Chapter Content
Okay, so, let's talk about nature, right? Like, the ultimate builder. We always think about things being, you know, super sophisticated, but, honestly, nature's where it's at. It's the best example we have of how to actually solve really, really tough problems.
I mean, think about it. Nature’s solutions are just, like, these organic and inorganic collections of stuff that deal with all these naturally difficult things, you know? Like, a river? It handles water distribution, erosion control, even moves nutrients around, creates habitats, and, oh yeah, keeps the floods from being, like, totally insane. Or a mountain? Catches water, boosts biodiversity, regulates the climate, captures carbon... It's kinda mind-blowing, isn't it? And, of course, you have beavers. Wood harvesting, dam building, food storage, territorial defense, and even structuring their families are handled by these little engineers.
It's funny, we look at nature and we get *so* jealous, don’t we? See a bird fly and suddenly *we* want to fly. Watch a dolphin zoom through the water or a duck just chilling on the surface and...boom, we wanna do that too. But, let's be real, our bodies? Not exactly equipped for survival in the wild. We don't have fur to keep us warm, claws to defend ourselves, or, you know, huge teeth to take down prey. But, hey, we're pretty good at working together and turning ideas into tools.
And that's the cool part, right? Observing nature and then, like, recreating it. That’s a big part of what makes us human. We look at a fish fin and immediately understand the connection between surface area and, you know, making it go. And we don't just leave it there. We go and make our *own* fins. Seeing how water flows in a river makes us want to build our own water distribution systems to deal with floods. Looking at a mountain inspires us to build high-standing buildings with cool shapes to save energy, or to protect us from the weather, or even other people. We even try to copy beaver dams with our landscaping and engineering. We try to copy the fur or skin of animals with our clothing and other textiles.
Humans have always looked to nature, you know, for inspiration to make the things we need to survive. But, here's the thing, our stuff? It's not even *close* to as sophisticated as nature’s solutions. Is a car faster than a cheetah? Well, depends on what you mean by "fast". In a straight line on a flat surface? Yeah, sure, the car's faster. But that kind of setup is completely fake. Humans invented it because the car can't really move in *most* environments. A car is useless in almost every natural setting. Can't exactly speed through a jungle. Tires are going to slip in a desert. I mean, sure, you can change the tires and, like, add better parts, but it's still not going to move like nature does.
A cheetah can run around corners, turn super fast, accelerate quickly, and keep its balance in all sorts of situations. Way more capable than a car in pretty much every way. And it's the same with *all* the things we make. They might *look* sophisticated, but they just don’t compare to what nature can do. Our stuff only looks good in artificial environments. A car needs a road to be fast, a plane needs a clear sky to be efficient, a drug needs us to ignore the side effects to call it “targeted.” Nature's stuff is just... better.
We've added a ton of pieces to our solutions, right? Bridges, planes, microchips... But they're still, like, totally predictable. They only do what they're supposed to do based on the instructions they're given and how we designed them. Every single piece has known interactions and clear pathways. Inside a car's engine, even if it’s super advanced, everything is planned, everything is deterministic. Sure, there could be errors that cause things to go wrong, but that comes from an outside source, not from the system itself.
A lot of human progress has been about changing the world so it fits what we can build, instead of, you know, creating actually amazing innovations. When we're honest, we see that we haven't been creating *true* sophistication. Besides cities, markets, and AI, most of our solutions don't have the hallmarks of complexity. That means they don’t solve really, really hard problems. Our creations only seem sophisticated because we've built a world to suit them.
Think about a rocket engine. People call it “complex” because they're comparing it to other things we've built. It has way more parts than an old gunpowder rocket. But now compare that rocket to a squid's propulsion system. They both push through fluids, but the squid has control that rocket engineers can only dream of. Squids can precisely control their speed and how they move. They can do bursts of speed or travel for a long time without wasting energy. Their nervous system helps them sense their environment and change how they move in real time. They can even regenerate and maintain their propulsion system on their own.
The rocket's only "better" at one super specific thing: going to space. The rocket has to use up components and fuel to do that. Sure, the squid isn’t going to the moon, but that wasn't what it evolved to do.
When we look at nature, we don't see all the same intricacies we see in human-made things. That’s because nature solves its problems with complexity. Complexity smooths out all those details. A rocket engine is not complex, not even close. It doesn’t matter how many pieces are added, it'll never be complex if its inputs (fuel and commands) always lead to the same outputs (thrust and stability).
True sophistication is needed to solve hard problems, but it's, like, a wet and slimy sophistication, not one of detailed intricacy. When things look intricate, they’re actually simple, because we can see how they work. Complicated is not complex.
Human-invented things aren’t truly sophisticated because they have clear causes and effects. You can debug them when they go wrong. We can draw diagrams and write equations to explain how they work. That's not sophistication, that’s just blatant simplicity. You cannot debug complexity.
So, how does nature create such incredibly sophisticated things? How does it create solutions with way more pieces than anything humans can make? How does it create the cheetah and the squid?
Well, we've already seen hints of it. Cities and AI both create their structures without design. Complexity in human-made things happens when we step back and let things develop on their own. And that’s exactly what nature’s been doing all along.
True sophistication comes from nature's recipe: natural selection. That's variation, iteration, and selection. Variation is just the differences between individuals in a population, like their traits and behavior. Iteration is the repeated cycles or generations, when those variations are put to the test by the environment. And selection is when certain traits are favored or, you know, not favored by the environment, which changes how often those traits show up over time.
Variation comes from a bunch of places. Genetics, sure, things like mutation, recombination, gene flow, and genetic drift. But things like habitat, climate, food, and other factors can also cause different behaviors and traits in different populations. Even things like mating habits, social structures, ways of communication, and group dynamics can play a role. And environmental factors during development can have lasting effects. There is a ton of variation.
But just having a lot of variation isn't enough. You need *a lot* of iterations to filter out the variations that don't work, the ones that can't solve survival challenges. Iteration happens in nature through generations. No one lives forever. That forces organisms to have offspring to continue their lineage.
Variation and iteration work together to make sure there are a ton of options and a ton of attempts. But there's one more ingredient needed: selection. There needs to be a way to decide what "good" means. And that comes from selection pressures, which is just all the things life has to do to survive. If a new generation has a change that doesn't do well in the environment, it's less likely to survive, and vice versa.
The key thing about nature's recipe is that it works from the *outside*. There's no deliberate piecing together of inner details. The inner workings just emerge naturally from the process of iterative change and convergence.
Some people might say DNA goes against this, right? We often call DNA genetic instructions for growth, development, and reproduction. DNA stores the code for building proteins, which are needed for organic matter. But DNA itself is a *result* of natural selection. It's not the starting point. Its molecular structure developed because simpler molecules were better at replicating and preserving the information needed to sustain life. DNA emerged over billions of years through chemical evolution. So the DNA we see is just one moment in a long process.
Plus, DNA is a human definition, a label we've put on it. It doesn't mean it's not real, just that we choose to focus on its isolated structure. But that structure means nothing without the surrounding physical context that makes its role possible. Complexity does not have root causes.
Natural selection shows us that operating from the outside is how true sophistication is achieved. By being outside any kind of inner knowledge about how things work, natural selection allows the chaos of matter and information to naturally organize itself into what's needed to survive. That's how solutions to hard problems are made.
That's why arguments like intelligent design don't work. Not just because they're unfalsifiable, but because they start with the idea that complexity would be designed. Believing complexity would be designed doesn't align with how complexity actually works. This isn’t about being for or against God, it's about the wrong framework of complexity. Nature doesn't design, and the reason is that design only works under deterministic settings, which is not the case for complexity. If there is a God, He would not have designed the universe in the deterministic sense, He would have put in place a high-level, external process that allows nature to converge automatically, an approach far more beautiful.
Nature has always been our inspiration. We've always looked to it for ideas and tried to mimic its solutions. But the real message here is that nature shows us that to create flexible and dynamic solutions that solve really hard problems, you *need* an external process of variation, iteration, and selection. That means stepping outside the systems you're trying to create. It means accepting that you can't figure out how something works from the inside. The absence of causality in complex systems isn't a matter of how difficult it is to figure it out. It’s a fundamental shift that changes how you solve problems.
And even though we usually talk about natural selection in biology, it's not just about biology. It's a universal process for how complexity is achieved and evolves. No matter the system, if you want to achieve something beyond the simplistic complication we see in human-made systems, you have to use massive levels of variation, iteration, and selection to get to things that are truly sophisticated.
Natural selection is nature's version of trial-and-error. Nature doesn't just deduce its way to its solutions. Deduction alone can’t produce an answer to a hard problem because it can’t foresee the trade-offs that occur between features inside a complex space of possibilities. There's no way to analyze how the countless features in a complex situation interact.
The chaos that starts with the three-body problem makes it impossible to know exactly how a thing functions mathematically and analytically, and that uncertainty increases exponentially with the number of pieces in the system. The only way to get the right arrangement of matter is to operate externally to the inner details.
Nature's recipe isn't about aligning details in a specific way. It's about arriving at arrangements automatically. Arrangements that compute the right output because that's what survived. That means nature, above all, is about computation. It creates solutions that compute answers to the hardest problems, and it does it in a way that's completely different from human-made machines. In fact, thinking about nature in terms of computation is a way more intellectually honest way to understand it compared to traditional science. It's also a good starting point for demystifying what emergence truly is.
So, we always think about nature through the lens of biology, right? Well, what if we looked at it through the lens of computation? All systems in nature compute. There are inputs, outputs, and a process in between. A mountain is a structure that takes wind and rain as inputs and turns them into climate regulation and carbon capture as outputs. A beaver takes predators, food, territory, temperature fluctuations, droughts, and parasites as inputs. All of these inputs must get transformed into something that leads to outputs that enable a beaver’s survival; gnawing, food storage and lodge construction.
And this isn’t just some forced analogy. Computation isn’t just for man-made machines. We usually think of computing in terms of doing calculations, processing data, or running algorithms to get a result. But computing doesn't actually need algorithms or logic gates. Computing is just transforming input information into output information through one or more operations. An operation is just doing something to achieve an outcome in a system. That action doesn't even have to be something that moves. It can be done by something that stands completely still, by being next to something that moves. This relative idea of action means things like mountains and riverbeds also compute. In the midst of all the activity, inorganic and organic objects transform matter, energy, and information into new things. That's how nature's problems are solved.
That changes how we think about computation. Computing isn't about procedures, rules, or steps. It's objectively about transforming information. This better definition lets us talk about computation more precisely and understand the universal properties it brings to nature. The most important property isn't about procedures at all, but about abstractions.
I've said that progress through abstraction is a universal truth. That means any process that solves harder problems over time has to do it by abstracting inner details into higher-level constructs. That's true for inanimate objects, humans, and the things we build. It might seem weird to think of a mountain or a beaver as being made of physical abstractions, but in computational terms, that's exactly what they are. Mountains and beavers don't produce their outputs using deterministic steps. It's not algorithms that turn wind and rain into climate regulation, or territory into lodge construction. Nature's solutions are collections of matter arranged through evolution to turn inputs into outputs, without simple causal pathways.
That makes physical abstractions the main computing constructs of complex things. Not logic gates or algorithms, but physical abstractions. That's because computing in nature has to do what all abstractions do, which is to map many possible inputs down to a few needed outputs. Only physical abstractions can produce what the mountain, the river, the cheetah, and the beaver produce.
Mountains might look simple from a distance, but they're incredibly complex structures with numerous intricate facets and sophisticated features. Tilted, folded, and faulted layers are a result of tectonic forces. Mountains often have different types of rock that represent different geological processes and periods. Mountains aren't static objects at all. They're often located at tectonic plate boundaries, which leads to interactions like subduction, rifting, and continental collision. Mountains also undergo physical, chemical, and biological activity. These things break down rock and reshape the mountain and surrounding landscape, leading to a unique structure that solves hard problems.
When we look at nature, we're looking at computation on a massive scale. Not the kind of computation we see in traditional computing, but the kind of computing that only complex objects can do. Computation that relies on physical abstractions that compress information to solve problems.
Nature's recipe of variation, iteration, and selection isn't how nature computes answers to problems. It's how nature finds the physical solutions that compute answers to problems. That's the difference between an internal and external process. Natural selection is an external process that doesn't care about the internals needed to solve a problem.
An internal process is one that does the actual computation to produce the needed outputs. In most human-made objects, the internal computation is set deliberately. This means that for almost all the things we have created throughout human history, discovery is only used to find the pieces that end up getting strung together via design, to make the computation happen. But for complex things, the discovery process is used right up until the point the object functions, when the object becomes usable. Whereas traditional engineering must use design to create its computing constructs (e.g. the interaction between rifle components to compute the firing of a bullet), complexity sees its computational constructs emerge automatically, through discovery alone. With natural selection, by the time the solution is discovered it is already assembled, with all the necessary guts required to compute in the wild. Discovered, not designed.
We saw the same external process used in deep learning. Deep learning isn't about engineering the specific inner details to compute outputs. It's about programming a scaffold that implements the external process of variation, iteration, and selection. A variety of data, millions of iterations, and selection against optimization criteria make deep learning a narrow reconstruction of natural selection. That's why deep learning can realize inner computations that engineers never put there. Deep learning works because of emergent abstractions that compress information, like all truly complex solutions do. Deep learning is possible because of a building approach that steps outside the system it looks to create, allowing things to converge on their own, until the necessary computational constructs arise automatically.
With technologies like deep learning, humans are starting to see what it means to build things that solve naturally hard problems. But it's not enough to just appreciate different processes that solve different problems. To understand how complexity works, and to demystify what emergence is, we also have to understand where the “hard” in hard problem comes from.