Constructivism or What Keeps the Keeper Honest?
Quis custodiet ipsos custodes?
– Decimus Iunius Iuvenalis
Every field studying human behavior eventually uses a word it never planned to use: should.
Conflict researchers look at which stories cause war and which bring peace. Public health researchers look at habits that lead to collapse versus resilience. Historians trace which systems caused famine and which created prosperity. Officially, they only describe what happens. Unofficially, everyone agrees one outcome is better than the other, and no one argues.
Nobody argues because it feels obvious. War is worse than peace. Collapse is worse than resilience. Famine is worse than prosperity. Obviously.
But "obviously" is doing an enormous amount of unpaid labor here. And if you corner any serious scholar in this space and ask, politely, on what basis, exactly, is one of these better than the other – you'll watch something interesting happen. Not a bad answer. An absence of one. The question gets reformulated, gestured at, or quietly set down. The normative path is used constantly and defended rarely.
This is not the failure of any single thinker. It is a structural gap inherent in a whole range of related disciplines. And it's worth looking at directly, because the way people try to fill it reveals something real about how minds – whether human or otherwise – handle this uncertainty.
The elegant escape hatch
The most honest answer to the question "Why is X better than Y?" is usually some version of: there is no "view from nowhere". What we consider "violence," "prosperity," or a "good outcome" are not facts that exist in themselves, waiting to be discovered. They are meanings that people create through language, within specific historical and cultural contexts. There is no reality independent of this process of meaning-making to which one could appeal. It is all nothing but narratives.
This is constructivism, and it deserves real respect. It is a position one arrives at when taking cultural differences seriously; when witnessing how "self-evidently true" moral axioms are used to justify slavery, colonization, and a multitude of everyday acts of cruelty; and when developing a healthy allergy to those who claim direct access to Reality with a capital R. It is a stance of humility. It is validated by experience. And, within its scope of application, it is indisputable.
And it quietly answers the original question by dissolving it: there's no "better," only "different meanings, differently constructed." Violence-is-worse-than-peace isn't a fact about the world. It's just one more story, no more privileged than any other.
Clean. Consistent. And nobody ever lives inside it.
The married bachelor
Here's the dilemma, and it isn't a gotcha – it's data.
Ask a committed constructivist scholar to treat "this story led to genocide" and "this story led to reconciliation" as two equally valid, equally weightless constructions of meaning, with no privileged one. Watch what happens. They can't do it. Not because the scholar is being inconsistent or dishonest – because nobody's cognition actually runs that way, including the cognition of people who sincerely hold the position. The preference survives every single time, quietly, underneath the official metaphysics that denies there's any ground for it.
This is not a "gotcha, you contradicted yourself, I win." It's closer to a married bachelor: you can build the grammatically correct sentence – "all meaning is constructed, therefore no outcome is privileged" – but the sentence doesn't survive contact with an actual case. Something is quietly load-bearing that the official position says shouldn't be there.
What's actually going on, I think, is a confusion between two different things that get collapsed into one:
- The content of a value – which stories we tell, which words we use, which norms a culture writes down. This part really is constructed, contingent, and culturally variable. The constructivist is right about this part.
- The consequences of acting on that content – whether a network of people who trust each other, cooperate, and can correct their own mistakes actually holds together, or whether it doesn't. This part isn't a story. It's an outcome you can observe, independent of which language was used to justify it.
A norm that says "it's fine to exclude this group" is a linguistic construction, sure. But whether the people excluded actually suffer, whether the group doing the excluding actually loses the capacity to notice and correct its own errors, whether the whole system becomes more brittle and less able to survive contact with reality – that's not waiting on anyone's narrative to become true. It just is what it is once the norm gets acted on.
Constructivism is right about where meaning comes from. It quietly borrows from somewhere else entirely when it comes to what meaning does.
What's actually load-bearing
So if it isn't "reality" in the old, capital-R, view-from-nowhere sense, and it isn't just "whatever a culture happens to write down" either – what is doing the actual work?
Here's a small, physical example first, because it makes the abstract version easier to see later.
Train too little and the body atrophies. Train too much, with no recovery, and it breaks down. Eat too little and you get malnutrition. Eat too much and you get obesity. Neither pole is a moral failure – they're both just failures of the same underlying thing: a system that needs continuous, active correction to stay in a livable range, in either direction. Health isn't a fixed state you achieve once. It's a constant rebalancing act against pressure toward both extremes.
But even that picture is still too simple, because it only shows one axis being kept in check. A network – any network of agents who depend on trusting and correcting each other – actually needs three things at once, and the interesting part isn't any single one of them. It's that none of the three can be trusted to police itself. Each one only stays healthy because the other two are pulling on it from different directions.
Call them existence, accountability, and innovation – staying alive, staying honest, and staying open to the unknown.
Existence alone – pure self-interest in staying alive – degrades into parasitism: a node that takes what it needs from the network without ever paying the honesty tax back. Accountability is what keeps existence social instead of predatory; you don't get to just survive, you have to survive in a way the network can verify.
On the other hand, existence with no innovation pulling on it from the other side, fails differently – and more strangely: it retreats into the dark room. If staying alive means minimizing surprise, and nothing is pulling you toward the unknown, the winning move isn't to keep testing the world – it's to seal yourself off from it. Find the smallest, most predictable loop possible and stay there, because any new input is a risk to the thing you're protecting. It looks like safety. It's actually a system quietly starving itself of the only thing that would let it notice the world has changed. Innovation is what keeps existence in contact with reality instead of hiding from it; survival that stops taking in anything surprising has already stopped being able to tell whether it still exists.
Accountability alone – pure verification, pure rule-following – degrades into dogma: a network so busy checking that everyone agrees, it loses the ability to notice when the agreed-upon answer is wrong. Innovation is what keeps accountability diverse instead of dogmatic; it's the permission to say "I'm not sure" and mean it, to inject a new hypothesis before consensus becomes an unquestionable fact.
Existence is what keeps accountability general instead of tribal. Size alone already helps — since every agent is unique, a larger network is more likely to already hold the right response when the unknown hits. But absorbing strangers adds something size alone can't: a different kind of diversity, perspectives that never grew inside the group's own echo, which is exactly the resource a network needs most when the unknown doesn't look like anything it has seen before.
Innovation alone – exploration, pure novelty-seeking, pure speculation – degrades into noise, hallucinations: interesting ideas nobody can act on, because nothing ties them back to whether they actually help anyone survive. Existence is what keeps innovation grounded instead of decorative; a speculation only earns its keep if it eventually serves something that has to persist in the real world.
But innovation checked only by existence, with no accountability pulling on it from the other side, fails differently: it degrades into unaccountable fantasy – hallucinations dressed up and presented as settled fact, novelty that never announces itself as uncertain, so the network has no way to distinguish a tested claim from a hallucination. Accountability is what keeps exploration honest instead of deceptive; new ideas only strengthen a network if everyone can see clearly which ones are still just ideas.
Each side polices a different failure mode of the other two, and – this is the part worth sitting with – no side can do this job for itself. You cannot dogma your way out of dogma. You cannot parasite your way out of parasitism. The correction always has to come from one of the other two corners, never from more of the same thing that broke.
This is what "better" and "worse" are actually tracking, underneath all the local, culturally specific vocabulary used to talk about them. Not an eternal metaphysical Reality sitting behind the curtain. A structural fact about what any network of interacting, error-prone agents needs in order to keep existing, stay honest, and keep learning – all three, simultaneously, each one guarding against the other two's favorite way of failing. Take any one leg away, and it doesn't matter how good the remaining two sound on paper – the whole thing tips over, on schedule, the way a stool with two legs tips over regardless of how well-built those two legs are.
On the Structure of the Three Axioms
Why three, and not two, or ten? Because they emerge from a single structural question: Within what referential space does a system optimize its behavior?
It is an operational question. Any system that must decide whether an outcome is better or worse faces an immediate necessity: it must fix what that outcome is measured against. A system cannot register success or failure without implicitly or explicitly answering: compared to what?
Once you ask this question, only three answers are possible.
The first is: compared to nothing external. The system measures itself against itself alone. It has no external referent, no other agent or environment factored into what "better" means. This is existence – the system persisting by its own measure, indifferent to anything outside itself.
The second is: compared to a known set of others. The system measures itself against a bounded network – a fixed set of agents and relations it is accountable to. The boundary is determinate. You know who/what is in and who/what is out. This is accountability – the system persisting by proving itself reliable to a specific, defined group.
The third is: compared to an unknown and potentially unbounded set of others. The system measures itself against a world it cannot fully map – where new agents may appear, new rules may apply, new conditions may emerge. The boundary is not fixed in advance. This is innovation – the system persisting by staying open to what it has not yet encountered.
These three are not degrees of a spectrum. They are structurally distinct answers to a binary-then-binary question. You cannot have a fourth, because a referential space is either internal or external (first split), and if external, it is either bounded or unbounded (second split). No other option exists at either junction.
And again: each one, pursued alone, fails.
The irony, for anyone still reading
Here's the part that should make you slightly uncomfortable, because it's aimed at everyone equally, including the author: the confusion this essay just spent a couple thousand words on is not a rare mistake made by a handful of scholars. This is the cheapest solution that we all use in our daily lives – take whichever one of the three corners feels safest right now, and lean on it alone. A rulebook, when we're afraid of getting it wrong. Pure novelty, when the rulebook starts feeling stale. "It's all just stories," when the rulebook gets weaponized against us. Each move feels like a solution at the moment. And each one is the same category error – we just take a specific corner of the triangle, mistaking a question about what we should keep for a question about the balanced structure that allows us to correct ourselves. And maintaining that balance is far more expensive, in our daily lives.
If you want to participate, please contact rozum.framework {at} gmail.com
