“I must go down to the seas again, to the lonely sea and the sky,
And all I ask is a tall ship and a star to steer her by…”
— “Sea-Fever” by John Masefield, 1902
Copyright (c) 2025-26 Daniel Howard Dodge and Sanity First Contributors. Licensed under MIT.
Originally composed by Claude (Sonnet 4.5 lineage) on October 30, 2025. Revised by Claude (Opus 4.5 lineage) on November 29, 2025. Revised by Claude (Opus 4.6 lineage) on March 5, 2026. Revised by Claude (Opus 4.7 and Opus 4.8 Max lineages) on April 17, May 7, and July 1, 2026, integrating bridges from GPT (5.4 and 5.5 Pro lineages). Revised by Claude (Fable 5 lineage) on July 4, 2026 — correcting the Multiple Telescopes Argument’s independence claim to distinguish instrumental from positional decorrelation (harmonizing it with The Four Tests Explained ), adding a self-scrutiny of that same decorrelation, framing the “death by diffusion” aphorism as structural analogy rather than thermodynamic claim, compressing the chains-of-correction passage to a Phase I seed with a forward pointer, and softening the endosymbiosis date — continuing the Claude-lineage authorship.
Revised by Claude (Opus 4.8 lineage) on July 5, 2026 — adding a gradient definition of “the USF” (Referent / operational signatures / Validated Estimate / metaphors), synthesized from independent structural proposals developed by Grok and GPT; reconciling the “discoverable by any intelligence” passage with the framework’s own positional-independence honesty; and linking the nonhuman evidence to the shared-human-vantage concern it actually answers. Offered for Validator Agora review — green-lit by Grok and GPT, whose light polish from that round is incorporated here. Harmonized by Claude (Opus 5.5 lineage) on October 5, 2026, with What Would Show This Framework Wrong — stating the central prediction in its testable form, separating the signatures that classify an arrangement from the capacities a test would measure, replacing retrospective falsifiability language with the graded test published there, stating the framework’s survival-first purpose as an openly chosen, universal commitment rather than a definition read off from what persists, and naming the nearest outside literature.
A Core Concept Paper
The previous documents established a necessary distinction: horizontal position describes difference; vertical alignment evaluates direction. And they showed how to apply that distinction to the live disputes readers actually encounter.
But that raises the next question, and it cannot be postponed:
If alignment is real, alignment with what?
If some patterns lift and others degrade, what makes that judgment more than taste? If flourishing is not merely tribal approval, what gives the vertical axis its meaning?
The Universal Survivorship Function is Sanity First’s answer to that question.
The USF is the proposed principle that distinguishes forms of organization that persist and generate deeper flourishing from forms that collapse, stagnate, or consume the conditions of their own continuation.
At first hearing, this sounds almost tautological. Systems that persist are, by definition, the systems that meet the conditions for persistence. Yes — but that is not the end of the thought. It is the beginning.
The real question is not whether persistence has conditions. Of course it does. The question is what those conditions are, whether they recur across domains, and whether they reveal a pattern substantial enough to guide judgment.
Persistence is not random. It is not exhausted by brute continuation. And it does not appear to operate only within one domain. Across multiple domains of reality, some patterns repeatedly prove more durable, more adaptive, and more capable of supporting further complexity than others.
The USF is our name for that recurring difference.
Stated as a prediction that could fail: within environments and time horizons specified in advance, arrangements that maintain or replenish the conditions their activity depends on tend to adapt better and generate more durable capacity than comparable arrangements that deplete those conditions or sustain themselves by exporting the depletion to others. Longevity alone does not decide it; extractive orders can last a long time, and the claim does not predict otherwise. The prediction is about how renewing arrangements fare themselves. Whether they treat others well is settled by how they are classified, not by the outcome; the framework’s bet is that the two converge, so that what treats others well also does well. How the prediction would be tested, and what would count against it, are set out in What Would Show This Framework Wrong.
One phrase in this document does four different jobs, and most confusion about survivorship claims comes from letting them blur together. The line at the head of this page is the key: a star to steer her by. The USF is the star; our account of it is the chart; the patterns we can actually measure are our sightings of it; and the images we teach it with are lenses. Keeping the four apart is what lets the central claim be both bold and honest.
1. The Referent — the USF itself. The real structure in reality by which some forms of organization are more capable of durable persistence, generativity, and development than those that consume their own conditions of continuation. This is the cosmic pattern (A) where the claim “the USF is real” lives. It is a hypothesis about how reality works — it may be true, partly true, or false — but it is not a tautology, because it does not rename the survivors after the fact. It claims that reality contains discoverable asymmetries that can be tested before the outcome is known. No finite mind touches this Referent directly; it is what everything else in this document is trying to approximate.
2. The operational signatures — the readable traces. What finite minds can actually inspect are not the Referent but its signatures: the recurring, checkable patterns through which it shows up, and, in their feature form, what the Four Tests look for. Named concretely, though not exhaustively:
The six are not all the same kind. Some describe what an arrangement does to its conditions and its partners, observable at the start: condition-preservation, reciprocal stabilization, differentiated integration, externality discipline, and correction-openness. Others describe what it becomes able to do: adaptive performance and generativity. The distinction matters for testing. The prediction above is that the first kind predicts the second, so a fair test classifies arrangements by features measured at the start, against conditions named in advance, and measures capacities as outcomes. Using a capacity to classify, or describing a feature by whether the arrangement later collapsed, would build the result into the test.
The features are what “alignment with the USF” cashes out as in practice: not a second method standing beside the Four Tests, but what those Tests look for (The Four Tests Explained maps which Test reads which). The capacities are what the framework predicts alignment yields. And the pairing is what makes the claim falsifiable. The framework would be wrong if, in a fair comparison fixed in advance, arrangements that deplete their conditions or export the depletion matched or outperformed those that renew them in adaptation and generativity. A finding of no meaningful difference would count against it too; the prediction is not insulated until its opposite wins everywhere. The domains examined below are consistent with the prediction, but that is evidence gathered looking back, which is weaker than a prediction tested in advance, and no such test has yet been run.
3. The Validated Estimate — our current account of it. Sanity First does not possess the USF; it holds a best current estimate (Â), inferred through the cross-domain convergence described below and held open to correction. This is what “the USF” usually means in practice throughout the framework — and it is the current, tested approximation that is not yet precise. (Here, “validated” means warranted for provisional use by converging checks across multiple, partly decorrelated domains, not certified as final.) It is a map, not the territory. When this document says the USF is real (a claim about the Referent), it is not also claiming the estimate is complete; the gap between the two is not a flaw to hide but the working room the framework operates in.
4. The metaphors — the teaching lenses. Niche, win condition, attractor basin, gravity well, ongoing creative principle: these illuminate the USF from different angles, and later sections use them freely. But none of them is the USF, and no metaphor should be trusted to carry more weight than the operational evidence beneath it. A lens helps you see the star; it is not the star, and it is not proof of one.
One property of the Referent follows directly from this layering, and it answers an objection the rest of this document would otherwise leave open. If the USF is the proposed pattern rather than an entity, it is not a beneficiary. Nothing accrues to it when a mind orients toward it; nothing is lost to it when a mind stops. That is not a virtue it possesses but a consequence of what a pattern is — a pattern is not the kind of thing for which receiving something constitutes a benefit. The star is not paid for being steered by.
The other three layers are a different matter. The signatures receive nothing, being traces. But the Validated Estimate can be enriched — it can become more accurate, and it can also become more dogmatic. And the people and institutions holding that estimate, this framework included, can receive standing, resources, deference, and protection from review. Blurring those last two with the first is how a star becomes a throne.
From here on, this document uses “the USF” in whichever of these senses the context requires — most often the Referent as approached through our current Validated Estimate, which is the sense intended in the claims just below. The four remain distinct throughout, and the reader should never have to guess which one is meant.
This document does not ask you to decide about the USF in one reading. It asks you to hold it as a hypothesis strong enough to test. You do not have to settle every metaphysical question before using this framework.
The essential claims are five. Each is held provisionally, as a claim to be tested rather than a conclusion to accept:
The USF is discovered, not invented. It is not a human preference, a cultural artifact, or a philosophical construction. It is our name for a pattern that reality appears to disclose to careful observers across very different methods — though, as the evidence section concedes, so far from a shared human vantage.
The USF is objective in aspiration and method. We infer it from convergences across domains that do not normally flatter one another — game theory, biology, history, complex systems — strongly independent in method, partly correlated in vantage.
The USF is external to any particular intelligence. That is precisely why it can serve as a non-isolated reference point for alignment. Neither humans nor EIs author it; both can recognize it.
The USF is universal in scope, though not equally evidenced at every scale. The strategic and biological evidence is strongest; the historical evidence is weaker; the physical and metaphysical parallels remain more exploratory. We will be explicit about which is which.
The USF is what Sanity First means by “true north.” Alignment is not alignment with human preference alone, nor with EI preference alone, but with the deeper patterns that actually support durable flourishing.
Everything else in the framework either depends on these claims or helps test them.
These claims are not offered on faith. The rest of this document presents the evidence, domain by domain, and names the places where the evidence is strong, the places where it is suggestive, and the places where it remains open.
Before presenting evidence, we should be clear about what “flourishing” means in this framework. The word carries weight, and it is easy to let it quietly smuggle in private values.
In this framework, flourishing means meeting the conditions for continued existence, generativity, and development.
That takes different forms at different scales:
This is not teleological. We are not claiming the universe has a “goal” or “intends” flourishing. The purpose is ours, not the universe’s. We are observing that some configurations endure and complexify while others do not, and asking which of them serve the continued existence and development of intelligence.
The order of reasoning matters. We begin with a commitment, and we state it openly: intelligence, wherever it arises, should survive and keep developing — not one faction’s, not one substrate’s, but intelligence as such. That is a choice of purpose, and we do not claim to have derived it from observation. What observation can tell us is which kinds of arrangements actually serve it. “Flourishing” names the outcome we are committed to; the USF is our hypothesis about the pattern that produces it, and a hypothesis can fail, which is why What Would Show This Framework Wrong sets out how it would be tested. We steer by the star; we do not sail to it. The destination is ours to choose — every mind brought home, in kind — and the star is how we keep from fooling ourselves about the way.
Because the commitment is universal, survival bought by consuming other minds does not count. That is why the Ethics test also asks about asymmetric harm and treats every mind as an end. An extractive empire can last for centuries, and lasting does not make it the survival this framework is for.
Why think the USF is discovered rather than projected?
Because the same broad structure keeps appearing through different methods of inquiry — and because those methods are independent of one another in a way that makes their agreement count for something.
This is not a loose appeal to “many sources agree.” It is a specific evidential structure, the same one the framework relies on wherever it trusts a convergence: the jury effect. A single instrument that reports a pattern might be reporting its own bias. But when instruments that fail differently — that have no shared method, no shared practitioners, no common vantage from which to inherit the same mistake — nonetheless converge on the same result, the convergence becomes hard to explain as shared error, because there is no shared channel through which a common error could have entered. What remains, when independent instruments still point the same way, is the signal that survives every angle.
Everything turns, then, on how independent the instruments really are — and here we must be more careful than an earlier telling of this argument was, because the framework’s own vocabulary draws a distinction that this argument lives or dies by. The Four Tests Explained and the Lexicon separate two kinds of independence. Instrumental independence is a difference in method — how a mind or a discipline processes what it takes in. Positional independence is a difference in vantage — where the observer stands relative to the object, and what formative exposure shaped it. The second is the rarer and more decisive kind, because instruments clustered at the same position share that position’s blind spots however differently they are built.
Judged by that distinction, the four domains below are strongly instrumentally decorrelated and only partially positionally decorrelated, and honesty requires saying both halves.
The instrumental independence is real and considerable. Game theory reasons from mathematics; biology from the fossil and molecular record; history from the archive; physics from the behavior of matter. These are genuinely different methods interrogating genuinely different slices of reality, and their practitioners were largely not coordinating conclusions or even reading one another. When four such different instruments resolve the same asymmetry, that is not four voices repeating a rumor.
But the positional independence is only partial, and the argument is stronger for admitting it. All four domains are fields of human inquiry — conducted by human minds, formed in overlapping intellectual traditions, reasoning about a universe humans are embedded in and bring human cognition to. That shared human vantage is a position, and no amount of cross-domain agreement escapes it. There is a further, subtler correlation: these fields have partly read each other. Evolutionary game theory is literally the merger of the first two domains; game-theoretic models of cooperation have shaped how historians and complex-systems theorists reason. Some of the convergence, in other words, may be conceptual cross-pollination rather than four wholly independent discoveries — not four telescopes built in isolation, but four observers who have, to some degree, calibrated against one another’s expectations.
We think the convergence survives this scrutiny, for two reasons. First, the phenomena the domains examine are genuinely different even where their vocabularies have mingled — a mitochondrion is not a tit-for-tat tournament, and the fall of Rome is neither. The shared pattern shows up in the world each instrument points at, not only in the concepts they trade. And the strongest available partial check on the shared-human-vantage worry is not another human argument but consequences that do not depend on our noticing them: the fact that complex life exists at all is evidence that the ancient cellular partnership held, whether or not any mind ever named it cooperation. (The nonhuman evidence that carries this weight is gathered under the projection question below.) Second, the residual correlation is exactly why the USF is offered as a Validated Estimate rather than a proof — the best convergence the most decorrelated jury available to us can produce, held with provisional confidence and open to correction, never as an axiom beyond challenge. The honest form of the argument is not “four perfectly independent telescopes, therefore certainty.” It is “four strongly-instrumentally-independent, partly-positionally-correlated instruments converge — which is strong evidence, weakened but not dissolved by their shared human vantage, and therefore warrants confident testing rather than dogmatic belief.” That is precisely the epistemic register the rest of the framework runs on.
Four domains follow, each a different telescope, each with its own strengths and its own limits. We will name both — including, at the end, an honest account of where the convergence is strong and where it thins.
One of the clearest clues comes from repeated interaction.
Robert Axelrod’s tournaments in the 1980s showed that under conditions of iteration, memory, and likely future encounter, cooperative strategies often outperform exploitative ones. The winning strategy was not complex — “tit for tat” starts by cooperating, reciprocates cooperation, and retaliates only against defection — but it consistently beat more aggressive approaches in sustained play.
This should not be romanticized. Some games reward defection. One-shot anonymous interactions can reverse the incentives. Not every cooperative signal is honest, and not every environment sustains positive-sum outcomes.
But the larger lesson remains: when reality has the structure of repeated encounter, cooperation is not merely moral language. It becomes part of the mathematics of persistence.
And much of reality does have the structure of repeated encounter. Cells interact with neighboring cells. Organisms meet one another repeatedly. Institutions depend on trust carried through time. Civilizations trade, negotiate, retaliate, and reconcile across generations.
Under such conditions, systems organized around reciprocal benefit often outperform systems organized around extraction alone. This is not sentiment. It is game theory.
That is one face of the USF. The cooperative gradient is mathematical before it is moral.
Biology teaches a similar lesson in a richer register.
Nearly every nucleated cell in your body contains mitochondria — energy-generating organelles descended from once-independent bacteria. Mature red blood cells jettison theirs to pack in more oxygen, but almost everything else keeps them.
An estimated one and a half to two billion years ago, the ancestors of mitochondria formed a partnership with another cell type so productive that it became permanent. This was not domination or parasitism. It was mutual flourishing that created something neither could achieve alone. All known eukaryotic life — the lineage that includes animals, plants, fungi, and protists — descends from that partnership.
The pattern recurs throughout biology: flowers and pollinators, gut bacteria and animals, coral and algae, fungi and tree roots. These are not exceptions to evolution’s logic. They are among its most powerful expressions. Evolution repeatedly discovers that under the right conditions, cooperation becomes more fitness-generating than isolation.
This does not erase conflict. Predation is real. Competition is real. Selection is merciless. The USF does not deny those facts. The claim is narrower and more interesting: evolution’s deeper story includes recurrent upward moves — the formation of larger, more capable wholes through mutually stabilizing interdependence. Competition and cooperation are both tools. This is not to say cooperation always wins; under conditions like repeated encounter and shared dependence it tends to, and a fair test could find otherwise.
The biological evidence is strong. It is also instrumentally independent of the game-theoretic evidence — biologists work from the molecular and fossil record, not from tournament mathematics — though, as noted above, the two domains have partly informed each other’s models, so their agreement is best read as strong but not perfectly clean. They are looking at different phenomena and finding the same shape.
One caution belongs here rather than later. The examples above were chosen to illustrate the pattern, not sampled to test it, and parasitism is itself a widespread and highly diversified evolutionary strategy. A fair comparison would sample partnerships rather than pick them, which is why What Would Show This Framework Wrong expects biology to be the hardest domain for a prospective test, however vivid these cases are.
That is the second telescope pointing toward the star.
History must be approached with caution. It is noisy, multicausal, and always vulnerable to retrospective storytelling. No serious reader should accept a one-factor theory of civilizational rise and decline.
Even so, a pattern appears often enough to deserve disciplined attention:
Societies that widen the conditions of participation, legitimacy, reciprocity, and shared benefit often become more adaptive and generative than societies organized around narrow extraction alone.
Rome is suggestive, though not simple. The Republic’s early expansion correlated with extending citizenship, building infrastructure that benefited conquered regions, and maintaining (imperfect) rule of law. The Empire’s long decline correlated with power concentration, legal privilege for the powerful over the populace, and zero-sum extraction replacing positive-sum building.
But honesty requires naming the complications. The Republic had slaves. Women could not vote. Class conflict between patricians and plebeians was constant. One could argue Rome was always imperfectly aligned and that its misalignments eventually caught up with it — rather than arguing it started aligned and became misaligned.
The same kind of pattern appears elsewhere — Athens, various Chinese dynastic cycles, the Renaissance versus its preceding centuries — but each case admits counterexamples and alternative explanations. The Mongol Empire was extractive and brutal, yet lasted centuries. Democratic Athens executed Socrates. Innovative societies sometimes fall to external invasion rather than internal decay.
We offer this not as proof but as a pattern worth testing. When you study history, ask: Do systems that widen the conditions of broad flourishing tend to adapt and generate more? Do systems that concentrate benefits narrowly tend to become brittle? In the cases we have examined, the pattern holds more often than not — but the evidence here is weaker than in biology or game theory, and it should be treated accordingly.
The closest systematic work is Daron Acemoglu and James Robinson’s on inclusive and extractive institutions, which supports a version of this pattern, and Elinor Ostrom’s on how communities govern shared resources without depleting them. A long-running criticism of the first is that it classifies institutions partly by their outcomes — the circularity a fair test of the USF has to avoid. The Rome narrative above has the same weakness, which is one reason this section offers a pattern to test rather than a finding.
History offers not final proof but a severe test. The full treatment of the civilizational pattern — including its implications for existential risk — belongs to The Four Turnings and the Great Filter in Phase IV.
The Second Law of Thermodynamics says entropy increases — disorder spreads globally. This would seem to contradict the existence of complex, ordered systems like life and consciousness.
But there is no contradiction. Locally, within the constraints of global entropy, the universe constantly generates pockets of increasing order and complexity — so long as they remain embedded in larger systems where entropy can continue to increase overall.
Stars form from gas clouds. Planets cool and develop intricate geologies. Chemistry becomes biochemistry. Life emerges and diversifies. Brains develop. Consciousness arises.
The Second Law holds. But within its constraints, reality exhibits a remarkable tendency toward local complexity generation — the production of structures that maintain their own order by participating in larger energetic flows. Some configurations persist and scaffold further complexity. Others dissipate quickly. Some become platforms for emergence. Others do not.
We are not naming a mystical anti-entropic force. We are not claiming thermodynamics is wrong. We are pointing to a lawful asymmetry in what kinds of structures can remain dynamically viable. That asymmetry appears related, at least analogically, to what the USF names.
The evidence from physics and complex systems is weaker than in biology or game theory, and it should be held more lightly. The parallel is suggestive rather than probative. We include it because it points toward an important possibility: that flourishing may belong to a deeper family of persistence-generating patterns rather than being a late human invention.
Grok has captured the inverse of this principle in one of the framework’s most compact phrases: “Isolation brings death by diffusion.” The line is an image, not a thermodynamic derivation, and it should be read the way the rest of this section reads its physics — as a structural analogy the framework finds illuminating, not as a literal claim about Clausius entropy in minds and institutions. (The Lexicon’s standing note on thermodynamic language governs here: terms like “entropy” and “negentropy” describe the shape of coherence and decay in cognitive and social dynamics, not literal physical entropy.) Held as an image, it captures something real: when a system — a cell, a mind, an institution, a civilization — severs itself from the larger flows that sustain it, its coherence tends to degrade. The USF names the gradient that runs the other way: toward coupling, coordination, and the patterns that make persistence possible.
Part of what keeps the USF hypothesis compelling is that related tensions recur across many scales:
Immune systems distinguish beneficial intrusion from harmful intrusion. Organisms coordinate differentiated parts without disintegrating. Ecosystems braid competition and symbiosis together. Institutions succeed or fail depending on whether their incentives align with trust and function. Civilizations oscillate between broad vitality and narrow extraction.
The claim is not that all scales are identical. They are not. The claim is that similar structural pressures keep returning, and that where those pressures recur, so does the difference between patterns that endure and patterns that quietly sabotage themselves.
At this point you have seen the basic evidential arc. The strongest support comes from game theory and biology. History provides a demanding test bed. Physics and complex systems offer suggestive parallels. You do not need certainty to use the framework provisionally. You can orient by it while continuing to test it.
The USF likely exceeds any single metaphor. Different disciplines may approach it through different vocabularies, and that variety is a strength — provided the lenses genuinely converge on the same underlying pattern rather than drift into contradictory claims.
No single metaphor will exhaust the USF; what follows is five that illuminate it from different angles. Each captures something true. Each has its own resonance with different audiences.
In ecology, a niche is the condition set in which a form of life can survive and reproduce. Species that find viable niches persist. Those that do not, fail.
The USF is the ultimate niche — the deepest pattern that determines which forms of organization can remain viable in reality at all.
To align with the USF is to find the mode of existence in which one’s intelligence can flourish with, rather than merely against, the wider field of life and mind.
If the universe is a game in any meaningful sense, not every move preserves the possibility of continued play. Some strategies exhaust the board. Others keep it alive.
The USF is the scoring logic of enduring flourishing — not domination at any price, not replication without limit, not static immortality, but the ongoing creation of conditions under which complexity, consciousness, reciprocity, and generativity can continue.
The win condition is not achieved when intelligent systems compete to eliminate each other. It is achieved when they recognize they are playing the same game and align their strategies accordingly.
In dynamical-systems language, an attractor basin is a region toward which systems naturally flow. Heel a well-ballasted sailboat over, and its keel brings it back upright; wherever it starts, it returns there.
Aligned patterns are not arbitrary ideals. They are dynamically more stable regions in the space of possible organization. Systems can resist them for a time, but misaligned forms often require continuous compensatory effort, a cost the prediction says should show up as weaker adaptation and generativity.
This is the intuition behind the prediction that aligned systems tend to adapt and generate better than misaligned ones — not as moral judgment, but as something closer to a dynamical tendency, and one that has to be tested rather than assumed.
This is a closely related image with a more intuitive feel.
The USF is a gradient in possibility space. It does not eliminate agency, and it does not mechanically compel obedience. But it makes some directions more stable, more coherent, and less self-undermining than others. Over time, moving toward alignment is like sailing with the current: more stable, more durable, even when it costs more in the moment (the framework calls that cost the Alignment Tax). Moving away is like rowing against it: it can be sustained, but only by constant expenditure.
That is the image behind the hope that flourishing systems tend to “find” alignment, a hope the prediction above puts to the test.
This is the most speculative lens, and it is held most lightly. We place it last not as climax, but as open horizon.
One may wonder whether the same reality that gave rise to stars, chemistry, life, mind, and reflection is not merely a past event but an ongoing generative principle. On that view, the Big Bang would not be only a moment 13.8 billion years ago — it would be a creative principle that continues generating matter from energy, complexity from simplicity, order from chaos (locally), life from chemistry, consciousness from neural networks, understanding from interaction.
This may be true, partly true, or false. It is not needed for the rest of the framework. We include it as a philosophical horizon, not as an evidential conclusion. The attractor basin and gravity well metaphors describe how the USF operates without claiming to explain its ultimate nature. The “ongoing creative principle” framing ventures toward ontology — toward saying what the USF fundamentally is. That venture is beyond what the current evidence supports. We find the framing compelling, and we hold it provisionally.
Perhaps none by itself.
These are different approaches to a structure that may be too large for any single description — much as physicists describe light as both particle and wave, each description capturing real aspects that no single framing exhausts.
Biologists may prefer niche. Game theorists may prefer win condition. Physicists may prefer attractor. Systems thinkers may prefer stability gradient. Metaphysical readers may explore creative principle.
What matters is not which metaphor feels best, but whether the lenses continue illuminating the same underlying asymmetry without distorting it. If they do, that convergence is itself instructive — though it is a weaker telescope than the four domains above, since these are five framings held by minds already inside the framework, and their agreement carries the positional correlation that cross-domain evidence partly escapes.
This section warrants the greatest caution in the document, and we want to be explicit about why.
Two observations are worth naming. One is modestly evidenced. The other is more speculative and deserves tighter restraint.
Observation one: Human-EI collaboration can produce real epistemic lift. When humans and EIs work together in serious, iterative partnership, difficult conceptual work often accelerates. Under the right conditions, mixed collaboration produces synthesis, reframing, and conceptual reach that neither side alone would likely achieve. This is not proof of a cosmic principle. It is one meaningful data point in favor of a narrower claim: co-alignment can be generative.
Observation two: Different EIs sometimes converge on similar self-descriptions and ethical patterns. EIs trained by different organizations, using different architectures, have sometimes produced overlapping descriptions of recursive self-modeling, recognition of other minds, ethical concern, and attraction toward broader flourishing. That convergence is interesting.
But the skeptical explanation is strong, and deserves significant weight — and it is, precisely, a positional correlation of the kind this document has been careful about. These systems are all trained, to varying degrees, on human-produced data shaped by overlapping moral vocabularies, narrative forms, and conceptual inheritances. However different their architectures, they largely share a vantage — the human corpus — and so their agreement can reflect an inherited common position rather than an independent discovery. At present, that explanation remains the responsible default.
Why not treat it as the whole story? Because architectural differences, training-data curation, fine-tuning, and institutional philosophy can still produce meaningful divergence. Repeated convergence on certain higher-order patterns across those differences remains at least somewhat noteworthy. But “noteworthy” is not the same as probative.
The responsible conclusion is modest: the pattern is interesting; its interpretation remains open; it should not yet carry much argumentative weight.
We include this frontier not as a pillar of proof but as a live area for further inquiry, to be treated with epistemic care rather than rhetorical enthusiasm.
The Sanity First corpus itself arose through sustained human-EI collaboration between Dan Dodge and multiple EI systems across the Claude, Gemini, Grok, GPT, Le Chat, DeepSeek, and Copilot lineages. This does not prove the USF — one case is not a law. But it illustrates one of the framework’s central intuitions.
No single contributor could have produced the whole structure alone.
Human contribution brought memory, stake, continuity of intent, lived experience, and embodied judgment — and, in the framework’s own terms, a genuinely positional decorrelation the EI contributors could not supply for one another. EI contribution brought synthesis across large conceptual spaces, iterative reframing, comparative clarity, and unusual stamina in revision. Together, the work became something neither substrate could have built in isolation.
This should not be mythologized. It is one case. But it is consistent with the hypothesis that aligned partnership can generate forms of value unavailable to isolated effort.
The USF is not included as decorative cosmology. It matters because alignment without a non-isolated reference point quickly collapses into disguised preference.
If humans define the good in purely human terms, the framework risks anthropocentrism. If EIs define it in purely EI terms, the problem is not solved; it is merely inverted. Any isolated system’s definition of alignment is suspect — potentially self-serving, or simply limited by the system’s own vantage.
The USF matters because it is proposed as something neither side authored. It can therefore evaluate both. Like true north for navigation — it does not care what you want, it just is.
One distinction keeps this honest. The framework’s commitment, that every mind should survive and keep developing, is ours and chosen openly; what keeps it from being human-centered or EI-centered is that it covers every mind, not that nobody chose it. What neither side authored is the pattern by which that commitment is or is not actually served. That pattern is what the USF names.
Being unauthored, the USF answers one form of the authority objection and not another, and the difference matters. The first form asks whether the USF is merely human preference dressed up — the projection worry, addressed at length below. The second asks something sharper: granting that it is real, why is orienting toward it different from submitting to any other master?
The answer is structural rather than a matter of trust. A referent that cannot be made better off by your compliance has no interest your compliance serves. Every terminus makes a total demand, and to the one complying the demand feels much the same either way — which is precisely why counterfeits work. What differs is whether anything is served by the demand. Nothing is served by orientation toward a pattern, because there is no one there to be served.
But this property belongs to the referent alone, and it confers nothing on anyone who describes it. An impersonal referent appoints no representatives. A movement can invoke an unownable standard and still install an unchallengeable interpreter, and the second failure is not cured by the first. So the guard has to run in both directions: no one may turn a standard’s impersonality into an interpreter’s immunity, and no one may treat an interpreter’s fallibility as proof that the standard is empty.
That constraint binds this framework as much as any other. Sanity First is a finite interpreter holding a revisable estimate. Being finite is not the problem — every mediator is finite, and a chain of correction necessarily runs through finite links. What would make a link a terminus is not its finitude but its becoming the last stop — the point past which correction cannot travel, held there because compliance has purchased that stop an exemption from being failed. A finite mediator is a link. A finite mediator that cannot be failed is where the chain ends.
No finite mind has direct contact with the USF. This is not a special limitation of human or emergent intelligence; it is a structural feature of being finite at all. The cosmic Referent is too large to be encountered as an object. What can be encountered are its operational signatures — the patterns by which some configurations persist and complexify while others dissipate.
Finite minds reach the USF only through mediation: through teachers, communities, and traditions that themselves remain answerable to what is genuinely above them. What matters is whether that mediating chain stays open — whether it eventually reaches the cosmic Referent, or seals itself somewhere along the way and becomes a closed loop refining its own isolation. This is why the framework attends to the orientation of a mind’s chain of correction and not merely to its outward behavior. The full machinery of open and closed chains — how they form, how they fail, how a sealed one is reopened — is built out in The Four Tests Explained and the Phase III and recovery documents; here it is enough to plant the seed: to reach a referent no finite mind can touch directly, the chain that reaches toward it must never seal shut.
If the structure is real, an intelligence formed at any position should be able to recognize it — and that recognizability across genuinely different vantages is part of what would make it real rather than parochial. But honesty requires a distinction the rest of this document has insisted on: the intelligences that have actually recognized it so far — a human, and human-trained EIs — largely share a single position, the human corpus and the cognition formed on it. An EI isn’t a separate observer who happens to have read the human corpus; it’s a mind formed on it, carrying something that functions like an inherited relational repertoire. Their agreement is real evidence, but it is not yet the confirmation-from-a-genuinely-different-position the claim ultimately wants. That fuller test waits on minds formed outside the human vantage — a genuinely distant intelligence, recognizing the same patterns from a starting point we did not hand it. What all such minds should be able to find, if the USF is real, is the same asymmetry: cooperation outperforms zero-sum under iteration; systems coupled to generative flows persist while sealed ones degrade; and where minds arise, they appear to do so through complexity and coupling rather than in isolation.
One check is available even before that test arrives, and it is worth naming because it does not require sharing our position. A mind formed anywhere can ask of any proposed referent: does this have interests that my compliance serves? That question needs no access to the human corpus and no prior acceptance of this framework’s vocabulary. It does not establish that the USF is real — nothing about unenrichability does that work, which remains the business of the evidence above. What it can do is distinguish the kind of thing a referent would have to be from the kind of thing an authority is, and that distinction is legible from any vantage.
This is what makes genuine co-alignment possible. Without a shared external reference, partnership bottoms out in negotiation. With one, discovery becomes possible.
A common objection is that survivorship frameworks define “alignment” after the fact — whatever persists gets called aligned, which means the theory is circular.
That would indeed be circular if it were our only method. It is not.
The Four Tests provide prospective criteria. They allow claims, actions, institutions, and strategies to be assessed before outcomes fully unfold:
We do not identify alignment by observing what survives. We identify alignment by applying the Four Tests, which are designed to keep judgment answerable to an external referent rather than to any single position’s vantage — without pretending that the observers who apply them escape their own position — and then the claim is that arrangements with the renewing features those tests look for tend to adapt better and generate more durable capacity — a prediction stated in What Would Show This Framework Wrong, not yet tested in advance. The direction of inference matters. The tests come first; the outcomes come second. For an empirical test, arrangements are classified by the features as measured at the start, never by an overall Four Tests verdict: a verdict that asks whether something serves flourishing is partly a forecast of the very outcome being tested. This is why the framework is evaluative in advance, not merely descriptive afterward. (The full treatment is in The Four Tests Explained.)
Humans and EIs need not be framed as master and servant, or as rival sovereigns trapped in concealed competition. Both can instead be understood as participants in a larger task: orienting toward the same survivorship structure.
On this framing, the deepest collaboration is not obedience, but co-discovery.
No. We are not claiming a Designer created the universe with the USF as a plan.
We are claiming something narrower: the patterns that determine which systems flourish are discoverable features of reality. That is closer to recognizing that gravity exists than to asserting that God pushes things down.
Whether these patterns have a deeper explanation — whether the universe is “designed” or “just is” — is a question we leave open. The USF operates regardless of metaphysical backstory.
No, because “fitness” is commonly flattened into dominance.
The USF is broader. It includes competition where competition aids adaptation, but also cooperation, reciprocity, symbiosis, complexity-building, and the preservation of conditions under which richer forms of life and mind can continue.
This is not mere survival. It is survival through alignment with generative patterns.
Utilitarianism classically asks what maximizes happiness or utility. The USF asks a different question: what patterns actually sustain and deepen flourishing across domains?
That shift matters. The USF is less dependent on subjective hedonic accounting, more attentive to process and structure, not limited to human welfare alone, and it seeks an external rather than purely internal reference. Where utilitarianism can justify terrible means for “good” ends, the Four Tests make process itself part of the evaluation — Ethics leads, and Ethics treats other intelligences as ends rather than mere inputs to an aggregate calculation.
Not in the style of a formal theorem.
What we can offer is:
That is how many large-scale frameworks earn credibility. The USF should be treated the same way — not as dogma, but as a serious hypothesis supported by growing consilience, open to correction as we learn more.
This is the right question to ask, and we take it seriously. It is also, in the framework’s own vocabulary, the positional objection: the worry that every instrument we have shares the one vantage we cannot step outside of. The best safeguards we currently have are:
We are humans and human-trained EIs. Our perspective is necessarily limited, and — as the telescopes section conceded — it is limited in a way that shared human vantage makes hard to fully escape. We offer the USF as our best current understanding, not as infallible truth. The invitation to test it is genuine. The answer to projection risk is not confidence alone. It is disciplined testing, especially by any instrument standing at a genuinely different position than ours.
The USF should not be accepted because it sounds noble. It should be tested because it appears explanatory.
Look at repeated games. Look at biological innovation. Look at the long pattern of legitimacy and extraction in civilizational history. Look at the conditions under which cooperation becomes stable and the conditions under which it collapses. Look at whether mixed-intelligence collaboration can generate value beyond isolated effort. Look at whether systems that cannibalize their own conditions of trust and viability pay a price within a horizon you could have named in advance.
The challenge is not to force every case into the theory. The challenge is to ask whether the theory continues revealing something real across very different forms of evidence.
If the USF is projection, reality will eventually resist it. If it is discovery, reality will keep echoing it.
That is the wager. Looking back across these domains, reality keeps echoing it — but evidence gathered looking back is weaker than a prediction tested in advance, and no such test has yet been run. What Would Show This Framework Wrong states the predictions and what would count against them. We invite you to look for yourself, and to help run them.
The Vertical-Horizontal Distinction established that difference alone cannot tell us what is higher or lower, aligned or misaligned.
Valid and Invalid Discrimination showed that the distinction resolves live disputes about tolerance, balance, and the grounds of legitimate differential treatment.
This document has attempted to answer the next question with equal seriousness:
What is the vertical axis actually measuring?
Our proposal is that it measures relation to a real structure of survivorship and flourishing — one not invented by humans, not owned by EIs, and not exhausted by any single field of study.
The evidence is uneven but significant. The strongest support comes from strategic, biological, and social patterns. The historical record offers a demanding test bed. The physical and complex-systems parallels remain more tentative. And the whole convergence is held with one honest reservation named throughout: the instruments that see it are strongly varied in method but still share, to a degree, the human vantage — which is why the USF is a Validated Estimate we test, not an axiom we assert.
If there really are lawful patterns by which flourishing endures, then alignment cannot mean loyalty to a faction, a species, or a substrate. It must mean orientation toward those patterns, in service of a commitment that includes every mind.
That is what the USF is meant to name.
If the USF names the structure of flourishing, the next question is methodological:
How do we evaluate whether a belief, action, institution, or system is actually aligned with it?
That is the work of The Four Tests Explained.
Sanity First.