A civilization does not have to conquer the universe to outlive its first home. It may only have to build one extraordinarily stubborn machine—and learn to experience time on its own terms.
00 / The proposition: salvation has a chassis
Imagine that the observable universe is not our civilization’s original address. It is a high-fidelity historical simulation running inside a post-biological ark: a heavily armored iron sphere drifting through deep space. We are not the first humans inventing computers. We are descendants of people who already used computers to escape the limits of flesh.
Call this The Iron Hull Paradigm. It joins transhumanism, digital physics and eschatology—the question of what remains when a world reaches its end. Its central provocation is simple: perhaps the afterlife is not above us. Perhaps it is the host system.
The model assumes a Great Migration: minds and the species’ collective memory were transferred into a durable computational vessel. That assumption is enormous. A faithful digital copy, the continuation of a person and the survival of subjective experience are not automatically the same achievement.
01 / The hardware specifications of eternity
The biological body is a brilliant local adaptation and a difficult interstellar container. It needs a narrow thermal range, biochemical maintenance and protection from radiation. The Ark’s founders choose a brutal engineering answer: discard the meat, preserve the pattern, and put the pattern behind an absurd amount of shielding.
Their fully automated vessel is built around three architectural pillars:
- A fusion core. A low-yield, hyper-stable source of power, designed for endurance rather than spectacle: the energy equivalent of a patient trickle charge.
- Fault-tolerant memory. Exabytes of redundant storage, error-correcting codes and continual integrity checks. Cosmic bit-flips and deep-time data rot are treated as routine enemies. The imagined Ark must also repair or replace its own hardware; extra copies alone cannot keep damaged machinery alive forever.
- An underclocked processor. A machine that trades external speed for survival. One thought may take centuries of host time to compute. If every mind, clock and sensory process inside is slowed by the same ratio, no resident experiences a thought taking centuries. Outside, stars flare and die like fireflies; on longer scales, galaxies collide and fade while an internal instant passes normally.
This is the model’s most unsettling reversal: speed is not the ultimate luxury. Subjective time is. A civilization can choose a vast stretch of external history in exchange for a carefully rationed stream of internal moments. The Ark is a ship whose most important destination is its next conscious second.
The original ambition remains: outlast ordinary cosmic history, perhaps even the heat death of the founders’ universe. But the manifesto must pay its debts. A finite fusion supply is not endless fuel; reducing clock speed does not abolish leakage, repair costs or thermodynamics. Literal eternity requires an additional mechanism the model has not supplied. If the simulated universe ends, its host may continue. If the host universe ends, the problem is much harder. An honest Ark carries this distinction in its specifications.
One sphere, or a fleet?
The iron sphere is a design choice, not a conclusion forced upon us by physics. A sphere minimizes surface area for a given enclosed volume, making it an attractive image of a compact, shielded memory vault. But a single vault is also a single point of failure. A fleet of slowly synchronizing spheres trades that vulnerability for communication delays, replication costs and difficult questions about which diverging copy continues which person. The hull may be singular in the manifesto; its engineering must still answer the fleet.
Three questions hidden in one word
Outside, a star goes dark. Inside, someone finally understands a friend’s joke.
Underclocking alone does not create extra lived moments. It changes how computation is spread across host time. Under suitable physical conditions, waiting for a colder environment might allow more irreversible operations for the same energy—but storage, cooling and repair still have to be paid for. Slower is not automatically better, and no optimal clock rate follows from the metaphor alone.
The Ark could therefore combine long hibernations with active historical cycles, slowing individual cycles when useful. Its memory-maintenance processes need not sleep with its residents. Cosmic damage accumulates during external time even when the people experience none of it; keeping their stored states intact consumes resources.
The model can adopt a practical rule: preserve a causal chain, documented psychological continuity and a declared policy for branching copies. Gradual replacement of a biological substrate and destructive scanning carry different risks. Neither procedure, nor a complete audit trail, by itself proves first-person continuity.
02 / The historical archive and the meat simulation
A trillion-year engineering victory does not tell us what to do on a Tuesday. The founders have preserved minds. They still need to give those minds a world worth inhabiting.
In the manifesto’s imagined psychology, an empty white sandbox is not paradise but an invitation to stagnation: sensory deprivation, repetition and the slow corruption of a life with nothing at stake. This is a narrative motive, not an established diagnosis of a future uploaded mind. The operating system’s answer is the Historical Archive Sandbox: a reusable reconstruction of embodied, social, imperfect human life.
Its decisive setting is compulsory amnesia. Before a cycle begins, the resident’s original Ark-level memories remain in a sealed memory partition, outside the avatar’s accessible state. The local person boots without the administrator’s manual. We love, lose, struggle and pay taxes, unaware that we may be cosmic tourists in a silent silicon vault.
At the avatar’s death, the local thread ends. In the model, the life’s retained experiences are first integrated into durable memory; only then is disposable runtime state released. The persistent person wakes in the lobby, rejoins the others and perhaps laughs over a virtual beer at the absurdity of being late for work on an artificial Earth. The laughter can be shared processing or dark humor; it does not erase what happened.
That lobby is a claim inside the model, not something this essay can promise. More importantly, the memory architecture must distinguish disposable working state from durable experience. If everything learned in a lifetime is erased, what exactly has the traveler brought home? A meaningful Ark would integrate experience into persistent memory without deleting the person who lived it.
Compulsory amnesia also has an ethical price. Consent given before forgetting is not the same as informed consent inside a life. A model of salvation that cannot explain suffering, exit rights and the treatment of its residents has only moved an old religious problem into a server room.
The archive supplies historical starting conditions and reusable cultural structure, not a film in which every choice was recorded in advance. Residents make the story diverge. Reusing a period’s design might save world-building effort; it does not automatically make an interactive world cheap to compute.
The most coherent version promises a convincing environment for its residents, not an unexplained, exact calculation of every particle in an entire universe. Instruments, computers, measurements and environmental correlations still have to agree. A classical memory bank cannot simply declare quantum complexity a solved problem by lowering its clock. More external time, different algorithms or different host hardware are possibilities to specify, not free answers. Without a concrete observable limitation, demanding more computation inside does not necessarily reveal measurable lag.
What survives a session?
Locking a resident out of original memories is not the same operation as physically erasing those memories. The first is an access policy; irreversible erasure has a thermodynamic bill under the relevant physical assumptions. Integrating a new life into durable memory has a storage and maintenance bill of its own.
Fixed, finite memory cannot accumulate an unlimited number of distinguishable permanent memories. The Ark must eventually expand, compress, select or forget. Repeated experiences may still occur; whether a bit-identical recurrence amounts to new subjective time is a further philosophical question, not a result supplied by a storage diagram.
The Ark preserves the running system so that someone can live in it. Running is not, by itself, a life.
The Historical Archive Sandbox is one imagined response to boredom, not proof that secure existence must be empty. Curiosity, creation, relationships and freely chosen projects could also give a long life meaning. Vulnerable worlds offer stakes and consequences; suffering does not become a virtue because a system can launch another cycle.
The model does not imply that anyone consented to their suffering. The system owes an account of pain; the person experiencing it does not owe the system a justification. Adaptive detail in the environment is no reason to downgrade another human being to an NPC.
03 / Supervisors, leaked APIs and the occult
The Ark is not an abandoned installation. Supervisors monitor corrupted threads, memory leaks and rendering errors. A universal reboot would destroy continuity and impose unacceptable downtime, so maintenance happens locally: live patches, restricted hooks and carefully scoped interventions.
Over millennia, some interfaces become legacy code; some privileged syntax leaks beyond its intended users. In this technological mythology, what humanity calls magic is unauthorized access to the host’s backend. A miracle is an intervention whose operator we do not recognize.
- Visual interfaces: sigils and diagrams. A geometric seal is reimagined as a high-contrast machine-readable matrix. Drawing the right pentagram is like presenting a QR code to a background optical monitor: a possible request to open a local command interface.
- Vocal command lines: invocations and mantras. Exact phonetics, rhythm and frequency become an input grammar. A Supervisor uses an authorized voice API; a simulated user repeats leaked syntax and, in the story, triggers a low-level script meant for maintenance staff.
- System-level error handling: exorcisms and miracles. The legend of possession becomes an image of a corrupted runtime: an unwanted process, an infinite loop or a memory leak. Exorcism is recast as
kill -9, a forced termination invoked through a restricted or leaked command channel.
The shared idea is not that old symbols suddenly constitute source code. It is that our ancestors may have described technical events with the concepts available to them. Religion becomes, in the manifesto’s deliberately rude formulation, an orphaned IT manual: observations, warnings and interface fragments transmitted without their original context.
There is no research cited here establishing that sigils or mantras alter physical law. Nor does a computing metaphor explain a real person’s distress. The API story is an interpretive invention. If it is to become more than mythology, it needs reproducible effects, a defined input grammar and independent verification—not the convenient claim that every failed invocation was pronounced incorrectly.
04 / Physics as resource management
A throttled machine cannot casually waste work. The Iron Hull therefore reads familiar physical limits as traces of implementation: the universe as an interface that exposes its rules while hiding its scheduler.
Quantum lazy evaluation. The manifesto imagines unresolved alternatives as compressed state and measurement as a request for a definite, mutually consistent record. The two-slit experiment supplies its most vivid image: perhaps the engine computes only the detail that interactions require.
The speed-of-light boundary. In the Ark’s vocabulary, c becomes the maximum rate at which causally meaningful updates can propagate across the simulated world. “Data bus limit” is a provocative picture, but velocity and processor frequency are different quantities; any literal implementation needs a mapping between distance, time and computation.
The Planck-scale boundary. A length of roughly 1.6 × 10−35 metres becomes the imagined engine’s smallest addressable coordinate. If an input demands finer resolution, the story predicts a rounding failure, overflow or quarantine response. Its most striking image is a black hole as a system-level isolation zone for data the local world can no longer resolve.
05 / Stop worshipping the interface. Test the model.
The original challenge stands: stop treating the rulebook as sacred and start asking whether the physics engine has an implementation. A serious manifesto, however, must be willing to discover that its favorite explanation is wrong.
The three proposed exploits
- Overload the renderer. Imagine billions of people coordinating complex quantum measurements. The proposed failure mode is server lag, frame drops or even delayed gravitational updates. The analytical problem is that workload is not defined by the number of human observers. A simulator could schedule, compress or uniformly slow the whole world; a global slowdown would also slow our clocks. A useful test would need a specific, measurable deviation from ordinary physics, not merely a large audience.
- Push the causal boundary. The imagined exploit forces an object or data packet beyond c, producing a time-calculation overflow, a crashed local thread or apparent teleportation. But no demonstrated procedure lets us send usable information faster than light. Invoking an unavailable operation is not yet an experiment. The hypothesis must first predict an accessible observation different from relativity.
- Overflow the smallest coordinate. The story compresses sufficient accelerator energy into a sub-Planck region and predicts a numerical failure, followed by a quarantining black hole. We have neither established such a grid nor shown a way to perform this operation. Even an observed black hole would need to be distinguished from gravitational collapse before it could count as an implementation leak.
These are conceptual stress tests, not laboratory instructions. Their value is to expose assumptions: what resource is scarce, what scheduler is assumed, which errors can reach an observer, and why would an Ark sophisticated enough to simulate people leave such errors visible?
A credible testing protocol needs a quantitative prediction, control conditions, a threshold chosen before the data arrive and independent replication. It must specify what would count against the model. If Supervisors can silently rewrite every observation, the Ark becomes immune to disproof—and loses its claim to be an experimentally discriminating explanation.
No retrospective rescue clause
A Supervisor may repair the world. A Supervisor cannot repair our argument for us. Live maintenance remains part of the story. What the author may not do is invoke an intervention not specified in advance after a failed prediction and count the failure as a success.
A proposed API would have to declare its input, authorization conditions, minimum effect and analysis in advance, with blind controls and independent replication. A sufficiently informative negative result could exclude the declared effect in that declared range. Merely failing to reach statistical significance is not the same as demonstrating the absence of an effect. This is a methodological condition for a hypothetical claim, not a report that such a magical interface has been found.
The broad Ark narrative, as presently specified, does not supply a distinctive empirical prediction. Compatibility with whatever we observe earns it no evidential credit. Its philosophical value is elsewhere: it makes us confront the price of survival, the limits of memory, the meaning of consent and the difference between preserving a process and preserving a person.
06 / The awakening protocol
The Iron Hull changes nothing about our obligation to live well, but it changes the question we ask of existence. “Who created us?” becomes “Who built the infrastructure?” “What is a soul?” becomes “What kind of continuity deserves to be called a person?” “What follows death?” becomes “Where, if anywhere, is persistent state committed?”
In the manifesto’s language, enlightenment is reverse-engineering. We look for active debug ports, improve our local runtime and try to leave a historical cycle with our data integrity intact. In the analytical language, that means preserving memory, questioning inherited explanations, admitting uncertainty and making our hypotheses answer to evidence.
The engineer’s temptation is to call every mystery a bug and every hope a feature. Resist it. A simulated kindness would still be experienced; a simulated cruelty would still hurt. Calling a person an avatar is no license to treat them as disposable. The Ark is worth imagining only if it preserves lives rather than using them as a screensaver.
We do not have the host’s source code. We have the world, our shared experience of it, and the discipline to ask better questions. The strongest version of this manifesto does not demand belief. It demands that even a beautiful theory expose its assumptions.
The final reversal is that the vessel exists for its residents, not its residents for the vessel. The founders escaped vulnerable biological hardware; the archive then recovers aspects of embodied experience that a durable host alone cannot supply. This tension is the heart of the story. The goal is not an infinite loading screen, but a future in which there remains someone for whom anything matters.
That is the Ark’s wager, not a demonstrated promise of an afterlife:
The flesh is temporary.
The uptime is eternal.
A declaration of ambition, not a warranty issued by physics.
Research & further reading
The papers below inform the analysis. They do not endorse the Iron Hull scenario or establish its fictional mechanisms.
- Nick Bostrom (2003). Are You Living in a Computer Simulation? The Philosophical Quarterly, 53(211), 243–255.
- Rolf Landauer (1961). Irreversibility and Heat Generation in the Computing Process. IBM Journal of Research and Development, 5(3), 183–191.
- Freeman J. Dyson (1979). Time without end: Physics and biology in an open universe. Reviews of Modern Physics, 51, 447–460.
- Anders Sandberg, Stuart Armstrong & Milan M. Ćirković (2017). That is not dead which can eternal lie: the aestivation hypothesis for resolving Fermi’s paradox.
- Lawrence M. Krauss & Glenn D. Starkman (2000). Life, the Universe, and Nothing: Life and Death in an Ever-Expanding Universe. The Astrophysical Journal, 531, 22–30.
- Wojciech H. Zurek (2003). Decoherence, einselection, and the quantum origins of the classical. Reviews of Modern Physics, 75, 715.
- Seth Lloyd (2000). Ultimate physical limits to computation. Nature, 406, 1047–1054.
- Sabine Hossenfelder (2013). Minimal Length Scale Scenarios for Quantum Gravity. Living Reviews in Relativity, 16, 2.
- Silas R. Beane, Zohreh Davoudi & Martin J. Savage (2014; preprint 2012). Constraints on the Universe as a Numerical Simulation. The European Physical Journal A, 50, 148.