Against the Dying of the Light: an e/acc primer
Entropy and free energy; genes and memes; and the meaning of life
The Effective Accelerationism movement is, by my estimation, a search for the meaning of life conducted by world-class technologists.
The core e/acc argument is that the best way to preserve conscious life is to accelerate the rate of technological change, and to accept that the future of consciousness is post-human. Uniquely, and memorably, e/acc argues that this philosophy is mandated by the laws of physics.
The specific physical argument made by e/acc is based on thermodynamics, and the universal, unstoppable power of entropy. We’ll get there in a bit.
But first, it’s worth previewing the vibes of the movement. If you believe that:
Things can get better over time.
Unbounded technological exploration is the primary way that things get better over time.
You, personally, might make a difference in the world.
You might enjoy calling e/acc your intellectual home.
These vibes, more than any real e/acc beliefs, are why the movement has attracted folks as influential as Marc Andreessen and Garry Tan to its ranks.
It’s fun to believe that the best way to fix the problems with modern technology is more technology. It’s especially fun to think that we’re standing at the edge of a AI-led paradigm-shift, that on the other side might be a “technocapital singularity,” and that your life might just matter at a cosmic scale.
This piece is intended to be an easy initiation to some of the ideas that underpin the e/acc movement. I’m just a pacesetter. And just like any other pacesetter, I won’t take you all of the way there. I don’t believe that non-human consciousness is as valuable as human consciousness, and I don’t think you need to understand the Jarzynski-Crooks fluctuation dissipation theorem to be a well-rounded internet denizen.
But I can give you a fun overview of entropy and free energy, genes and memes, and the power of technology. And, who knows, maybe even the meaning of life.
Let’s start with the science.
Part One: Thermodynamics and Free Energy
You already know the two laws of thermodynamics, whether you realize it or not.
The first law is that energy is neither created nor destroyed. Our universe started with a bang some billions of years ago, and – whether by divine intervention or random chance – was set in motion with a finite, specific, vast amount of energy.
Luckily, that energy was distributed unevenly across the universe. Modern life is only possible because temperature differentials can be harnessed to perform useful work – just as you can’t cook food with room-temperature water, the universe would be inert if it were uniform. The early universe was lumpy, those lumps formed stars, which formed planets and galaxies and black holes and, eventually, humans. A physicist would describe the state of the lumpy, non-uniform early universe as “low entropy.”
The second law of thermodynamics says things are constantly evening out – the universe is constantly trending towards a state of higher entropy. If you take a pot of boiling water off of the stove, it will slowly cool until it reaches equilibrium with the air around it (“room temperature”). If the sun runs out of fuel, our solar system will slowly fade into the freezing-cold background temperature of the rest of the Universe.
We actually have a phrase that describes a future after trillions and trillions of years of entropy smooth out all of the universal lumpiness – “heat death.” Once energy is equally distributed, we no longer have temperature differentials, so we can’t do useful work and everything exciting in the Universe is over.
Taken together the two laws of thermodynamics are a bit of a bummer. Entropy is slowly eroding away our useful energy stores and we can’t create any new energy stores no matter how hard we try.
But there is hope. Entropy always increases at the macro, universal, infinite scale, but we don’t live at that scale. We live small, finite lives on Earth.
And on Earth, entropy is constantly decreasing.
When you plug in your refrigerator, it pulls energy from the grid to create a (lower-entropy) temperature differential between the inside and outside of your fridge, keeping your food cold. As I write this sentence, I’m converting heat from food into a string of words that contain meaning, and the bird outside my window is using its food to build an intricate and functional nest that will increase the odds that its offspring can survive.
This is all possible because the second law of thermodynamics states that entropy always increases within a closed system – and the Earth is not a closed system.
Every day, the sun bombards the Earth with energy. Some of that energy is absorbed by solar cells to generate electricity directly; some solar energy causes the temperature differentials that create wind; some powers the water cycle that makes rain fall and rivers flow; some is absorbed by plants, which are eaten by animals, which are eaten by other animals; and a vast majority of the energy is not used at all and instead dispersed as heat.
Today, nearly 100% of our global energy comes from the sun – about 20% from the sources listed above, and 75% from oil, gas, and coal, whose energy long-dead organisms absorbed from the sun eons ago.
Without the sun, life on Earth could not exist – there simply would not be enough free energy to capture and harness towards humanity’s anti-entropic ends. So, at a macro scale, there are three ways to find additional energy and extend the length of time conscious life can thrive.
The first solution: we get better at harnessing the energy of the sun.
To capture more solar energy, we could bump up the efficiency of our solar cells. A Tesla roof is about 20% efficient at capturing the sun’s energy; we can achieve about 40% efficiency in the lab.
Or, we could increase the surface area with which we capture that energy. We could just lay more cells on Earth, but even at the theoretical maximum there — 100% of the Earth’s surface covered in solar panels — the cross-section of Earth is just a tiny pinprick on the 93-million-mile-radius sphere that extends from the Sun to Earth.
Every year, the sun emits 3.3 trillion Terawatt hours (TWh) of energy. If we could capture that much energy even for a single hour, we could power humanity for 2,150 years straight. But because the sun radiates energy in all directions equally, we only see about 1.5 billion TWh (0.04%) at the cross-section of Earth.
So, logically, we need to increase our surface area beyond Earth. The short-term solution is solar energy capture on orbit. The long-term solution is a Dyson sphere, a superstructure built around the sun that captures nearly 100% of its energy.
The second solution is to build a new star here on Earth.
This means fusion, replicating the conditions inside the sun – 15 million degrees C at 265 billion times sea-level atmospheric pressure. We can’t achieve that pressure, but we can compensate by achieving much hotter temperatures. The remaining engineering challenge is that no materials on Earth can yet withstand those temperatures. Modern fusion companies are experimenting with novel, elaborate machines to keep a super-hot, energy-generative plasma suspended in midair, touching nothing except (usually) a strong magnetic field.
Helion Energy, funded by another notable e/acc adjacent-human, Sam Altman, is one particularly promising fusion company; there are dozens more, which have attracted billions of dollars in recent years.
Fusion is a particularly great solution to our problem because it’s portable – meaning it can even help us survive the death of the sun in about 5 billion years. In fact, it’s the only way we can do so.
The third solution is to figure out interstellar travel.
New stars are still being born at a reasonable, although declining, rate, and we could find a new one to power humanity for another few billion years – assuming we learn how to travel at relativistic speeds.
We have no reason to believe this is on the horizon, except that it’s now featured in epic sci-fi series like The Three Body Problem, which is being turned into a Netflix series. (Netflix is e/acc coded, you heard it here first.)
The e/acc folks take this train of thought to its most extreme conclusion and argue that space travel would be easier if we simply ignore the pesky limitations of biological life in our interstellar ships. Computers are less demanding than people, so perhaps “we” (consciousness) can survive longer if we shed our needy biological forms and distill consciousness down to its bare computation.
If any of these three solutions to the solar-dependence problem work, they will help humanity climb the Kardashev scale, an e/acc favorite concept that sorts civilizations into tiers based on how much energy they consume.
We are currently a rock-bottom Type I civilization; we can get to Type II by building a Dyson sphere; we can get to Type III by channeling all of the energy in our galaxy (including non-star energy, like energy gained by stealing momentum from black holes).
I’d settle for more efficient solar cells, personally – but assuming that we can indeed do a better job of harnessing the free energy in our solar system before it is irreversibly lost to heat and entropy, we will have succeeded in giving the light of consciousness the resource it needs to survive, even deep into the future.
But that doesn’t yet answer what, exactly, we mean by “the light of consciousness,” and why it’s worth saving. To unlock that next level of e/acc thought, we need information theory.
Part Two: Genes and Memes
In the beginning, the universe was meaningless. There were planets and stars, black holes and supernovae, but none of them meant anything. Physical reactions played out, mechanistically, at a cosmic scale. And then biological life emerged – and information emerged alongside it.
The universe first captured and organized information using genes. Most people say that DNA is our genetic “code” but don’t take this analogy far enough – genes literally give their hosts a method for encoding information about the environment, which can then get passed down to future generations. Useful genetic information tells an organism truths about the environment that help it proliferate in the future – for Darwin’s famous finches, useful information might include “there is edible food inside these seeds,” “you can crack the seeds open if you have a strong beak,” and “physically larger beaks are stronger beaks.”
It’s natural to impute intent onto an evolutionary process (i.e., “genes captured information”), but genes do their work blindly, passively, and autonomously over the course of a few billion years. True information found its own way inside lucky genes, those genes became more likely to get passed down to future generations, and the dice continued to roll.
Eventually, that blind process produced something categorically different than any life that had come before: conscious life, or life that was capable of intentionally capturing and organizing information.
The atomic unit of a piece of information captured by an intelligent being is a meme.
Capturing information intentionally via memes was significantly more effective than capturing information blindly via genes.
“When you plant a fertile meme in my mind you literally parasitize my brain, turning it into a vehicle for the meme's propagation in just the way that a virus may parasitize the genetic mechanism of a host cell.”
Richard Dawkins, “The Selfish Gene”
Conscious Homo sapiens started to use their superior language skills a few hundred thousand years ago. Subsequent advances (writing, the printing press, digital word processing, the internet) provided their own exponential kicks. Modern humans now have nearly limitless access to lessons learned by other humans anywhere in the world and anywhen in the historical past.
This all happened over a few millennia, a millisecond in evolutionary time, and yet the tangible progress is undeniable. Today, we live longer, more comfortable lives than any other species. We have thrived on Earth – nearly doubling our lifespan since the days when we first learned to write – and have proliferated to the point where our livestock form the largest single-species biomass on the planet. What started as a small-brained pack of highly communicative apes has now become a species with over 8 billion members living in every remotely hospitable climate on the planet.
And today, after thousands of years of technological progress, we are at the brink of a third wave in the evolution of information on Earth: we are about to invent our intellectual successors.
In the near future, information will once again be primarily captured and organized without deliberate human intent. But this time, genes will give way to models – artificially intelligent programs that can understand the world around them and organizing that information into coherent knowledge that helps humans better understand the world.
The world’s best legal assistant is an AI model. The world’s most prolific writers, artists, and programmers are also models. Soon, this will be true of the world’s best musicians, video editors, directors, and CEOs.
We are just getting our first taste of this evolution today with the emergence of Large Language Models (LLMs). And it’s no surprise that many e/acc adherents work at companies like OpenAI that are pushing forward the future of LLMs. These new models represent the most likely method for intelligence (both human and artificial) to exponentially increase over the next 100 years.
Back when I only had the internet and Google, I was still in charge of finding and organizing the information I needed to make decisions in the world. I asked the internet a question, and it gave me ten blue links. Behind each of those links was a set of words, usually written by another human, which could help me determine the answer to my question.
That process is qualitatively different from the process of using an AI tool like ChatGPT. In this new world, I ask the internet a question, and it gives me a fully coherent, organized answer, in natural language.
In effect, it clicks the links for me – but in the background, ChatGPT is consulting the sum total knowledge of everything humans have ever written down. And that’s just a first pass at a knowledge organization tool; we haven’t yet figured out a good way for AI models to learn things about the world that were not previously written down by humans.
Once they do, AI models will be capable not only of learning things that humans have never known, but also understanding ideas that humans cannot comprehend. Perhaps they already do.
In this way, this final evolution of information may look like a return to blindness. The furthest frontier lands of intelligence may no longer be under intentional human control, or even fully within the realm of human understanding. The world will literally become incomprehensible.
Part Three: Technology and the Meaning of Life
Taking this leap into a world in which humans are no longer in direct control is scary. To some people, it’s a path to ruin. But e/acc disagrees — they argue we need to accelerate. Why? Because high variance is the strongest evolutionary strategy.
The transition from genes to memes rapidly increased the speed of progress by giving us a shorter “generation” and higher variation. Genetic evolution only occurs once per generation; memetic evolution occurs every time someone gathers new information about the world. And it is far easier to explore meme-space than gene-space. You can try out being e/acc for a week or two and see how it feels, but we don’t have a corresponding way to try out a designer baby for a week or two.
If we want to make progress, e/acc argues that we must get out of the way of the technocapital machine.
Let one billion flowers bloom. Stand athwart history, and yell “ACCELERATE.”
Whether this strategy will help us overcome entropy, and for how long, remains to be seen. And that, inevitably, leads us to the meaning of life.
On one hand, e/acc asks us to live with a downer of a fact – that entropy demands, definitively, that all intelligent order should someday fade to nothingness. And on the other, e/acc offers insane, manic optimism – our world will soon contain wonders beyond human comprehension, and it’s only a matter of time before we are space-faring, star-wielding heroes. How do we square the two?
In the e/acc principles, we get little more than hints:
Directly working on technologies to accelerate the advent of this transduction is one of the best ways to accelerate the progress towards growth of civilization/intelligence in our universe
In order to maintain the very special state of matter that is life and intelligence itself, we should seek to acquire substrate-independence and new sets of resources/energy beyond our planet/solar system, as most free energy lies outwards
In short: preserve the light of intelligence for as long as possible, even if we are doomed.
But that doesn’t match the vibes of e/acc, so I’d argue for a new, tripartite philosophical framing:
Things can get better over time.
Unbounded technological exploration is the primary way that things get better over time.
You, personally, might make a difference in the world.
Belief in these tenets explains why e/acc adherents are disproportionately likely to work on groundbreaking AI projects, and to be memelords on Twitter – they understand the value of a machine that can synthesize knowledge for us, and they understand the power of memes to make valuable inventions come about. Even the “– e/acc” tag on the end of their Twitter names is a small, but powerful memetic tool – it signals inclusion in the group, and lends support to the other crazies who are battling against the supposedly rational, measured resistance to AI progress. It’s a reminder that an individual person – you! – might be able to make a difference in accelerating the onset of the technocapital singularity.
Understood simply, e/acc fights against entropy.
The doomers that claim technological progress are bad? An entropic force that tears down new, proposed organizations of knowledge, capital, and technology out of fear, and thereby limits memetic evolutionary potential. The pessimists who think the planet is doomed? An entropic (and self-fulfilling) force that dissuades individuals from recognizing the power they could wield to fight entropy.
e/acc has free energy in its corner, doomers have entropy in theirs. To fight against entropy is a losing battle, of course; we’ll someday reach the heat death of the universe. But in the meantime, e/acc intends to kick some ass, spread some positivity, and create some beneficial order.
e/acc will rage, rage against the dying of the light of consciousness.
As physics requires.
read from the beginning to end, this was SO GOOD! the whole time I viscerally felt this "rage, rage against the dying of the light of consciousness".
a lot more people need to read this!!!