nearby hospitals if there's empty spots and so forth. And very quickly, the story is, boom, there it is. So by about 1000 EPs, about one KEPs, those are steps in simulation time. Once again, all the infections are gone. We have, you know, 2% of the sites in the entire world are dead bodies. But there's no more infected ones. There's 19% recovered and still 2 uninfected ones that managed to avoid this whole debacle. The comparison is, let's suppose we turn all these things on now. Don't go jumping around. Step back if you're next to a person and let hospitals move stuff around and see what it looks like there. And at the same speed, all right, here we go. Now, it's going much slower and there's much less dying going on. There still is dying, particularly in the middle there, where there's that dark area, which is a connection. It's kind of like mountainous terrain. It's sort of difficult to move around in there. And I didn't know that was going to happen. Well, but there it is. So there's the demo. And here's the bottom line. The original case where we just business as usual, do whatever you want to do. Cases surged up. The infections jumped up. The patients in the hospital, is that what these are? The blue ones are patients in the hospital. But they're all these critical cases. If you start out as critical and you end up next to a hospital and you turn into a patient, that's how we can tell the difference. So all those critical ones were guys that wanted to be in a hospital, but there wasn't room for them. And it was all over very rapidly. By contrast, you know, if we turn on the various, this is if we're allowed to make the transfer change from hospitals, just change the infrastructure stuff, people get to do whatever they want to do, doesn't help. Well, actually, in this case, it was actually a little bit worse. But the way the world works, it's, you know, the role of the dice exactly where the hospitals land the same approximate percentage of hospitals, people's and so forth, but it changes each time. In this one, it was actually worse than nothing at all. But look at that reduce the contacts, you know, step back, and also move patients around. We've cut the deaths down from almost 2000 deaths in this situation in this simulation to well under 1500. And finally, when we turn all of the rules on, stay local, reduce your contacts, let the hospital infrastructure move stuff around ways that they might not normally do, this is flattening the curve. And we can superimpose them. Right. So this is the surge is the no treatment, the no change in behavior case. This is the aggressive change in behavior case. It's hundreds of almost a 50% reduction in mortality for this little world. So that's it. You know, once again, as far as epidemiology is concerned, it's all about are not the average reproduction rate or zero of a disease when it comes in. And are not makes it seem like it's some big scientific thing. But the important thing is that it includes our behavior. If you think you have nothing to do with it, and then you're completely independent of whether the disease succeeds or not, you are not.