Hello, and welcome to the Physics World Stories podcast. I'm Andre Glesser. And in this episode, we're gonna be exploring the physics of crowds. Specifically, or at least to begin with, looking at the physics of sheep movement.
And whilst this is a slightly wooly topic, and you will find the odd sheep based joke in here, there is also some really intriguing physics not just a value to those who are animal herders but anyone interested in animal or human behavior later in the podcast we'll hear from someone who won for their work on human crowds the ignobel prize the prize given for research that makes you laugh and then think but on the physics world website you'll find an article by Philip
Ball a science writer based in London entitled field work, the physics of sheep from phase transitions to collective motion. And through a window out of which I often gaze, there is a flock of sheep. I must admit at no point when I've looked at these sheep have I thought of physics. So I wondered quite why it had occurred to Philip Paul. So gazing out of that same window at that flock of sheep, I gave him a call.
I've had a long interest in how ideas from physics can be, can be used to understand systems that don't sound at all like physics. And animal motion is one of them, actually, including the motion of people. So actually, years ago, 20 years ago, I wrote a book about, the this topic of how we could use ideas from physics to understand aspects of human society.
And, you know, as I say in the piece, compared to trying to do that, trying to understand sheep might seem like, you know, an easy thing to do because, they have somewhat less, sophistication than, than humans. But it it's actually it's a very interesting problem to understand how sheep herd, but also there's this added twist that sheep are herded by dogs.
You know? And that seems like a completely kind of idiosyncratic thing that is miles away from anything physics could handle, but it turns out that it isn't. And these models that are used, they're ones that have, that sort of have their roots in work on particularly on more obviously sort of collective flocking motions like, flocking of
birds, like the schooling of fish. Things that, you know, when we look at them, there seems we we just kind of intuit that there has to be some deeper principle going on that gives them this kind of coherence, that, you know, doesn't seem to come from from individual animals. And in fact, I'd seen before I even encountered these papers, I'd seen, just, you know, video footage on on the web of sheep
fluids. So it's got the kind of these aerial photos, probably drone photos of sheep, her you know, wandering around in fields, speeded up and kind of going through gates and, you know, going around obstacles and so on. And it's really weird to see it because it really does look like the flow of some kind of, you know, slightly granular fluid.
So there too, if we see it if we look at it on the right sort of time scale and from the right perspective, you know, again, there's this sort of intuitive feeling that there's something there's some grand principle, there's some physics in this system. That's what this work that I was writing about is trying to explore. So I say I'm looking out the window, and there are a flock of sheep. You know, there's walls, there's hedges,
There's occasionally the odd sheepdog. And at this time of year, when the tourists have gone, the sheep are sort of, yeah, just left to their own devices. When there are tourists here walking their dogs, the sheep are scared all over the place. But it's it's I've to be perfectly honest, as a bit of a nerd, I have enjoyed watching the sheep, more at this time of year where they're just left to their own devices, as I say, and they have this kind of well, yeah, they are a flock, aren't they?
Yeah. It is. And, you know, what so so what does that mean? It, you know, it means there's a a load of them, but it means something more than that. It means that they are somehow interacting with an with each other to stay coherent. And that's something that pretty much all, groups of animals do or groups of living organisms. Actually, even cells do that. You can
see it with bacteria. There are some people who develop models like this to try to understand bacteria and how they sort of flow around, and they too can kind of sense each other's presence. And, you know, I I I kind of we we if we think of how we move around in space, we're clearly doing that as well. We like to think, you know, we're all individuals. But if we're moving down a crowded pavement or something, then we are
interacting with others. We're certainly, we're trying to avoid collisions if we're not staring at our phones. We're trying to avoid bumping into, each other. But also, occasionally, we might be moving with a group of friends or with family. And so there's a kind of attraction, a kind of cohesion there. We're trying to to stick together.
So, you know, there are these these principles that aren't totally unlike the way inanimate particles might interact through forces of attraction and repulsion, repulsion being this tendency to avoid collisions. And and this was the kind of idea that, you know, lies behind all of these attempts to try to understand group motion.
Is it is it something that we can describe just in terms of forces of attraction and repulsion between the individuals, who each of which, you know, have their own kind of agenda to some extent, but it may be a fairly simple agenda. I mean, even for us, if we're walking down a pavement, we're generally trying to get from one place to another, you know, probably as as as quickly as possible while navigating obstacles and while, you know, avoiding collisions.
So that's that's a, in principle, that's a fairly simple situation to to try to model and, you know, that's kind of what you're probably seeing sheep do, except that there's, you know, there's also something else that I imagine you're kind of seeing them doing, which is, you know, they're not just trying to get from a to b. On the whole, what they tend to be doing is is wanting to to eat. They spend their most of their lives doing that, you know, chewing on on grass.
So, you know, that's another aspect of the problem that's specific to grazing animals that they're not just trying to move. They will move occasionally, and in particular, they'll move to try to find more food, more grass. But there's that, there there's that sort of complex, you know, interplay between just trying to, you know, be left alone to graze, to eat, and moving around together. And with sheep in particular, as you say, they're nervous of other animals, particularly of dogs, but also
of humans. So there's also that aspect. They're kind of looking out for what they might, you know, imagine are predators. So those are the kind of components of the behavior that need to go into trying to model this situation. As we've been talking, the sheep have gone from lying down behind a sort of hillock on the hillside, if you see what I mean. It's it's it's a a mound on the hillside, which is protecting them from the wind. That's why they're all there as far as I can see.
Then the sun's come out, and they've all spread out across the field. Physics doesn't explain that, does it? I mean, apart from the sun coming out. Well, it it it I could see how it could in that, you know, if you were wanting to, to to include in a model like this, their wish to be in sunshine, then the you know, there's an attraction there. There's an attraction to a particular kind of, you know, you can model it as a as a force, as a field. I mean, there's literally a field as well.
So you can see how you can, you know, build aspects like that into it. The other, interesting thing about sheep and about, you know, larger animals like that, which actually doesn't apply so much to us. You know, sheep are elongated. So from above, we're kind of, you know, circular blobs, and you can model us as circular particles. Sheep are kind of more like ellipsoidal particles. And so there's an orientational
aspect to that. It's a bit like the contrast between, you know, simple atoms or or, you know, globular molecules and liquid crystals, which also have that tendency to kind of align, to respond to one another's presence by sort of aligning their their axis. And so, you know, that's a question we might ask and and the researchers have asked about sheep.
Is there a a tendency to align? And if there is, then that looks a little bit like the way magnetic spins align in magnetic materials, and you can build that force into it as well. A force that, will make one sheep have a kind of tendency to align its body, with the direction of what its neighbors are doing. Why? Well, I I I guess I mean, certainly, if if you're if we're thinking about the way sheep move, you know, they seem to the the flocks, seem to occasionally to sort of wander as
groups in the same direction. That's something that you just observe happening. So sheep, you know, have this tendency to follow one another. And if that's the case, they're clearly gonna have to be aligned in order to do that. So they're not, you know, literally, they're they're trying to align in the walking direction. So while they're moving, you know, they'll be
trying to align. But also, you know, if they were in a dense flock and they were being, for example, if they were being shepherded by a dog, then again, they're they're one they're they're not gonna be wanting to bump into each other. So, you know, there's probably there too going to be a tendency to kind of line themselves up.
So, you know, that it's really, just either a matter of, doing that in order to, coordinate their motions or doing it so that they can pack more efficiently in a small space. I do there's a a behavior that I observe here. I don't know if you've seen an answer to this, but, very regularly in the summer months, there's, you know, a tourist dog who should be on a lead isn't on a lead, but we'll we'll sketch over that particular issue. But one of the sheep gets isolated from the rest of the flock.
Then the dog goes away and that sheep the isolated sheep then calls out and tries to find its way back to the to the flock. It well, it seems that they do. I mean, a lot of animals do, particularly if there's a predator around because it's the kind of safety in numbers thing, you know, that that actually a sheep on its own is gonna feel much more exposed than a sheep in a flock and particularly a
sheep in the center of the flock. So, you know, there's clearly a, there's a sort of attraction if you like there for the sheep to a sort of cohesive force that will keep them together for safety. And, you know, more than that, that what you actually see in, sheep flocks and in other sometimes other groups of animals that are exposed to predators is they they all wanna be in the middle, which kind of makes sense, you know, that that's, that's the safest place.
And so there's this kind of constant churn, you know, going on. And in fact, it's, some, biologists have, you know, talked about this in terms of this kind of selfish flock theory, where, you know, everyone wants to be in the middle. And so what what kind of dynamics, result from that? So, yeah, that that that, you know, if if if, an individual or small groups get isolated, then there is this sort of attraction to bring them back into the flock
for the the the purposes of safety. In the selfish flock, my understanding is that at least it's the way it's talked about, I think, in the march of the penguins. They they take it in turns, and that's less selfish and more, altruistic, more part of the group kind of looking after each other. Is that just a kind of a nice way of looking at it? Well, that you know, that's a good question because it's quite possible, and I don't know if this is the
case for penguins. I think it's more probably the case for for sheep, certainly. It's quite possible that if you have a model where there's the the the the individuals are moving around, they have this tendency to try to get into the center. That actually spontaneously out of that will come this kind of churn, this kind of turnover of who's on the edge. So it's not as though the individuals on the edge are kind of at any time consciously
thinking, okay. I've had my turn now. You know, I've done my my my part to be on the edge now. Let me back in the middle. I don't think, it seem it seems unlikely that anything like that is going on. It seems more probable that something like this kind of behavior would emerge spontaneously simply from those movement and from those kind of psychological rules.
And the other thing that seems to be spontaneous from what I can see is the who the leader is at any moment with the sheep in terms of the the leader of the movement, who takes this the flock in a particular direction or parts of the flock in a particular direction?
Yeah. Sure. I mean, that that is something, that that is observed, and it's something that was looked at, in particular in one of the models that I discussed in the piece, work done by Fernando Peruana and colleagues, at one of the universities in Paris, where they they were watching. So they started off empirically. They were just watching small groups, sometimes really tiny groups of of sheep, you
know, move around in a field grazing. And they found that there was this kind of tendency to occasionally develop to adopt to kind of, you know, moving mode where they all walk in some single file to a different part of the field. But what they observed was that there so there's a leader there, by Natesti if there's a line, but they they found that the sheep seem to take it in turns. So there's no, single leader who decides that in the flock each time. It seems to happen more or less at
random. So pretty much each of the sheep gets a turn, at being the leader. And, again, you know, that sounds like that's all very kind of egalitarian and, you know, it's you could imagine the sheep saying, okay. Now it's my turn to be leader. But actually, you know, it seems like it doesn't happen or at least we
don't need to assume that. If we simply assume that whoever, becomes the leader, that that is, you know, something that just happens randomly, that there is a particular chance at any moment that any individual will become the leader of one of those moving groups, that that's enough to account for for what we see.
So, yeah, there there absolutely is. And the you you see the the advantage there's an advantage in in having that kind of, if you like, that kind of cheap algorithm, you know, inherent in in their behaviors, relative to their always being the same leader because, it means that there's more pooling of the information that all of the individuals have. So, you know, it may be that one of them knows or has has realized that one part of the field seems to be more attractive or less, you know,
chewed on than another. And the ideal is that the flock has access to the information of all the individuals. You know, this is something that we might also see in in in flocking birds or in the other sort of flocking animals where it might be that 1 or 2 individuals spot a predator where the rest of the the group hasn't. And so you want to make use of that. You want to pool that information that everyone has.
So this idea with sheep, this idea of constantly changing the the leader in these sort of episodes of walking, It's it's an it's an efficient way of working, in the sense that it's efficient in information sharing. I I can't help wondering if it's not that, efficient if you're trying to run a political party or a government to keep changing your computer.
Well, you know, I mean, even there, that's although that's true, you know, there is a discussion about whether there are more sort of egalitarian, you know, whether it would be more sensible, for example, instead of having a House of Lords to have, you know, people randomly, you know, chosen from the population to lead a body like that and that
there's a constant turnover there. And maybe that would be more efficient in terms of being able to make better use of the information that everyone has rather than that a few privileged individuals have for life. But wouldn't the concern there be that you could get someone that was sort of, you know, a a conspiracy theorist running an important part of the world. I mean,
absolutely. You know, questions like that, they're they're they're complex ones, but I suppose the the the principle there is, you know, you could equally get them and you could equally get that in the house. And when you do get that in the house of lords now, and they're there for life.
So, you know, as long as you have that constant sort of turnover, the the the idea would be that, you know, on average, you're going to, get a better sort of representation and a fairer representation of all the information and all the views in the population. There's that kind of angle to it, but I wonder, is this actually useful for the farmer who's herding the sheep out there? Well, I I think I mean, first of all, I guess the question is, can a model like this even reproduce what
we see in terms of herding? And, you know, that's the thing I found quite remarkable that it can. That without any assumptions and here I'm talking about a model, that's being developed by, peep researchers at Harvard, and others to try to look at shepherding behavior. And, you know, here, as as I said initially, it just seems like, how on earth do you make a physics based model of a sheepdog actually trying to herd, you know, a
flock of sheep. But it seems that you can, that it's an optimization problem is how they present it. So you're you know, the the the challenge is to try to get the flock from a to b as quickly as possible without losing any sheep on the way. So what what is the optimum strategy for a dog to, to undertake to encourage that motion? Assuming that there's this repulsion between the dog and the sheep so that that's what's going to kind of move the the flock along.
And there's there's this attraction between the sheep themselves and that's what's going to get to keep the flock coherent. So given those constraints, you know, what's the most efficient way of doing it? And they come up with three possibilities that do the job. One is that the dog just sort of runs from side to side of the flock, keeping it, you know, in order but gradually kind of getting it to move, in a single direction.
Another is that the dog runs in circles around the flock, but they're kind of corkscrewing circles so that it's gradually getting somewhere. Both of those strategies are ones that are seen, literally in the field. And then there's a third strategy, which isn't seen as far as the researchers are aware, which is that the dog actually it almost kind of burrows into the flock. But, because the flock is coherent, it doesn't
just scatter. It sort of stays together, but, you know, it it sort of draws back from the dog within it. And so by the by the dog moving, the flock, it kind of moves with it as if the dog is like a driver in a car, and so they call this strategy driving. So it seems to be one way that the the sheepdog could work, but but doesn't. And it may be that the the flocks just aren't really sort of cohesive enough to make that a viable strategy. We we, you know, we don't know.
But, you know, it does seem to capture what actual sheepdog do. And so, you know, that's a kind of it seems like that's a kind of validation of the model itself. And so if you've got something like that that works, then you can start to think, okay. What are there ways that, you know, are there better ways of of of doing this or are there ways that we could improve the efficiency with
which the the dogs work? Or in particular, are there ways that we might automate this process using, for example, drones or using, you know, robotic, you know, entities to to do the herding instead. And I, you know, I I loved it that, what some of the researchers involved in this work said, well, if you if you speak to farmers about that possibility of using drones or whatever, they just laugh, because, you know, the technology just isn't ready.
But in particular, it seems that sheepdog you know, sheepdog are really smart dogs. That's why they hear that they're they're so good at doing this job. So they have a kind of an instinct it seems to be able to respond on the fly to what the sheep are doing. If you've got one that wanders off, you know, the sheepdog can respond to it in ways that it's probably gonna be very hard to program into robots or drones.
But, you know, if you're ever going to do that, then models like this are precisely the kind of thing you'd need to figure out what are the basic rules of behavior that a an automated shepherd or sheepdog, would have to follow in order to do the job efficiently and reliably. I can't help but wonder why on earth you'd want to replace sheepdogs, though. I mean, what would be the reason? Yeah. Well, they require a lot of training, and, you know, they if if and they have their limitations.
They can only go at a certain you know, they can't go too fast. And, you know, it it could be that particularly if you're working with bigger flocks that, you know, sheepdogs would a single sheepdog would struggle to, to to contain, you know, might it be then that you'd have to use something more systematic, and with great a greater range of capabilities in order to do the job. So, you know, I suppose you could imagine situations where where that's, where where that's the case.
You know, drones never tire, drones never need feeding. But, I mean, you know, I I suspect that behind your question is the same feeling for me that actually it's amazing and lovely that sheepdog are able to do this. And, you know, what a shame it would be if they were replaced by something automatic.
I know. I think the only possible reason I can think for it is if you had a sort of robotic dog that had some form of artificial intelligence, and it worked with the sheep so much that eventually you could ask it whether androids dream of electric sheep. Brilliant. Of course, you would need to get
that in there. I guess, you know, I I mean, I I think the other, answer to that might be, well, you know, it may be that you don't wanna do this with sheep, but it may be that you wanna do it with other systems, including maybe, you know, sort of nanoscopic inorganic systems where you've got a load of what are are called active active particles that can move under their own steam, and you want to get them somewhere and you want some other, you know, active particle or
nanorobot or something to, you know, be able to move them or other animal systems or possibly even cells. You know, if you want to marshal cells in a particular direction, can you do that with some kind of artificial entity that is programmed in certain way? So, you know, these these, questions of how to shepherd a load of, you know, moving possibly quite idiosyncratic particles, there's a broader applicability to it than just she. Can we apply
this to to people? Can we is there some way we can help with crowding at, I don't know, concerts or football matches or something by looking at this? It's it's already
being done. Models like this to try to understand crowd behavior have been used to try to plan things like events like Notting Hill Carnival where you can, you know, literally map out the real route that the carnival takes and, you know, get some sense of what the crowd motions might be like, where you might have pressure points and, you know, difficulties that might need extra policing or extra guidance or so on.
It's also been applied models like this have been applied to the, the Muslim pilgrimage of the the Hajj, which has had problems, you know, some years because so many people come to it. It's this massive crowd event that has, you know, the real danger of there being sort of buildups of pressure in the crowd. And there have been, disasters really that have happened in the past
because of this. And so this sort of modeling has been applied also to to events like that to see if you can create better safety measures. And just sort of crowd panic in general, you know, is is a really difficult problem, that, you know, models like this can can can inform at least.
So, you know, it's not necessarily that you'd, you'd be thinking of a a sort of shepherding problem, but that you'd be anticipating problems that might arise in these sort of mass crowd movements that are possibly quite predictable ones just as they are for road traffic, and then think about the kind of safety measures that might alleviate that. You mentioned your book that you wrote a a few years ago, but haven't you got a new one that's sort of looking in this area? Oh, yeah. How life
works. That's looking more at the kind of molecular details of, of of that question. But it does, you know, it it it does then sort of, open out into looking at higher levels of the system. So it's actually yeah. I mean, in the sense that the that book is looking at questions of emergence in biological systems, you know, that's what we're talking about. It's absolutely what we're talking about with these collective motions of
animals. I mean, you know, the murmurations of of birds are one of the classic examples of of emergence but we absolutely see in biological systems in in all sorts of ways how you get these higher levels of organization that are somehow somehow seem to be autonomous from the really lower level details of what individual molecules are doing and what genes are
switched on. So that broader question of how you get emergent sort of coherence and robustness in the face of microscopic details that might be quite sort of random and idiosyncratic. That's absolutely the same kind of question, that we we see in, you know, in these cases. And in fact, it's actually you could say the same kind of problem that, you know, in a sense, that emergence in biology continues beyond the level of the organism to the ecosystem and for us to the society.
And there are similar kind of principles involved in the sense of how do you get a high level system that is reliable and robust and isn't going to break down as soon as something microscopic goes awry. A walk among the sheep. They they don't worry about me walking amongst them. They just stay where they are. But it the movements that they have, as you say, you know, sort of that bit of field looks tastier or whatever it is, It it it doesn't seem to make sense. It seems to be just sort of
I'm going to move now. It doesn't seem to be any more to it than that. Right? It just seems like, yeah. I'm moving moving over there and then the others follow because they, oh, maybe they found something. It doesn't seem to be kind of, you know, in any way, decisive based on evidence about just sort of going somewhere. The information bit of it, that's really lovely. Right? That that they're sharing that information, but maybe the information is dodgy. It isn't better.
Mhmm. Well, I mean, okay. So maybe the first thing to say about that is that, and this is something that I heard in in writing this article that, you know, farmers, will say, actually, sheep, they're more intelligent than you think they are. They're and they're more individual than you think they are. So for example, it seems that sometimes, sheep flocks respond to what the farmer has told the sheepdog to do. So the farmer can, you know, has whistles or whatever to instruct the sheepdog to
do particular things. The sheep get to to recognize those calls and start behaving, you know, as the dog would want them to behave before the dog has really done anything, you know. So then they are picking up on these things. But also it seems like there's quite a lot of variability in the way they behave. Some sheep just look at the sheepdog and think whatever and ignore it. Whereas others think, oh my god, and run like mad the moment they see
it. So that, you know, that's, that variability you need to build into, the into into these models. But it also undermines the the, you know, the the kind of meme like stereotype really is, you know, sheep are just they just all do the same thing. Well, they really don't. But the question of, you know, why they're doing this in the first place, it's it's it's all about foraging strategies, and each animal has, you know, has
some. And the this has been, a very sort of rich area of research of how do animals decide what strategy to follow, what's what's efficient. Some of them seem to, undergo what are called levy flights where, you know, the idea is you go to one location and you forage around a little bit around there with random walk, and then you make a big jump to somewhere else and then you forage a little bit around there. That's something that you could describe using a particular kind of statistics.
You know, other animal foragers. I mean, if we think the classic example really is bees. You know, that's that's extraordinary, the amount of cognition that's involved there that they will you know, they'll go out, they'll, you know, they'll find some, some sort of source of food, come back to the hive, and tell the others where it is but with this incredible waggle dance and the others will understand and off they'll go. And sometimes they won't. Sometimes
they'll think, you know what? I'm it doesn't I you know, I've been there and it's not I mean, I'm you know, this isn't what the bees are thinking, but it's how they seem to behave. They don't all just sort of respond in this programmed way. So even in the case of a bee with this kind of, you know, brain the size of a grain of salt, let alone a sheep, there's an amazing amount of cognition, of subtle cognition going on there.
So, you know, I I I think although, I mean in a way that's what these models are kind of showing us that, you know, we can be described using these particle based models in certain situations and yet we like to think that we have this extraordinary amount of cognition that, you know, we're making use of in other contexts.
And so we should probably not underestimate the amount of cognition that other animals are capable of, and the the way that the, the the way that that is harvested collectively. You know, that's the key thing that the these creatures aren't making decisions just by themselves. They're actually that that these strategies have this sort of collective ability to pool information and to make effective use of the group.
So it's it's a much richer environment than you might think understandably by going up to a sheep and thinking, there's not a lot going on in there. Actually, you know, there may or there may not be, but what comes out of it is really quite rich. With no offense to any sheep or farmers listening, I wanted to know more about the human interaction in crowds. Someone who won the Ignat Nobel Prize for the research into crowd movement is Alessandro Corveto.
I am an assistant professor at the Endoben University of Technology in the department of applied physics. I have here my own group that is called the AI for complex flows and traffic. And, what we do here is, well, essentially 2 things. So first, use tools of physics to explore, let's say, new physical systems, specifically traffic and, most importantly,
crowd traffic. And the second thing is we use AI tools to try to crack problems in in complex fluids and complex flows that were that are open for long. And, and therefore AI is giving new opportunities in this direction. We won the 2021, Iqnoebel Prize for a research that we started somewhere around 2013 and, that is, that is a very big part of my work, today as well. So we try to understand, whether we can establish some basic physical understanding of of the way pedestrian crowds moves. Okay?
And and this means try to establish a physical model that reproduce, that explain, what we see as we look at crowds. And let's say crowds are very, let's say, as you can as you can imagine, a very random, a very stochastic system. Right? So imagine that you are yourself in a in a train station. What you do today is different from what you'll do tomorrow, and what another person does today is different from what they will do tomorrow and what you will do today.
So it's an extremely stochastic system. And the the key idea of of our work was to explore this system, collecting as much data as possible, in this case, in a train station. Now imagine tracking, all the people passing in this train station for, like, 1 year. Now we we speak about, 100,000 people per day, roughly speaking. So you you multiply this by 360 days. You are immediately in the the millions scale. Well, many millions of scale.
And then, the endeavor was or the endeavor still is to find universal pattern or universal physical features that emerge as we look at this, dynamical system at this, this large scale. And let's say you need to start simple. Right? So this is a this is a extremely complex system. So we start simple. And, so first first step is to look at how individual particles, let's say, individual pedestrian move, what is their statistical
fingerprint. So how can you describe this with the with the stochastic differential equation? And, specifically, the ignoble was centered around the, let's say, the volume 2 that is, when you move from a single particle to 2 particles interacting. So we did, something, let's say, in some sense inspired by by by impact parameter studies in in scattering physics. That is imagine that now you have 2 particles that that are,
walking or are moving in opposite direction. And then, I mean, obviously, you don't collide typically with with the particle coming in in in the opposite direction. And so there is, there is a whole avoidance mechanism that is that is triggered. And we characterize this avoidance mechanism looking, again in the in the hundreds of thousands of trajectories. And, and we wrote the differential equation that
reproduce the same statistical fingerprint. So what happens So what are the the the the common way in which this process, happens? The fact that every now and then there are indeed collisions and how, let's say, the interaction decays in space and and so on. Okay. Okay. So, like, for example, one one time, I was walking along Tottenham Court Road in London. A taxi pulled up. I wasn't concentrating. It was before I had a mobile phone, so I wasn't looking at that. I don't
know. I was probably looking at a a Lego shop or something. And, the, the door opened of the of the taxi without me noticing, and somebody stepped out, and I bumped into, that person. And that person was Billy Piper, who at the time was in doctor who, which was terribly exciting for me. I imagine it wasn't terribly exciting for her. So that kind of collision has had a different reaction with the 2 different particles.
So one of those particles had a, you know, reaction of boredom and annoyance, and the other one had a reaction of excitement and can't wait to tell its other particle nerd friends about the interaction. Let's say this is, I think, connected with with the other Nobel Prize that has been that has been won the same year by the group in Japan of of Claudio Feliciani and the others. And, yeah, they they started, let's say, the the case in which which people are not paying attention because they
are distracted by by mobile phones. And they are examined essentially how this interaction mechanism gets penalized as as people, don't pay attention. What does it tell us about ourselves, crowds, or physics in more widely? Bumping or perhaps not bumping is, is the basic interaction mechanism that there is in this type of system. Let's say that that, you know, as physicists or as applied mathematician, what mathematicians, what we'd like to do is
start from something complex. Right? We have many agents, many people, and so on and so forth, and we try to reduce it using, the language of of mathematical physics. Right? And so immediately, you you would reduce a system of interacting pedestrians into a system of particles that, in a way or another, have, interactions that are in some sense similar to gravity, but probably anti gravity. Right? So you repel each other instead of, attracting,
each other. And, and, the interesting thing, is that considering a a particle system with with this very basic, repulsion mechanism, is enough to recover some of the qualitative features of this system. So you might be aware of of of the following. This is a very, very much used case to explain some of this emergent behavior that you have in this in this crowd system. Right? So,
okay. Okay. Maybe let let me add something that, one of the the connection point between between crowd dynamics and and physics is in the direction of complex system. Right? So complex system is about, how, let's say, complex pattern or complex structure emerge as basic entities interact with simple rules. So the fact that you can get macroscopically nontrivial patterns out of basic
rules between many, many agents. And in crowds, the very basic, emergent behavior that is considered is the following. So imagine that you have a crowd of people, let's say, with with red shirts that are moving along 1 one one street from one side to another and another crowd of people with blue shirts that are moving on the same street but in the opposite direction. Okay? So now you have the 2 big crowds, and they're facing each other and walking
in in opposite direction. And then the very basic, emergent behavior that this system has is that particles, or or people that have that are going in the same direction, which means that they they come with a a shirt of the same color, would align in stripes. So now imagine yourself looking up. Now these 2 crowds walked, and now they are crossing each other.
But then you wouldn't observe that, the system is completely disordered, so you wouldn't observe completely random distribution of red, of red and blue shirts. You would rather observe the fact that the system has ordered in in in stripes of shirts of the same color, meaning of stripes of people going in the same same
direction. And the the fascinating thing is that it's extremely well, it's surprisingly simple to regenerate or to to, explain, to reproduce this emerging behavior by just having particle moving in opposite direction that know only a single rule beside the the the need of moving that is the need of avoiding. So the need of not entering into very string stronger contact. Right? So so the mechanism of not bumping into each other is already enough to explain quite interesting emergent,
emergent behavior. This is only one. Right? There are there are others. And, Yeah. Yeah. It took a very long road here. Brilliant. So when you are you're doing all of this research into into crowds, what what's your goal? There are multiple goals. So, let's say, as a physicist, there is the challenge of of explaining something that is extremely complex, extremely stochastic. So there is the the the underlying question, can we put down
physical rules that that regulate this system? And here, one needs to go deep enough because, obviously, you cannot expect to have a mathematical model that reproduces what one person does because this is obviously, I mean, meaningless. So this is completely senseless because it's just unpredictable. But then, in in our case, going deep enough means having enough data of this system, in such a way that that statistical fingerprint emerged. Like, the fact that the average behavior
of the system is this. The fluctuation the typical fluctuation is that. Events that happen 1 in 1,000 are these and these and these. Events that happen 1 in 10000 are these and these and these. And I think at this point, it becomes not so surprising the fact that about this behavior, one can write, can write physical, equations.
And, and, and this aligns with, again, topics such as, physics of active matter that is, the physics that study all sort of matter that in a way or another is capable of of turning internal energy into motion spontaneously. Right? So we go from, bacteria. We go from, we consider, I don't know, flocks of birds, and so in general animal behavior and so on. It aligns with the physics of complexity that is, how, as I said, how nontrivial pattern emerged from simple inter from simple interaction.
And this this case also, especially when we go in a high density regime, it can align with with, some condensed matter structure. So but in general, there are there are very many connections with different fields of physics. And then, of course, there is also the the the societal slash engineering component that is, now we get to have a model that that is capable of reproducing the the stochastic dynamics of the system. How about we use it to in design phase perhaps of of a facility?
Or we use this to do real time control, which are things that actually are happening. So let's say you observe your system at a very large scale, and then you you have a model that tries to predict what's most likely going to happen in the next 5, 10 minutes, and then you use this to to nudge. Right? So to to exert some action in the system in such a way that that it behaves as it, I mean, to increase safety, for instance, or or
or comfort. And this, this is this might sound science fiction, but this is active research on one hand, and already there are many attempts in this direction already on the other. So Yeah. Do you do you have any examples of where it's been? Well, I can I can connect, with with my research? For instance, we're investigating, the usage of of illumination to nudge, people into, let's say, actually, into distributing in space to maximize some some property.
A typical application would be, for instance, in museums. Right? So you would like floor usage, to be uniform, and not, that, you know, your crowd is fully concentrated in one room and then all the rest of the museum is, empty. We've been working with, with with sound with positional sound as well. Let's say you, only in certain location, you can hear some sound.
And, only for instance, if you have a complete track this is something that we we published recently that, let's say, only if you are walking a specific trajectory, you could hear a full, let's say, melody. Right? And, and if you steer out from this, let's say, target trajectory, you stop hearing the melody. And so there we show that actually we could have people following the the, let's say, the target trajectory due to the fact that there was, an expectation by people of keep on hearing the
melody. You know? So it's like every step you hear a different tone. Right? And then people would be, I mean, spontaneously, you would like to hear what you expect to be the next tone. Right? So imagine that there is a there is a simple melody, and and this enable us to to push people to to follow, certain trajectories. But but there is more. I mean, there are
things that are very unexpected. And, I mean, let's say at times, you are maybe in a train station and and and you are annoyed by the fact that your mobile phone is is is not connecting to the Internet. And this is not because there was no Internet there, but because the signal is locally jammed to make sure to make sure that people don't clog in a in a given, location. Let's say we don't want people to stand here because maybe the the the flow would
would would clog. Right? Imagine, there is a staircase perhaps, and then people would would climb the staircase and just wait there. But obviously and let's say that this this staircase leads to to the platform of a train station. Yeah? And, obviously, you don't want people just to to climb the stairs and and wait there. You would like them to distribute on the platform. And, so how do you ensure that? Then you would locally jam the mobile phone signal in front of the staircase.
And now and now, you know, the person would would climb the stairs and then they would just stand there and get out their mobile phone and then they figure, oh, it's not working. Right? And then I move around to to get it to work. But, yeah, it was not it wasn't working not because there was no signal. It was on purpose. You know, you know and then it was not the efficiency of the mobile phone company. It was rather by design that you were not,
That's amazing. Is that something that's going to happen? No. This is something that is happening already. Really? Tell you this. Yes. Genuinely. People are using people are blocking that. So, basically, at halftime in the football, right, I I sit there and try and check the scores from elsewhere, and it it you just can't because everybody's doing exactly the same thing. Happily, in my own, football club that I support, they provide Wi
Fi, which does the job quite nicely. And and they put the scores on the screen. But there it does happen that it's blocked because there's a number of people in it as well. Right? Just so people are listening and not thinking every time it's blocked, it's some mobile phone company trying to move them on. Well, no. No. No. I mean, like, yeah. Of course of course, there you have a you have a network capacity issue. I I
think these these are different things. Here, I mean, there is a possibility of doing very local jamming. Right? The the the within a within a very limited area, mobile phone is is reception is not is not good. But, again, this is this is already in in place. And, that's extraordinary. That's extraordinary. Well, can you tell me just going back to the light bit, the illumination bit in the museum? Right. How does that would we are you sort of eliminating particular bits to draw people in?
Are you dimming bits? Yeah. We we we ran a couple of of big experiments, around here. Is, is the question, can we use light to to to provide information that is, to convey, let's say I I don't want to use the word subtle, but to convey, let's say, more and more implicit information. And it works, or at least the experiment that we did, were based on on on dimming. And, we showed indeed the following that, imagine you have you have this big crowd.
We did this, in in this very experiment that I'm referring to was that was run during a big festival that we have here in Andover. So it's it's, city walk throughout the entire city, let's say. So, again, tens of thousands of people. And now imagine these these people all entering into a corridor, then, we wanted them to go, let's say, to choose either the left or the right exit. And, in our experiment, we compared our effective it is to show people, for instance, a normal
signage, so big arrow that says, okay. Go left rather than go right. Or rather to change the illumination level between the two ends or the two exits. Yeah. And then we show that that, within some, let's say, crowd density limit, you can have more or less the same effectiveness in, in light and signage have more or less the same effectiveness. And then it happens that as cloud density, grows, as as you can expect, there is no indication that is really followed.
It doesn't matter whether it is based on light or signage. It's it's like, I mean, people at certain point when there are these many in a confined space, they just need to get out. So it doesn't matter whether there is an indication that points there them somewhere. But but, let's say, within operational, density level that you can find in museums and so on and so forth.
So in our experiment, we could show that that you can convey the same type of, signal or you can convey a signal that that has the same effect on people, either by using variable light intensity or or, signage. Could you tell me a bit more as well about the the the audio version? Do do you have the melody? In the paper, you could find the 4 the 4 tunes. It was it was, really piano chords. Yeah. Oh, okay. I'm literally sitting at a piano right now. So Okay. So we could see. Yeah. I I I yeah.
I I think yeah. If you want, I can I can look for it even right now? But it's, let's say in this paper, then then you would see that we divided the space into areas, and then each area, each small area, something like maybe half a meter times half a meter, came with different, piano tune. Right? With a different, and then this was played one after the other should you be, walking along the right path. Yes. Otherwise, you you wouldn't hear anything. And,
okay. I'm not a music person. Right? So now I would say something that I hope makes sense to you because it doesn't make much sense to me. The sequence was c4c5 and thene4e5 and then g4g5andc5c6. So should you be following the right path, you would be hearing this. As when you're walking around in crowds, are you sort of in a kind of Sherlock mind palace with equations appearing in your peripheral vision as you're walking, or are you just kind of wondering how can we make this
better? Walking around is is mostly like, about how to improve the models. So how to include, let's say, more factors in the models that, that that we are using or that we are making. And I think, yeah. I mean so that there are very there are a bunch of very interesting physical directions that are let me make an example that is more telling than others in in in my view. That is, for instance, when when a crowd walk, not necessarily the interaction is cooperative.
Right? So imagine that that you would like to to get, onto onto a train. Right? And you would like to to be inside the train first. And this eventually doesn't mean, for instance, that the people that need to get into the train with you, altogether will manage to get in in the most efficient way. Right? Because typically, the the behavior of of each particle here is is competitive instead of constructive. Right? So and then the question arises, so how to up deal with it, how
to model this? So what is the underlying physics? How can we represent it? I think this is, for instance, a very interesting, a very interesting, direction that is, how to to consider this particle system in which, which, there is an interplay between between competition and and and cooperation, as the the context changes. And so for instance, this is one of
the things that I'm asking myself lately. So what are efficient way to deal with this and then then the randomness in the crowds and and and so on. And about about the interest in general, I think I got into crowds because I wanted to do I wanted to explore a part of physics that had two characteristics. And one that was, close to people, like, that was, that I could communicate easily about, that was immediately impactful in society. That was new. Right?
Where new means that, that research, let's say, started not so long ago. I mean, for crowds, I mean, obviously, I mean, crowds, you know, have been I mean, people have been trying to optimize, crowd flow from the times of the Colosseum in Rome. Right? But nevertheless, let's say physicists started to look very seriously at crowds only since the beginning of nineties, and the first experiments were since the beginning of 2000. So I would say that this is relatively new.
Wanted to do something something new and it's also something that being new enables to I mean, needs to couple a lot of things. So how to do experiment is still an open question. How to model is an open question. How to control is an open question. So it has a very, very broad range of open questions. And this gave me the the opportunity to to to bridge many things from the experimental part to the modeling part. And this is, yeah, also one of the reasons I enjoy this.
I'm not sure whether I project the equations in my mind, but it goes in that direction for sure. I've never spoken to anyone who's won the Ig Nobel Prize before. How does that come about, and how does it feel? Well, it works, in in the in the following way that that you receive a mail from some colleague that you might or might not know that tells. Well, well, typically, I think you you you don't know the colleague and then, you receive this email this morning that says, hey. I read your work.
It's very interesting. I'd like to ask you some question. And then and then, it goes more or less like, the famous scene of the matrix, you know, the the blue and red pill thing. So, okay, I'd like to tell you that that, you got, I mean, you have you won this, Nobel Prize and then, but you can talk with your colleagues and decide whether or not to accept the price. And then, you you, yeah, you contact again this person that that wrote you in the 1st place and you go for the red
pill. And then it becomes yeah, afterwards, it becomes obviously extremely interesting because it's a great opportunity to to discuss with with very, very wide audience from scientists to to popular science channel about about your work. Okay. But so is that why you decided to accept the prize because of that opportunity for communicating about your work? Well, this is definitely well, because it's a prize. Right? So we're not. And and, but also, I mean, it's I think, it's an unprecedented
opportunity. I mean, I don't think, yeah, I don't think, anyone, could let it slip, honestly. Yeah. No. It's a it's a it's a lovely thing. I mean, my understanding of it, right, is is that it's interesting research that also makes you laugh. Right? Yes. Yeah. I mean, the the I think the line by by Mark Abrahams is, research that makes you laugh and then think.
And, I think the beauty of this, global prize is that, every year there are there are discoveries that, I mean, at first sight, they they might look, they might not look about science. Right? Because, you know, maybe seen by the general public, science needs to be extremely complex, very far from what we are doing in our daily life, but, often it's not the case. And, and, and, therefore, this is something that that people would find surprising. Therefore, in many cases,
it can make you laugh. There are, but then definitely, it's it's a channel for having people thinking about surprising, let's say, surprising emergence of science where they wouldn't expect. I think this is a growing field. I mean, the cryodynamics is a growing field in physics, engineering in general. It's something that is definitely challenging scientifically, but also impactful in society as, in in places that are very densely populated, the Netherlands, UK, and so on and so
forth. It often happens that you the the pressure on public infrastructure like the train station grows, by the year, so there are more people moving. But often, it's becoming impossible to extend train stations to to, cope with increased load. And so now the only way is is making the environment smarter, And this is, one of the reason we are doing this. I'd like to thank Alessandro and Philip for talking to me for this episode of the Physics World Stories podcast.
And, of course, we'll post links to their work on the Physics World website, physics world.com. Next month, we'll be looking at Pele's tears, which have more to do with volcanoes than football. Until then, thank you very much for listening.