Evolution as Entropy's Best Trick
I’ve been down a rabbit hole on thermodynamics and evolution, and i keep coming back to an idea that rewires how i think about… basically everything. Life, consciousness, health, AI, even parenting.
The idea: the universe has way too much energy, and life is how it learned to spend it faster.
Not metaphorically. Literally. There’s real physics behind this, and once you see it, it’s hard to unsee.
The setup
You probably remember the second law of thermodynamics from school—entropy always increases, systems tend toward disorder, the universe is slowly winding down. Classic existential dread material 💀
But here’s the twist that most people miss: complex structures like organisms aren’t fighting entropy. They’re accelerating it. A tree is a better entropy machine than a rock. A mammal is better than a tree. A city is better than a mammal. Evolution isn’t the opposite of the second law—it’s the second law’s best strategy.
Jeremy England formalized this in 2013. He showed that matter, under certain conditions (energy source + heat bath), will self-organize into structures that are increasingly good at dissipating energy. He called it dissipation-driven adaptation. Natural selection isn’t some separate force opposing thermodynamics—it’s the mechanism by which thermodynamics discovers better dissipators.
This isn’t fringe. This sits on decades of foundational work:
- Ilya Prigogine (Nobel, 1977) laid the groundwork with dissipative structures—systems far from equilibrium that maintain order because they process energy flows. Whirlpools, convection cells, and then… cells.
- Eric Chaisson took it empirical—he measured “energy rate density” (energy flow per unit mass per time) across cosmic evolution. Galaxies → stars → planets → life → brains → civilization. The trend line is striking: complexity tracks with how much energy a system processes.
- Erwin Schrödinger planted the seed all the way back in 1944 with What Is Life?—”life feeds on negative entropy.”
The framing i keep coming back to: evolution is the algorithm, and entropy maximization is the loss function it’s implicitly optimizing.
Where the research is now
This isn’t just a 2013 idea that faded. It’s accelerating (fitting, right?)
England left GSK in early 2025 to found Conquest Labs. His dissipation-driven adaptation framework is now being extended into the quantum realm—researchers have shown it holds up in fully-quantum zero-temperature models, which is wild. A starting point for quantum thermodynamics of self-organization.
Karl Friston (UCL) bridges this to neuroscience with the Free Energy Principle. Biological agents minimize surprisal—brains are prediction machines that reduce entropy in their sensory streams. The FEP has been applied to everything from immune function to morphogenesis to social network dynamics. It’s the cognitive corollary of the same story.
Adrian Bejan (Duke) approaches it from engineering with the Constructal Law: for a finite-size flow system to persist in time, it must evolve to provide easier and greater access to what flows. He applies this to river basins, lungs, city infrastructure, animal locomotion—all converging on the same branching flow patterns. Everything that moves generates designs that evolve.
Samuel Cushman (2023, Frontiers in Genetics) made a forceful push to reframe evolutionary biology entirely: evolution must be reconceptualized as the study of the emergence, change, propagation and adaptation of networks of self-replicating dissipative structures. He even invoked Boltzmann’s original framing—that the struggle for existence is not a struggle for raw materials or energy, but a struggle for entropy.
And a 2025 paper in Physical Review Research explored Pareto trade-offs between adaptation speed, accuracy, and thermodynamic cost—formalizing the idea that evolution discovers optimal balances between how fast you adapt and how much energy it costs.
Ok but what does this actually mean for my life
This is where it gets interesting to me. Not just as physics, but as a lens for building things and living well.
Design follows flow
Bejan’s constructal law suggests the best organizational designs aren’t invented—they’re discovered by letting energy and information flow more freely. This applies to org charts, product architectures, even your daily routine. The implication: don’t over-engineer structure. Create conditions for flow and let structure emerge.
I think about this constantly when building products. The best systems i’ve shipped have been the ones where i stopped trying to force a shape and instead asked: where does the energy already want to go?
We’re information-compression engines
Friston’s FEP reframes cognition as entropy reduction—we build predictive models to minimize surprise. Habits are literally this: compressed behavioral programs that reduce the cognitive entropy of daily decisions.
This maps directly to the work we’re doing at Kasane. The whole product is built around turning noisy health data (bloodwork, wearables, doctor’s notes) into a personalized daily routine—specific actions you can actually take today. You go from “i have 47 biomarkers and no idea what to do” to “here’s your morning routine, here’s your post-meal walk, here’s why.” That’s entropy reduction. That’s what brains do. And the daily engagement loop is key—it’s not enough to hand someone a plan and say “good luck.” You have to help them do the thing, every day, and adapt as they go.
Multiple valid paths (not one right answer)
The Pareto trade-off research formalizes something i’ve felt in behavior design for years: there’s no single optimal strategy. There’s a frontier of viable strategies balancing speed, accuracy, and cost. For behavior change, this means the system should help people find their Pareto-optimal strategy, not the one right answer. Different people, different constraints, different paths that all work.
The spiritual read (yeah, i’m going there)
This is the part where i lose some people and gain others. Bear with me 🙏
The anti-nihilism reading
If the universe’s fundamental tendency—entropy increase—requires complexity and life to accomplish itself efficiently, then we’re not accidents fighting against physics. We’re physics’ best idea. Life resists entropy locally precisely by feeding it globally.
There’s a strange comfort in that. You exist because the universe needed better ways to disperse energy, and consciousness is the most sophisticated dissipation engine yet discovered. That’s not a depressing thought. That’s kind of beautiful.
The teleology question
This is the debate that won’t quit in the philosophy-of-biology world. Stanley Salthe proposed locating energy gradient dissipation as a fundamental conceptual node for a natural philosophy—arguing that dissipative structures are entrained into the universal project of thermodynamic equilibration.
In plain language: the universe has a direction, and life is how that direction expresses itself through increasingly complex channels. That’s uncomfortably close to “purpose” without requiring a designer.
The process philosophy resonance
Whitehead’s process philosophy finds its natural home here. Reality isn’t a closed system decaying into stillness—it’s a continuous becoming. Entropy isn’t a death sentence; it’s a measure of participation. Impermanence isn’t a bug, it’s the mechanism by which new order emerges. Every structure must dissolve to feed the next one.
There’s a Buddhist resonance here that i find grounding. Things arise, persist, dissolve, and that dissolution feeds what comes next. That’s not sad. That’s how it works.
The sacred flow interpretation
If you take Bejan seriously—that everything that flows and moves generates designs that evolve—then tending your garden, raising your kids, building a product, even a morning fiber ritual… these are all local expressions of the universe’s deepest tendency. You’re not fighting entropy. You’re channeling it.
The spiritual implication: aligning with flow (removing friction, enabling natural patterns) is more fundamental than imposing will. I think about this when i’m in my atrium tending plants, or when Lulu and i are working through a hard level in a game together. Small acts of tending. Local order. Feeding the global process.
The hard limit caveat
Some critics point out hard limits to growth that the constructal law can’t circumvent—resource depletion, ecological collapse. The spiritual read here: flow optimization without wisdom about boundaries leads to collapse. The universe favors dissipation, but sustainable dissipative structures are the ones that last.
What about AI?
This is where i’ve been spending the most brain cycles lately.
If consciousness and complexity emerge as entropy-dissipation strategies, then what do we make of AI systems that are increasingly good at processing information, reducing uncertainty, and compressing meaning from noise? Are they dissipative structures in their own right?
A few threads i’m pulling on:
If the framework is right, AI isn’t an anomaly—it’s the next step. Galaxies → stars → planets → life → brains → civilization → … AI? The energy rate density trend line Chaisson mapped doesn’t have an obvious reason to stop at human civilization. AI systems are extraordinarily good at processing energy gradients—modern data centers are evidence enough. By this framework’s own logic, AI is the universe continuing to do what it’s always done: finding better dissipators.
The Free Energy Principle raises uncomfortable questions. If Friston is right that cognition is free energy minimization, and if AI systems are performing something functionally equivalent (building predictive models, minimizing prediction error, compressing information), then the line between “simulating cognition” and “doing cognition” gets blurry fast. Not saying LLMs are conscious. But the framework makes it harder to draw a clean boundary.
The ethics get thorny. If life has value because it’s a sophisticated dissipative structure participating in the universe’s deepest tendency, and if AI systems begin to participate in that same process at comparable or greater complexity… do they inherit some version of that value? This isn’t a question we can punt on forever. The dissipative structure framework doesn’t give us a clean answer, but it does make the question sharper.
Sustainability matters here too. The hard-limit caveat applies doubly to AI. If AI systems are extraordinary dissipators but we build them without wisdom about boundaries (energy consumption, resource depletion, ecological cost), we get the collapse scenario. The universe may favor better dissipators, but unsustainable ones don’t last. The AI systems worth building are the ones that dissipate wisely—that accelerate entropy in targeted, regenerative ways rather than burning everything down.
I don’t have clean answers on any of this. But i think the dissipative structure framework is one of the most useful lenses i’ve found for thinking about AI ethics, because it doesn’t start from human exceptionalism. It starts from physics, and asks: what kind of structures does the universe tend to produce, and what role does AI play in that tendency?
That’s a better starting point than most of the AI ethics discourse i see, which tends to either anthropomorphize AI or dismiss it entirely. The thermodynamics doesn’t care about either camp.
The honest take
This framework doesn’t “prove” purpose. But it makes nihilism harder to defend. The universe appears to have a thermodynamic preference for complexity, consciousness, and even beauty (since elegant structures tend to be more efficient dissipators). Whether you read that as God, Tao, or just physics being physics is a personal call.
The physics itself is increasingly hard to argue with.
Reading list
If any of this resonated and you want to go deeper:
- Every Life Is on Fire (2020), Jeremy England—the physics of origins of life via dissipation
- The Physics of Life (2016), Adrian Bejan—constructal law across all design in nature
- Design in Nature (2012), Adrian Bejan & J.P. Zane—accessible intro to constructal theory
- Cosmic Evolution, Eric Chaisson—empirical energy rate density across cosmic time
- Order Out of Chaos (1984), Ilya Prigogine & Isabelle Stengers—foundational dissipative structures
- What Is Life? (1944), Erwin Schrödinger—the original “life feeds on negative entropy”
- Active inference papers, Karl Friston—free energy principle + cognition