41 | Steven Strogatz on Synchronization, Networks, and the Emergence of Complex Behavior
What this covers
Sean Carroll interviews Steve Strogatz, a mathematician who has spent decades studying how order arises spontaneously from simple local rules. The conversation ranges across firefly synchronization, small-world networks, and related phenomena in biology and physics, with Strogatz and Carroll exploring both the mechanics of emergence and its broader philosophical implications. A secondary thread examines how mathematics—particularly calculus and nonlinear dynamics—serves as a tool for capturing the essence of complex systems, and touches on gaps in science education that leave young researchers unprepared to think about such problems.
The heart of the discussion sits on two concrete discoveries. First, fireflies achieve perfect synchrony without any conductor or external signal; each individual responds only to nearby flashes, yet the whole population converges. Second, real networks—the brain, social connections, disease transmission pathways—possess a paradoxical structure: highly clustered (your friends know each other) yet surprisingly small in path length (any two people are close). Strogatz and Duncan Watts showed that a tiny number of shortcuts connecting distant clusters can collapse the network geometry toward randomness. The conversation also covers why this matters: small worlds enable both coordination and catastrophe, since anything spreading—information or disease—propagates far faster than on a regular lattice. Strogatz elaborates on Mathematical Impressionism, the deliberate use of simplified models that capture essence over accuracy, contrasts genuine measure zero events with practical impossibility, and addresses why order emerging from disorder does not contradict the Second Law—entropy is actually part of the mechanism. The discussion includes applications to brain function, jet lag, sleep science, and brief treatment of why real degree distributions remain heavy-tailed even if strict power law distributions do not universally hold.
Strogatz argues that complex, ordered behavior—from firefly synchronization to brain function to small-world networks—emerges spontaneously from simple local interaction rules without any central designer, and that mathematics (especially calculus and nonlinear dynamics) is the impressionistic tool for capturing the essence of such emergence.
- Firefly synchronization emerges from each firefly individually adjusting its flash timing in response to others, provably converging without a leader
- Small-world networks combine high clustering with short path length via a few 'shortcut' links, a generic self-organizing property of real networks
- Order emerging from disorder does not violate the Second Law; it is a self-organization phenomenon, not something opposing entropy
Small-world networks arise from very few long-range shortcuts that dramatically compress path length despite high local clustering, making them nearly universal in real systems.
- The naive calculation of degrees of separation (if I know 100 people and each knows 100 others, the world collapses to a handful of steps) fails because of clustering: the people my friends know heavily overlap with the people I know, since we live near each other or share a profession, so the simple multiplication massively overcounts reach.
“The problem is that of the 100 people I know, when they know 100 people, a lot of those are the same people because we live near each other or because we're in the same profession”
- Worlds do not have to be small: a checkerboard or ring world of 10 billion people who each know only their nearest neighbors would have on the order of 100,000 degrees of separation, demonstrating that the small-world property is a non-trivial feature requiring explanation rather than a given.
“The point being that worlds don't have to be small. The checkerboard world is not small and yet our world is small.”
- Twenty years on, the small-world property has held up as nearly a universal law for real networks, while the scale-free power-law claim has not held up reliably; what does survive is that real degree distributions are heavy-tailed—containing far more highly-connected hubs than a bell curve would predict—even if they don't strictly follow a power law.
“The small-world law is pretty much a universal law that almost every real world network is gonna be a small world by our criteria... Scale-free has not held up as well”
- Real-world networks are paradoxically both highly clustered (most of your friends know each other, unlike a random network) and 'small' in path length (almost as small as a random network), and the central question Watts and Strogatz solved was how a network could simultaneously possess both properties.
“it feels like our lives are very clustered. Most of our friends do know each other... But yet the world seems small, almost as small as if it were random. And so that wasn't obvious. Could you have both?”
- Very few long-range 'shortcut' connections (like a chess friend in Holland or a high-school friend now in classical music) are enough to make a network's path length collapse to nearly that of a random network, and because so few are needed and they don't operate in our consciousness, most real networks end up being small worlds unless shortcuts are scrupulously avoided.
“this shortcut mechanism, we showed that very, very few shortcuts were enough to make the world incredibly small, and that seemed to be a very generic phenomenon”
- Because the full connectome of the human brain is unmapped, Watts and Strogatz tested their prediction on the only fully mapped nervous system, the worm C. elegans with about 300 neurons, and showed it satisfied the small-world criteria: much more clustered than a random network but with path length comparable to a random network.
“by looking at the nervous system of this worm, we show that it actually satisfied our criteria for a small world. That it was much more clustered than a random world, but it also had path length about comparable to a random world.”
Synchronization emerges spontaneously from local interactions without central control, solving binding and coordination problems across fireflies, neural circuits, and other coupled oscillators.
- The brain's 'binding problem'—how separately processed features like color and shape are unified into the perception of a single coherent object—is solved by synchronous oscillation: the neurons detecting different features fire at the same frequency, and most brain researchers now believe synchronization is how binding is accomplished.
“synchronous oscillation is the brain's way of, the biologists call it the binding problem. How do you bind all the different features of an object into a coherent single object”
- Firefly flashing can be modeled as a relaxation oscillator (analogous to an RC circuit) in which something builds toward a threshold and then suddenly discharges as a flash; when one firefly flashes it nudges all others closer to their firing threshold, and this mutual coupling is what produces synchronization.
“it's a bit analogous to something we all learn in the first year of physics where there's a charge between two, on capacitor plates that's building up. It's like in the jargon an RC circuit or a relaxation oscillator.”
- Firefly synchronization is an emergent, self-organizing phenomenon that requires no master firefly, no conductor, and no external trigger like lightning; it arises purely from each firefly reacting to and emitting flashes according to local interaction rules.
“the prevailing view today is that it's done, this spontaneous synchronization is an emergent phenomenon that as you say, it self-organizes. You don't need an external signal like lightning in the jungle.”
- Synchronization in the brain is not always desirable: epilepsy is an example of pathological synchrony, where the rhythmic convulsions of a seizure result from millions or billions of neurons discharging in perfect step when they are not supposed to.
“The rhythmic twitching and convulsions is that many millions or billions of neurons are discharging in perfect step when they're not supposed to.”
- For the infinite-range model in which every firefly interacts with every other, Strogatz and Mirollo actually proved a theorem that the population always synchronizes with probability one—non-synchronization remains technically possible but occurs with measure zero, like a pencil thrown in the air landing balanced on its point.
“We proved that for this model, they would always synchronize in the sense that... it had probability zero. It's not the same... So what's a good analogy for this?”
Simplified mathematical models deliberately capture essence rather than photorealistic accuracy, like impressionist painting, enabling understanding when full complexity proves intractable.
- Modeling firefly synchronization is mathematically hard because the system mixes continuous charging toward a threshold with sudden discontinuous flashes and discharges, whereas the standard apparatus of calculus assumes everything changes smoothly and continuously rather than jumping.
“when you have jumps in any kind of system that you're trying to describe with math, unfortunately, the apparatus that we use in math assumes everything is smooth and continuous and doesn't jump around discontinuously.”
- Rather than building photo-realistic, fully accurate models, good applied math (like impressionist painting) deliberately uses simpler models that capture the essence of a real phenomenon; Strogatz's firefly model was not an accurate model of real fireflies yet captured the essence of what they were doing.
“I like to get inspired by something in nature or in the world, and then do something that's much simpler that's the essence of what's out there. Sort of like the way the impressionist painters didn't try to capture the details.”
- Strogatz confesses that, unlike a physicist who ultimately cares about getting the science right, he cares more about math than nature—an 'evil part' of him is okay with a model that is beautiful but a little bit wrong, which is precisely why he is not a physicist.
“if I'm really being honest about it, I care more about math than nature.”
- Biologists could not demonstrate that firefly synchronization arises from interaction because they lacked the computational tools and mathematical training; solving the problem required a physicist or mathematician who thinks differently, illustrating how progress on emergence problems depends on cross-disciplinary methods.
“no biologist could figure out how to demonstrate that because they didn't have the computers that could do... It's not the way that they think. They have different training. So you really need a physicist or a mathematician to try to work on this.”
Calculus, developed over two millennia, is humanity's greatest achievement by systematizing infinity to solve hard problems and enabling transformative technologies like wireless.
- Calculus, broadly construed, is the systematic use of infinity to solve hard problems: you chop a hard problem into infinitely many infinitesimally small pieces (which are easier to solve), then reassemble them through integration—an idea developed over 2000 years from Archimedes to Newton and Leibniz that literally changed the world by enabling things like Maxwell's prediction of wireless.
“roughly speaking, the systematic use of infinity to solve hard problems... you can take a hard problem and chop it up into infinitely small... pieces, and then, those turn out to be easier problems to solve”
- Calculus is one of humanity's greatest achievements—rankable alongside evolution and quantum theory—yet over a million US students take it each year without understanding why; the rich human story and world-changing significance of calculus should be accessible to curious educated people, not just professional scientists.
“the great human story of this fantastic idea that I would rank right up there with evolution and quantum theory, this is a fantastic thing that I want people to understand”
- Mathematics comes in two flavors: pure math that looks inward at itself and is concerned with proving theorems as the be-all and end-all, and applied math that looks outward at the world (broadly construed—oceanography, biology, the universe) as math in the service of, or inspired by, something else.
“there's math that looks inward at itself, which is what I think of as pure math, and there's math that looks outward at the world”
Small-world networks propagate information and disease much faster than lattices, and synchronization can become pathological, illustrating dual benefits and dangers of network topology.
- Small-world networks enable very fast propagation of information through shortcuts, which makes them excellent for anything needing to coordinate as a unit despite being large, but equally dangerous because anything spreading—including disease like HIV—propagates much faster on a small world than on a lattice or ring.
“Anything that can spread for good or for bad, will spread much faster on a small world than it would on, let's say, on a lattice or a ring.”
Jet lag and circadian desynchronization, measure-zero events, and scale-free network predictions illustrate conceptual distinctions in probability and mathematical phenomena.
- Strogatz is skeptical of melatonin pills because experiments show picograms of melatonin are biologically active, yet pills contain perhaps a trillion times more than the brain needs, making him wary of taking such an ungodly amount—though he concedes he may be wrong since he hears of no problems from users.
“in experiments, picograms of melatonin are enough to be biologically active... I think it's like a trillion times more than your brain wants.”
- An event having probability zero (measure zero) is not the same as being impossible; outcomes like a thrown pencil landing balanced on its point, or a presidential popular vote being exactly tied, remain possible but are something you will never see in practice.
“zero probability doesn't mean that there aren't possibilities of it happening, it's just that you'd never see them in practice.”
- Watts and Strogatz deliberately ignored the possibility of making a world small via super-hubs (universally known 'connectors' like Lois Weisberg) because it felt like cheating and seemed energetically costly and unnatural, but this was a missed opportunity—Barabási and Albert's scale-free networks showed a year later that real systems like airports do use hub mechanisms.
“It turns out that there are a lot of networks that use hubs, the most obvious being airports... So we missed the boat.”
- One biological theory for why male fireflies synchronize is that flashing in unison produces a very bright collective signal that escapes the darkness of the jungle, acting as a beacon that lets females locate the males from far away—though from a Darwinian perspective it remains puzzling why an individual benefits from looking just like his neighbor.
“all the fireflies of a certain species would get in sync because it would send their message the farthest possible distance. 'Cause you would make a very bright signal that could escape the darkness of the jungle.”
- Jet lag is not merely about sleep disruption; it is the desynchronization of multiple internal circadian rhythms (body temperature, alertness, hunger, bathroom timing) from each other and from local time—even if sleep adjusts to local time, the other internal rhythms remain on home time, which is the actual 'lag' in jet lag.
“What happens during jet lag is that even if your sleep gets on to the local time, your internal rhythms are still back at home time. That's the lag in jet lag.”
- Purcell's electromagnetism textbook shows the electrostatic force between two charges must lie along the line between them not as a mere property of the force but as a consequence of the symmetry of space: in an otherwise empty universe with no preferred direction, the only available direction is the line connecting the charges.
“he said, that follows from the symmetry of space. But that's not just a property of this force, that's a property about space”
Sleep involves accumulating chemical substances and remains biologically mysterious despite leading theories and a century-old experimental evidence for sleep-inducing factors.
- Despite extensive sleep research, no one has ever really figured out why we need to sleep; leading theories include the buildup of measurable sleep-inducing substances (peptides) during wakefulness and ecological reasons related to being predator or prey, but the fundamental purpose remains unresolved.
“nobody has ever really, believe it or not, figured out why we need to sleep.”
- A roughly century-old experiment demonstrated that sleep is associated with a measurable chemical: blood taken from a sleep-deprived sheep and injected into another sheep causes the recipient to fall asleep right away, supporting the existence of accumulating sleep-inducing substances.
“take a little bit of the blood of that animal, that is sleep-deprived, and inject it into another sheep and that other sheep falls asleep right away.”