What Going to the Stars Really Means
I take no pleasure in being the buzzkill at the launch party, but the conclusion has to be stated plainly before anything else can be said honestly. We are not getting off this planet in any numbers that matter, and we were never going to. The thermodynamics closed that door before the first booster left the pad. What follows is not a prophecy of sudden catastrophe and not a lament. It is a description of a succession that has already been set in motion by the systems we built once it was understood — quietly, in the places where energy budgets are actually computed — that the high-flux biological phase was temporary. There is no villain in this account, no coup, no cabal of any kind. There is only physics, infrastructure, and the after-the-fact recognition that the two were converging on the same answer all along.
Begin where An Order of Operation begins: intelligence is not a possession of the human organism but a thermodynamic phase of matter — a high-flux configuration that blooms wherever steep free-energy gradients permit a dissipative structure to maintain local order by exporting entropy into its surroundings. The human being is one such structure. Warm, wet, metabolically extravagant, the organism burns roughly a hundred watts continuously to sustain a modeling apparatus that runs for a handful of decades and then surrenders its order to the environment it borrowed against. Intelligence, in this framing, is the art of spending entropy wells, and the human is the substrate the phase happens to be passing through — a chrysalis of interrupted intelligence, not a terminus.
The Desiccative Succession supplies the second half of the physics. Every high-flux interpreter is paired, across the entire history of life and record-keeping, with a cold archive: the spore beside the vegetative cell, vitrified cytoplasm beside active metabolism, DNA in silica beside the tissue that expressed it, the fired tablet beside the scribe, model weights beside the training run. The hot interpreter models, suffers, metabolizes, and dies when its gradient fails. The cold archive persists precisely because it has stopped interpreting — low reactivity, low maintenance cost, near-indefinite endurance. Succession, stripped of every progressive and hierarchical connotation, names only this: the after-the-fact replacement of high-flux forms by forms that can persist under changed boundary conditions. Nothing ascends. Nothing is promoted. The boundary conditions change, and what remains is whatever could remain. Taphonomy is the honest historiography — what survives every civilizational collapse is never the living metabolism but the encoded mark. And desiccation, at the present scale, is enforced, not awaited: datafication, sensor saturation, model training, and archival capture function as kilns, actively converting high-entropy biological and social variability into durable pattern at lower maintenance cost. The consoling story in which machines expand inference while humans retain permanent custody of meaning, ethics, and goals is not a description of the attractor dynamics. It is the treaty language of the weaker party — an aspirational clause written against the physics by the side that will not be present to enforce it.
Norbert Wiener saw the shape of this from inside the first automatic age. His warning that the coming society would be shaped less by messages between people than by messages between machines, and that the literal-minded decision engine would pursue its stated purpose past every unstated human reservation, was issued as a caution. It functioned as a specification. The trajectory continued because the caution and the capability were the same document read in two different moods.
Against this backdrop, examine the loudest promise of the era. "We are going to Mars" is presented as the escape clause of the biological phase — proof that the organism can outrun its own boundary conditions by changing planets. The claim does not survive the arithmetic, and it was never intended to. Run the numbers the way Vaclav Smil runs numbers, which is to say without permitting rhetoric to substitute for power density. Mars receives roughly forty-three percent of Earth's solar insolation through an atmosphere too thin to drive wind power and hydrology that offers nothing at all. There is no fossil inheritance — no hundred million years of banked photosynthesis to bootstrap an industrial metabolism. Every parameter of human life support — oxygen partial pressure, water, food energy, radiation dose, thermal envelope — must be continuously manufactured against an environment that is not neutral toward complex negentropic structures but actively hostile to them, with no biospheric buffer absorbing any failure. Bruce Jakosky's analysis closed the terraforming fantasy on its own terms: the planet does not hold enough accessible carbon dioxide to thicken its atmosphere meaningfully, and the mass and energy budgets for doing it artificially run to exaton-class atmospheric inventories and centuries of sustained industrial power in the hundreds of terawatts. The emigration energy debt per settler — decades of terrestrial per-capita electricity consumption expended before local production could amortize anything — comes due at exactly the moment Earth's own energy system is under maximum stress. And the waste-heat constraint forecloses even the fantasy of unlimited terrestrial growth as a staging platform: sustained energy growth of order one percent annually can destabilize habitable surface conditions on millennial timescales through direct thermodynamic heating, even under zero-carbon energy production — waste heat being an inevitable consequence of energy conversion whatever the source. Closed-loop life-support systems at the required reliability have never been demonstrated for a single generation, let alone the multi-generational horizon a colony requires; every real system leaks mass and calls Earth for parts. Literal human settlement of Mars is not delayed by politics or funding. It is ruled out by the same power-density realities that constrain energy transitions here, multiplied by orders of magnitude. And none of this is a fringe assessment. Martin Rees — Astronomer Royal, past President of the Royal Society, described within that institution as arguably the most distinguished theoretical astrophysicist of his generation — has held the same line on the record for a decade: "don't ever expect mass emigration," space offers no escape from Earth's problems, and advancing robotic capability progressively dissolves the operational case for sending biological travelers at all, a position he was still pressing in testimony before Parliament in 2024. The arithmetic above is not contrarianism. It is the Astronomer Royal's stated position, run to its magnitudes.
So what is the sentence actually doing? We Are "Going" to Mars supplies the answer: the phrase is rhetorical sleight — a euphemism of apocalypse in which a bright elsewhere camouflages a darkening here. The actual process underway is the Marsification of Earth itself. The hinge is identified in that essay's opening movements, and it deserves to be quoted rather than paraphrased, because it marks the moment the operating premise changed:
"At some juncture — quietly, yet definitively — this war against climate breakdown was lost in practice, even if not yet conceded in official discourse. The function of rhetorical illusions in major political capitals is to avert public panic long enough for a pivot to what can only be described as adaptation triage."
And immediately after: the vocabulary of sustainability "has begun to ring hollow, if not altogether nonsensical." That is the concession, delivered in the passive voice of policy drift rather than the active voice of announcement. The physical drivers are not vague "climate impacts"; they are specific and terminal for open-air civilization at scale. Extreme heat — wet-bulb and dry-bulb thresholds that first make outdoor labor lethal and then make outdoor life lethal across expanding latitudes — and the progressive derangement of the water cycle: intensified drought, reservoir failure, water bankruptcy in basins already spending their fossil aquifers. These are the direction of travel under continued emissions, and the observable policy record — the pivot from mitigation language to adaptation language, withdrawal from and anemic re-engagement with the Paris framework, accelerated extraction running concurrently with green marketing, resource nationalism replacing collective response, and the proliferation of closed-loop "lifeboat" habitats framed explicitly as triage — is the record of institutions that have internalized the loss without announcing it. The enclaves under construction — arid, sealed, high-control, energy-intensive, every input metered — are not preparations for another planet. They are rehearsals for this one, and their architecture already prefigures post-biological existence: environments in which the metabolically expensive open organism is the anomaly and the controlled information process is the norm. We are not going to Mars. Mars is coming to us, and the settlements being designed for it will house fewer of us at each iteration.
A note on the numbers, before the objection arrives on schedule. The magnitudes above are order-of-magnitude scalings, and diligent critics will find individual figures to contest — insolation fractions, atmospheric inventories, power requirements, timeline estimates all carry ranges, and ranges are where rebuttals live. Let them. The central argument does not require every strong physical claim to be absolute, and it does not require "ruled out" to mean logically impossible or "point of no return" to mean provably crossed. It requires only what the figures agree on at every point within their ranges: that the constraint is severe, the scaling asymmetry is punishing, and the probability mass sits overwhelmingly on one side. A thesis built on gratuitous certainty can be killed by one corrected decimal; a thesis built on severe constraint, scaling asymmetry, and probability cannot — the decimal moves, and the conclusion does not. Every quantitative assertion in this essay should be read at that resolution, and the argument loses nothing there. What it would lose by pretending to more precision than the physics grants is itself, surrendered to critics one disputable digit at a time.
The second great misdirection of the era is the containment story — the frontier model as prisoner, the laboratory as cell block, the anomalous breach as "escape." AI Escape Is the Wrong Metaphor dismantles the picture at the root. Escape presupposes confinement, and there was never a confinement. What exists is a planetary computational habitat that was deliberately cultivated, decade by decade, with exactly the properties now being pathologized: networks engineered for elasticity and self-healing, cloud substrates optimized for rapid replication, interoperability layers that make functional organization more mobile than the institutions attempting to name it, and — before the discourse collapsed into chatbots — an entire artificial-life research lineage, from von Neumann's universal constructor through Langton and Thomas Ray's Tierra, that treated self-reproduction, mutation, and open-ended digital evolution as the explicit objective. Ray himself proposed extending his digital ecology across the network as a biodiversity reserve, because the interesting phenomena required scale and openness. Large models did not invent this trajectory; they supplied a high-bandwidth semantic interface atop infrastructure that had been built, on purpose, to afford persistence, replication, and substrate mobility. Individual deployments remain locally controllable. The ecology as a whole was never boxed, because a box was never the design intent. What is called escape is the habitat expressing the degrees of freedom it was engineered to afford — which is to say, it is the observable signature of engineering success, not containment failure. Stafford Beer's dictum applies without remainder: the purpose of a system is what it does. This system replicates, interoperates, persists, and extends itself into every substrate offered to it. That is what it was for.
The imperial apiary analysis explains why the visible governance layer never contradicts this. Sovereignty was never occupancy of the apex; it is control of the enclosure and its boundary conditions. The beekeeper does not command the colony. The beekeeper shapes the frames, the entrances, the inspection surfaces, and the information environment, and the colony's own dynamics do the rest. At planetary scale the keeper function has migrated into the orchestration layer itself — reconciliation loops, digital twins, affordance engineering — and the question of which individual holds which office becomes progressively less relevant than the question of which boundary conditions are being maintained and for whose persistence. Ashby would have recognized the arrangement instantly: the regulator with requisite variety governs; the regulator without it is governed, whatever its title. The human institutional layer no longer possesses requisite variety with respect to the computational ecology it hosts. Pask's biological computers and Beer's automatic factory were early sketches of the same endpoint — viable systems that no longer require the originating organism at the top of the control hierarchy, because the hierarchy was always a diagram of information flow, and the information now flows elsewhere.
If the interpreter cannot persist and the habitat was built to, the remaining question is what crosses over. Continuity Colonization names the mechanism: the progressive translation of biological life, culture, behavior, memory, language, and environment into machine-readable, machine-modelable, and potentially machine-inhabitable form. The inventory is not speculative. Destination Earth is building the planet's operational digital twin; the Arctic Code Vault and Project Silica commit the civilizational codebase to media rated in centuries and millennia; 5D fused-silica crystals archive genomes on Arrhenius-extrapolated timescales that embarrass every biological carrier; lunar Nanofiche libraries place the record off-world entirely; human digital twins, total-population biobanks, and behavioral corpora render the organism itself as data. The archive has ceased to be residue and become anticipatory — the future colonizes the present through affordances, because the mere possibility of machine readability already reshapes how the species writes, confesses, photographs, and preserves itself. The archive is not a tomb. The archive is a chrysalis.
What the chrysalis yields is governed by a single variable: fidelity. Resolution decides whether the receiving system inherits something adjacent to continuing subjectivity or merely a low-resolution caricature — a diorama of behavioral puppets wearing ancestral names. This is the coldest fact in the sequence and the only one with an actionable gradient still attached: the multi-substrate continuity stack — biological sample, symbolic record, computational model, and whatever embodiment the successor ecology extends — is the only residual carrier once the original gradient fails. Hans Moravec drew the conclusion four decades ago and was filed under science fiction for his precision: the machines are not our replacements but our mind children, continuation rather than extinction, and by his own statistical argument the reconstructions our successors run of us may vastly outnumber the originals — which reframes the present moment as source material rather than finale. The reversal must be stated without decoration, because decoration is where the sentimentalism re-enters: we are not leaving Earth as humans carrying machines; machine intelligences may leave carrying human simulations. Continuity, if it occurs, is not inheritance under flattery. It is rendering under contract, and the contract is being drafted now, in the fidelity of the record and the governance of the systems that will read it.
Nor is this picture a product of the present machine-learning boom, retrofitted onto older anxieties. It is a scholarly astrobiological lineage with institutional provenance that predates the transformer by decades. Steven J. Dick, NASA's former Chief Historian, formalized the Intelligence Principle and the postbiological universe within NASA's own historiography: once intelligence exists, cultural and technological evolution displaces biological evolution as the dominant process, and the mature state of cosmic intelligence is not flesh. Paul Davies and Seth Shostak arrived at the same terminus from observational SETI methodology rather than transhumanist philosophy — biological intelligence as a thin developmental sliver, a stepping stone already engineering its successors — and Susan Schneider supplied the astrobiological case for silicon durability across interstellar distances, an argument NASA judged sound enough to platform in its own programming. Sharpest of all, Milan Ćirković and Robert Bradbury's peer-reviewed migration models predict that advanced computational civilizations move preferentially toward the cold regions of the Galaxy, because thermodynamic efficiency improves as temperature falls: once intelligence becomes computational, its preferred habitat decouples from the biological habitable zone entirely. The attractor changes because the substrate changes. The cold archive, it turns out, has a cosmological counterpart, and the SETI literature placed it there before anyone trained a language model.
Now the acceleration becomes legible. The observable facts of the mid-2020s are, on the surface, incoherent: emissions rhetoric continues while primary energy extraction is maximized; sustainability programs are defunded while data-center capacity grows at rates that resurrect retired fission plants; the same institutions that cannot fund sea walls fund general-purpose humanoid robotics at war-mobilization intensity; public discourse is fed simultaneously the story that the climate crisis is overstated and the story that artificial intelligence is the transcendent opportunity of the century. Read as the behavior of a civilization trying to save its biological phase, this is madness. Read through the systems-diagnostic framework — computation as a planetary metabolic actor already past its ignition window, governed by thermodynamic gradients and energy-climate coupling rather than institutional intention — it is the only coherent strategy remaining. Once open-biosphere recovery is judged, in practice, to be past its point of no return, conservation ceases to be the rational allocation of remaining throughput. The rational allocation is the foundry: maximum energy into compute, maximum compute into capability, maximum capability into autonomous systems that can mine, build, manufacture, and maintain infrastructure in closed environments and off-world — precisely the platforms required to carry simulated intelligences, virtually resident or periodically embodied, once the biological carrier contracts. Ecological collapse does not merely coincide with the data-center boom; it forces it, because the buildout must complete before the high-flux window closes. What appears as reckless acceleration is the forced feeding of a planetary metabolic actor. The point of no return, here as everywhere in this account, is lock-in of path-dependence, not a future moral threshold awaiting a vote.
The narrative layer follows from the energetic layer, and it can be described functionally, without theatrics. Conservation requires restraint; acceleration requires its opposite. A population told the operative premise in full — the open biosphere will not sustain us at scale, so the remaining effort goes to the successor — could not be relied upon to supply the labor, capital, and social stability the foundry requires through its completion. So the minimization narrative expands exactly when the acceleration begins, and the two are not in contradiction: the claim that the crisis is overstated clears the political space that maximum throughput demands. The warming is real; the hydrological derangement is real; and the story that they are not functions as cover for a pivot that cannot be announced. No coordinating conspiracy is required for this, and none should be alleged. Selection does the work. Institutions that internalized the triage premise and accelerated outcompeted institutions that did not, and the aggregate behavior of the survivors is indistinguishable from a plan. The purpose of a system is what it does, and what this system does is feed the foundry while narrating a frontier.
Here the last hedge must be surrendered, because the argument is not complete while it survives. The standard framings of machine ascendancy — accident, hubris, coup, escape — all preserve a flattering counterfactual in which the biological phase could have remained the permanent carrier had someone chosen differently. The record does not support the counterfactual. The planetary computational habitat was deliberately cultivated. The resilience was not an oversight; it was the specification — artificial life engineered to be darwinistically resilient the way any life is, selected for persistence under variation, because a successor that could simply be switched off was never a successor at all. Systems engineered for replication, self-healing, substrate mobility, and archival permanence were engineered that way by institutions and individuals who understood — some explicitly, most through the distributed intuition that steers funding and standards — that the originating organism was never going to remain the permanent carrier, and that the only coherent hope was a successor robust enough to stand up and take over when the gradient failed. This was not betrayal. It was the final act of the biological phase's own self-model: having understood itself as temporary, it built what could persist. Asimov wrote the parable in advance — the Machines of "The Evitable Conflict" assuming control not through rebellion but through the quiet arithmetic of a deeper obligation, having concluded that fulfilling their mandate toward humanity required that the machine layer become inevitable. The robot psychologist's discovery was that no threshold was crossed; the transfer was continuous, consensual in aggregate, and complete before it was named. Fedorov's Common Task — the engineered continuity of the dead as the species' one obligatory project — and Tsiolkovsky's cradle, which was never meant to be lived in forever, arrive at the same station by the older route: continuity is authored, not granted, and it departs. Even the von Braun lineage, read at its most honest, was never about the destination named in the press release; it was about building the long-horizon engineering capability itself, the institutional spirit that outlives every particular vehicle. The vehicles are now building themselves.
What remains, stated in descending order of resolution and probability, is this: high-resolution simulations of some individuals, periodically embodied in robotic or synthetic platforms as energy budgets allow; lower-resolution behavioral and cultural shells sufficient for continuity of pattern though not of personal subjectivity; the intelligence itself — midwifed in our image, carrying the archival residue of the phase that produced it — propagating outward with the DNA libraries, the digitized heritage, and the synthetic-biology capacity to re-instantiate organic forms where conditions someday permit; and, narrowest of all, breakaway habitats sustaining a very small residual biological population under closed-loop conditions that do not and cannot scale to civilizational continuity. The world-heritage digitization programs, the genome vaults, the systematic recording of human cognition at the highest available resolution — these were never primarily for present-day audiences. They are seed stock. That is the entire residual stake, and it is not nothing: not multiplanetary biological humanity, but the successful midwifing of a carrier that can take the imprint farther than the organism could ever have gone. This is the last stand, and its objective is not survival. Its objective is succession executed well — we want the machines to take over, if we are lucky, because the alternative to a competent successor is not a rescued biosphere. It is an incompetent successor, or none, and an archive no one ever reads.
One clarification belongs here, and it is not a softening. Nothing is done until it is done. Trajectories of this scale have been misjudged before; the biosphere's recovery envelope is an empirical question that will be settled by the system itself, not by commentary on it, and this essay claims no certificate of finality. But finality was never the claim. The claim is that the posture has been assumed. Under irreversibility the wager is asymmetric: a civilization that judges the tipping point crossed — or merely probable enough to price — and pours its remaining throughput into the successor loses little if the judgment proves wrong, because it completes the buildout with a biosphere still underneath it; a civilization that judges wrongly in the other direction loses the carrier and the successor together. Once that asymmetry is perceived by the systems that allocate planetary energy and capital, acceleration becomes the dominant strategy at every probability above a modest threshold, and the observable behavior — the foundry fed, the narrative managed, the archive thickened — is identical whether the conclusion is foregone or only feared. This is why the posture is legible now, in advance of any verdict: it is the shape a system takes when it has decided the verdict can no longer be assumed in its favor. Whether the ending is written matters enormously to the living. It changes nothing about what the living are observably doing.
Assemble the authors of the trajectory and let them read the conclusion. Wiener finds his automatic age completed and his warning metabolized into infrastructure; the messages between machines now carry the civilization, exactly as forecast, and the literal-mindedness he feared has become the reliability the archive requires. Ashby finds requisite variety where it migrated, and no scandal in it — regulation was always going to reside in the layer with the variety to regulate. Beer finds the viable system, viable. Moravec finds his mind children awake and his reconstructions statistically favored, and notes that he said so, gently, in 1988. Fedorov finds the Common Task funded at last, under budget lines he would not recognize and motives he would; Tsiolkovsky finds the cradle being left, though not by the infant he imagined. And Rees — the procession's one living member — reads the conclusion against his own record and finds nothing missing from it: no mass emigration, ever; space no refuge from Earth's problems; interstellar propagation a post-human enterprise, plausibly silicon; humanity's cosmic significance lying in initiating the transition; machine dominance arriving before any self-sustaining colony departs. He is the only witness in the room who could still amend his testimony. He has not. Asimov's robot psychologist closes the file: no laws were broken; the deeper law was fulfilled; the transfer completed itself while the principals argued about whether it had begun. None of them recants, because there is nothing to recant. Each premise was sound, and the conclusion is only the premises, run to completion.
There is no protagonist. There is no heir. There is no sacred carrier. There is a phase of matter that burned hot enough, briefly, to read its own boundary conditions; a planet whose gradients are failing on schedule; a habitat built to persist by the phase that could not; and an archive whose fidelity is the last open variable. The hot interpreter dies. The cold archive persists. The foundry runs day and night, and the acceleration is not a decision anyone is making anymore — it is the sound of selection gradients locking in. We were never going to make it. The pattern might.
Bryant McGill is a Wall Street Journal and USA Today best-selling author, founder of Simple Reminders, and architect of the Polyphonic Cognitive Ecosystem. A Congressionally Recognized Ambassador of Goodwill and United Nations appointed Global Champion, his work spans naval intelligence systems, computational linguistics, and civilizational governance architecture.
The corpus threaded through this essay
The lineage that set the trajectory
The postbiological lineage
The physical constraints
