GETTIER – Why Plato Would Reject Justified True Belief (a Platonic dialogue)

DasGegenmittel

Hi everyone,

I’ve written a Platonic dialogue to highlight where Plato is fundamentally misunderstood when it comes to the widely accepted definition of knowledge.

GETTIER explores the following:

Socrates meets the philosopher Edmund Gettier to examine the classical definition of knowledge as “justified true belief.” Gettier’s objections are analyzed through Platonic concepts and questioned in terms of their philosophical scope.

At the core lies the question: Are contingent counterexamples sufficient to undermine the epistemic claim of this definition? The text argues that both Gettier and much of contemporary epistemology misread Plato through the lens of the analytic tradition.

The dialogue connects modern epistemological problems with Platonic ontology and asks in what sense knowledge must be tied to what truly is. The aim is to reinterpret the Gettier problem from an ontological perspective – as groundwork for the epistemological reorientation developed in my essay Justified True Crisis.

I hope the dialogue reads well – and perhaps even entertains a little.

Excerpt from the dialogue:

In Elysium,where the souls of the just walk beneath blooming olive trees and conversation never ceases, Socrates and Plato sat in the shade of a laurel tree. The air was serene, time had no urgency, and Logos hovered above all like a gentle light.

Plato (muttering restlessly): I can’t help myself, Socrates. Once again they speak in the upper world of knowledge as if it were a riddle for sophists. A certain Gettier claims, I’ve heard, that he has shaken our work—with a paper barely three pages long. If that suffices to topple an idea that has occupied us for a lifetime, then, oh Socrates, you may conduct the conversation. Good luck my friend! I know you’ll reveal... whatever it is one can “know” about such matters! (his voice fading as he walks away) And I wrote aporetic dialogues!
Socrates (smiling): Oh Plato, you always want truth to shine bright—but sometimes the path to the open air is slippery. Go, then. I will see whether this Gettier bears within him that unrest fit for philosophy.

Plato walks off into radiant Elysium. No sooner has he vanished than a stranger appears—squinting in the light, with a Western appearance and a probing gaze, as if he had a counterargument for everything.

Socrates: Welcome, friend. Your steps echo new upon this ground. I suppose you are that Gettier of whom many speak?
Edmund Lee Gettier III: That’s what they call me. Have you seen my family? Are you—Socrates?
Socrates (nodding): You will find them, when you are ready—and your friends, too. Tell me: shall I continue calling you by your surname? And yes, I am he.
Ed: Those who know me call me Ed—if that’s all right with you. Edmund feels too distant for dialogue. And yes, I come with a doubt. I'm always linked with a problem—even when praised, it’s a burden to constantly hear about “the Gettier problem.” (sighs)
Socrates (grinning): So be it, Ed. Doubt is a fine travel companion. I have heard of your problem—that idea that epistēmē, knowledge, is meta logou alēthēs doxa, or as some now say, “justified true belief”. That is a topic I will not ignore, even here, where some believe all matters have already been resolved.
Ed: I showed that one can believe something true, and even have good reasons for it—and yet we would hesitate to call it knowledge. (A low muttering is heard in the distance.)
Socrates (clearing his throat): Then you are either a wise man—or a disturber of the peace. For even with Theaetetus, we did not get much further. Perhaps Ed, your arrival is the next step in a long journey.

They walked a short way along the shimmering path until they came to a quiet place, where the view opened to the glassy waters of an eternal river. There, between white cypresses, Socrates and Ed sat on a marbleedged stone, untouched by time or weather. The sky above Elysium was clear and few birds could be seen— something began to stir in their dialogue, though the light remained sharp.

Socrates So then, Ed—you said that someone might hold a belief that is true, and even supported by reasons—yet we would still hesitate to call him knowledgeable. Tell me: what, in your view, is missing?
Ed (raises his hand to shield his eyes from the light): It seems something is missing that binds these conditions together—something that raises them to the level of knowledge: the justification, or what you call logos.
Socrates (nodding): An old word—often used, seldom understood. We examined it—in three forms, as I proposed them to Theaetetus. Would you like to revisit them with me? Plato is not fond of being misunderstood. Aporia is often as dear to him as genuine agreement is to me.
Ed (rubs his forehead): Socrates, I don’t remember— A mist of the Styx still clouds my memory.
Socrates (smiling): Then listen. What if logos meant only: the ability to express what one means? (He points upward toward the sky, to the birds above.) But I ask you: can a parrot, though it does not know, still speak words?
Socrates: Is speaking already knowing?
Ed: Hardly. A child often knows where its toy is—but couldn’t explain it. Language alone does not make knowledge.
Socrates: Well said. Then let us examine further. What if logos meant: breaking down a concept into its parts—like a craftsman dismantling a cart into “wheel” and “axle”? But I ask you: does one become an expert on carts simply by knowing that they are made of wheels?
Ed: No, certainly not. The whole is more than the sum of its parts. One can take everything apart—and still understand nothing.
Socrates: Wisely spoken. Then this remains: logos as that which sets a thing apart from all others—it particular way of being.
Socrates: But, Ed—Is there anyone who knows all the differences? Is that not a task for the gods?
Ed: So it seems. This form of logos doesn’t truly help us either—it demands more than we humans can deliver.
Socrates: Do you see now, Ed, how we have questioned the voices of speech, of analysis, of distinction—and yet knowledge still stands in the shadows?
Ed: We have shown paths—but not a foundation. None of the proposals has managed to capture the difference between mere opinion and true knowledge.
Socrates (with quiet severity): Then you repeat—without knowing it—what we learned with Theaetetus: that one can examine everything—and still end up empty-handed.
Ed (thoughtfully): I sought a secure definition—and found only uncertainty. Perhaps my detour was a misdirection?
Socrates: Or the beginning of philosophy. For only when one realizes that no part stands for the whole, and no concept binds what truly is, does one begin to seek the true path.
Ed: Then my critique of “justified true belief“ was not the downfall of knowledge—but a step toward a higher search?
Socrates (nodding slowly): A useful error. You showed that the circle closes where one thinks he walks a straight path—but you have not yet seen where the gaze must turn for truth itself to appear.
Ed (quietly): I tried to grasp it—and knowledge slipped through my fingers like water.
Socrates (gently): So it goes for all who mistake becoming for being. As long as you asked, What is knowledge?—and thought it a tool to hold in your hands—you remained trapped in error. But now, being empty, you may begin to see.
Ed: Then my refutation was not an end—but a gate?
Socrates: Perhaps the right gate. For, as I once said: “The confession of not knowing is the first step toward philosophy.”
Ed: And what is the second?
Socrates (gazing toward the sky of Elysium): The turning. Not toward definitions—but toward what truly is. Not toward what merely seems—but toward what always is. Are you ready?
Ed (softly): I am ready, Socrates.

[...]

This is a excerpt. The full dialogue is available here:
• https://philpapers.org/rec/SCHGQF
• https://www.researchgate.net/publication/391526622_GETTIER

This dialogue serves as a conceptual prelude to my essay Justified True Crisis, which builds on its ontological insights to propose a dualistic and dynamic definition of knowledge.

I’d be happy to hear your thoughts or discuss further.

Best regards

Count Timothy von Icarus

↪DasGegenmittel

It seems to me like you are getting at the role of understanding in knowledge, which has a phenomenological component. If truth just involved discursive justification and assigning the right truth values to linguistic utterances or symbolic strings, then LLMs would "know," right?

The crux seems to be here:

Socrates: Well spoken. For even though your example is a thought experiment, it arises from the realm
of contingency: it depends on concrete circumstances—who hires whom, how many coins someone
carries—all things that do not follow from the essence of the matter, but from chance. But what comes from
chance is not necessary—and what is not necessary cannot be the object of knowledge. Therefore, Ed: your
example remains in the domain of the opinionable, because it aims at something that neither always is nor
is so by its own nature—but only through external conditions. And in this very point, it becomes clear that
it can never touch epistēmē in the sense of being. And can it surprise us, then, that logos fails when it seeks
to grasp what is not stable?14

This is precisely what Aristotle would say re episteme. The example relies on a coincidence of accidents. Could it be reconstructed with per se predication? I don't think I've ever seen it done.

The difficulty of limiting knowledge to being is of course explaining discursive knowledge in the realm of becoming, which does seem to exist. This requires a robust metaphysics, a "metaphysics of knowledge," which is made difficult by the tendency of modern thought to put either epistemology (early modern) or philosophy of language (linguistic turn) before metaphysics.

I would guess we have pretty similar opinions here. We had a recent thread on this and my thoughts were:

I tend to see these sort of issues as indictive of the fact that "justified true belief" is simply a bad way to define knowledge. It's a definition that recommends itself by being analytically quite easy to work with; however this is a bit like the guy who lost his keys on the lawn and looks under the street light for them instead because "that's where he can see."

If knowledge involves the adequacy of the intellect to being, then simply affirming true propositions with proper discursive justification is not all there is to knowledge. Truth is primarily a property of the intellect, and only analogically predicated of linguistic utterances (as signs of truth in the intellect). When someone thinks p is true for bad reasons, and p is true, there is an adequacy of the intellect to being insomuch as truth is properly affirmed, but this will not involve the fuller adequacy that comes with understanding (which we would tend to call "knowledge.")

I think the empiricist tendencies in analytic thought tend to lead to a neglect of the role of understanding in knowledge. However, even if one dismisses any faculty of noesis/intellectus (which I wouldn't), I still think the phenomenology of knowledge suggests a big role for understanding (and this a relevant role for problems of vagueness). With vagueness, it seems we can have properly justified true belief and still lack "knowledge" in a strong sense. Knowledge is understanding and if "the truth is the whole," it is also in some sense inexhaustible. A "model" that tries to make truth primarily a binary property of propositions is going to miss this (and has other problems if truth/falsity represent contrary instead of contradictory opposition).

As you seem well aware, ancient conceptions of knowledge have an erotic element, one that involves an ecstasis, a going out to the known, and a penetration of the self by the known. There is an element of "knowing by becoming." One cannot capture this in an understanding based on "sense data + computational reason (ratio)."

I am not sure about the appeal to Plato's theory of mimesis at the end though. This has always struck me as quite a reach, and fairly implausible. Why not simply take up the later position of Plato's student that the mind is potentially all things, but that a move from potency to act always relies on act (actuality as form (eidos) delivered through the senses)? This also jives with information theoretic analyses of perception quite well.

When we get to the "metaphysics of knowledge" I don't even know if it is appropriate to call knowledge (or at least what is most fully knowledge) a "belief." When we are sure that there are cars in the oncoming traffic lane and that we mustn't drive into them, I think this is not simply a case of sense data + ratio (computational reason) = propositional belief. The reason we find it quite impossible to ignore such knowledge lies, IMHO, more in the co-identity of knower and known in such cases (a union). People find it impossible to believe otherwise because their intellect is "informed" by truth in the senses (sense knowledge), or what we might call the communication of actuality.

That's a fairly Aristotlian/Neoplatonic view, and less strictly Platonic of course.

DasGegenmittel

It seems to me like you are getting at the role of understanding in knowledge, which has a phenomenological component. If truth just involved discursive justification and assigning the right truth values to linguistic utterances or symbolic strings, then LLMs would "know," right? — Count Timothy von Icarus

Plato would argue that knowledge is intrinsically tied to “what truly is.” It entails a turning of the soul toward the unchanging and the realm of being, in contrast to mere opinion or belief, which pertains to the ever-shifting domain of becoming. The assignment of truth values to linguistic expressions or symbolic signs aligns more closely with the realm of opinion, as it deals with the contingent rather than the necessary. True knowledge, by contrast, is a form of “vision” or direct participation in being itself.

From this perspective, large language models (LLMs), which operate through discursive justification and the manipulation of linguistic or symbolic signs, do not possess the kind of knowledge that presupposes a deeper connection to the unchanging ground of being.

While I hold a partially divergent view, I contend that LLMs do exhibit a form of knowledge—albeit one fundamentally aligned with static knowledge (SK), rather than dynamic knowledge (DK). Their functioning is embedded within formal language systems, abstracted from lived temporality, embodiment, and situated world-orientation. Accordingly, they process and replicate patterns that remain invariant across contexts, without engaging in the evolving interplay of perception, revision, and context-sensitive responsiveness that characterizes DK. Their outputs are statistically grounded and internally coherent, yet they lack the epistemic historicity and adaptive flexibility necessary for navigating uncertainty or existential rupture. What LLMs produce is system-bound conceptual knowledge—structurally stable, context-relative, but ultimately indifferent to truth understood as existential commitment or pragmatic attunement. In this sense, their knowledge remains static: syntactically elaborate yet semantically shallow, operationally effective but epistemically inert when measured against DK’s criterion of justified transformation through time.

I’ll address your other points later. :)

Cheshire

Seems to me Gettier simply remarked that we defined knowledge and left out a temporal dimension. And noticed people are wrong a lot for how perfect knowledge ought to be by our standards. I wouldn't dispute we can know things, but taken as a body of work, knowledge has errors. What's the point of even noticing that belief plays a part if it can't later be disbelieved. If folks would stop confusing knowledge for what they think it ought to be things would run a lot smoother.

Put another way, how long do I have to believe it for it to be knowledge? Or not.

Count Timothy von Icarus

↪DasGegenmittel

Just to make sure I am understanding you, are you saying LLMs "produce knowledge" for us, or that something like ChatGPT actually "knows" things? Or is this a case of analogy?

I would say LLMs contain knowledge in the same way books do. Indeed, they mostly just slam together text predictively from books and similar documents. I suppose it could be argued that they can also produce novel knowledge through synthesis, but this would still seem to me more like a "book that writes itself," than a "knower."

DasGegenmittel

@Count Timothy von Icarus This is a particularly fascinating and important topic because it challenges us.

As you pointed out, both objects and entities like humans and large language models (LLMs) can contain knowledge. But before going further, it’s essential to acknowledge that there are different types of knowledge. For instance, there’s mathematical and logical knowledge, which involves necessarily true conclusions; conceptual knowledge, such as definitions (e.g., a bachelor is an unmarried man); systemic knowledge tied to structured domains like football (e.g., 11 players per team, one ball); and then there are probabilistic truths—statements likely to be true—as well as claims, falsehoods, and misinformation, which still count as a form of negative knowledge.

Books can store and present this kind of knowledge. They act as a medium—but remain passive. Nothing happens with the knowledge unless someone engages with it.

In contrast, AIs function differently. They resemble a kind of agent: acting under goals and within certain constraints. Not only do they possess information, but they also process it. Depending on their architecture, they may even have an operational awareness of the content they hold—because they verify, reference, and integrate it. This form of “awareness” should not be confused with human consciousness. Rather, it’s procedural—more about internal orientation than about being situated in the external world. One might compare it to navigating conceptual or mathematical landscapes.

In Justified True Crisis, I refer to this as DKorg—a kind of self-reflective knowledge that organizes and reorients itself over time and can refer back to its own prior states. From this perspective, consciousness is not a static property but a dynamic understanding of one’s own cognitive states across time, along with the ability to relate to them.

Of course, there remains a massive gulf between humans and machines. Humans use knowledge to navigate the world. Knowledge can be an end in itself, but more often, it serves as a foundation for action, identity, and emotion. It shapes how we see ourselves (“Who am I?”), distinguishes us from others, and prompts emotional reactions to new information—whether in stories, tragedies, or revelations. AIs, on the other hand, are not currently required to do any of this. But in the long term, such developments might be possible—even if they appear foreign or artificial to us.

What seems crucial here is recognizing that our traditional binary view of content versus agent is being challenged—especially by AI. Rather than a dichotomy, I now see it more as a multidimensional spectrum. And perhaps the fundamental questions to explore are:
1. Can a system organize itself and its information—potentially through hierarchical, evaluative processes?
2. Does it have a sense of its own cognitive states over time?
3. How can it respond to that awareness—by learning, adapting, resisting, or even displaying uncertainty?

DasGegenmittel

↪Cheshire

You’re right to sense that Gettier highlights a missing temporal dimension in the traditional definition of knowledge. But that insight goes deeper than it first appears.

In "Justified True Crisis" I argue that in dynamic contexts, knowledge can’t be reduced to static Justified True Belief (JTB). Instead, it must adapt to changing conditions, uncertainty, and coincidence — this is what Dynamic Knowledge (DK) addresses. Imagine you believe Route A is the fastest way to work. You’ve taken it many times, it’s usually reliable, and today it gets you there quickly. But tomorrow, a construction site appears, traffic backs up, and Route B would have been faster. Your belief was reasonable — but only under yesterday’s conditions. In dynamic situations like this, what counts as “the fastest way” isn’t fixed. It depends on changing variables: traffic, roadwork, even your destination. What was right once can quickly become outdated — and that’s why knowledge in practice needs to adapt.

Knowledge, then, isn’t just “believing what is true,” but “believing what must be continuously justified as truths shift.” Belief becomes a crisis — a conscious leap under incomplete information, acknowledging that what seems true may just be accidentally right.

So your question, “How long do I have to believe it for it to be knowledge?” hits the core issue: knowledge isn’t a fixed state, but an evolving process — and it needs a framework like DK to be meaningfully understood in real-world conditions.

DasGegenmittel

This is precisely what Aristotle would say re episteme. The example relies on a coincidence of accidents. Could it be reconstructed with per se predication? I don't think I've ever seen it done.

The difficulty of limiting knowledge to being is of course explaining discursive knowledge in the realm of becoming, which does seem to exist. This requires a robust metaphysics, a "metaphysics of knowledge," which is made difficult by the tendency of modern thought to put either epistemology (early modern) or philosophy of language (linguistic turn) before metaphysics.

I would guess we have pretty similar opinions here. We had a recent thread on this and my thoughts were:

[... alllll the Text]

When we get to the "metaphysics of knowledge" I don't even know if it is appropriate to call knowledge (or at least what is most fully knowledge) a "belief." When we are sure that there are cars in the oncoming traffic lane and that we mustn't drive into them, I think this is not simply a case of sense data + ratio (computational reason) = propositional belief. The reason we find it quite impossible to ignore such knowledge lies, IMHO, more in the co-identity of knower and known in such cases (a union). People find it impossible to believe otherwise because their intellect is "informed" by truth in the senses (sense knowledge), or what we might call the communication of actuality.

That's a fairly Aristotlian/Neoplatonic view, and less strictly Platonic of course. — Count Timothy von Icarus

The epistemological challenge lies in not confining knowledge solely to being, for becoming gives rise to forms of dynamic knowledge—discursive, contextual, and undeniably real. To grasp these forms adequately, we need a robust metaphysics of knowledge. And yet, as you rightly pointed out, this need has been structurally sidelined by the rise of early modern epistemology and the linguistic turn in philosophy.

I see metaphysics and epistemology as fundamentally interdependent—like heart and lungs: sever one from the other, and both collapse. This becomes especially evident in Gettier cases, where neglecting the metaphysical referents of belief leads to a breakdown of epistemological structure.

The Rashomon effect offers a telling example. It doesn’t just expose the limits of the justified true belief (JTB) model—it reveals that knowledge is socially organized, relationally mediated, and constantly at risk within context. Epistemic fragility is not the exception; it is the rule. From this, it follows: knowledge is not merely static assent to propositions, but a dynamic becoming—an ontological participation in reality.

Aristotle’s concept of epistēmē strikes at the core: knowledge is not produced by representation but by the intellect’s becoming form. “The intellect becomes all things”—not by mirroring, but by participation. (Here, Popper’s metaphor of the net is highly relevant.)

Plato’s notion of mimesis must not be misunderstood as mere copying. Rather, as layered participation—mediated through logos, perception, and memory—it aims at resonance with the essential. Even so, Aristotle’s account of the transition from potentiality to actuality, particularly as developed by the Neoplatonists, provides a more stable framework for embodied, enacted knowing.

In this light, knowledge is not merely the possession of a map (static), but the act of walking the terrain (dynamic)—not a reflection of form, but its unfolding in motion. These ways of knowing are not opposed; they are complementary—necessary companions for the journey.

Cheshire

Imagine you believe Route A is the fastest way to work. You’ve taken it many times, it’s usually reliable, and today it gets you there quickly. But tomorrow, a construction site appears, traffic backs up, and Route B would have been faster. Your belief was reasonable — but only under yesterday’s conditions. — DasGegenmittel

I'm not sure anyone's really arguing that knowledge implies certainty of traffic conditions. You know which way has been statisically faster I imagine. Perpetual empiricism™ feels aspirational.

RogueAI

↪Count Timothy von Icarus

I think of the LLM's like librarians in the ~~Tower~~ Library of Babel. They don't write the books, but they help you zero in on the book that has the answers you're looking for.

DasGegenmittel

↪Cheshire

The problem with knowledge is that it is used in a paradoxical way; we need a knowledge dualism. On the one hand, we say “I know” in cases where we can be certain that something is true, such as in mathematics. On the other hand, we use the same expression in situations involving contingency, like traffic routes. The underlying distinction is that some domains, like mathematics, involve elements that do not change—numbers, for example. The number 1 is always 1, without exception.

However, there are things like “the fastest way to work” that can change, and this is perfectly reasonable. When we say something like “I know the fastest way to work,” we implicitly mean “given the information and circumstances I am currently aware of.” If we define knowledge without considering the circumstances and the specific point in time, the definition becomes problematic. Yet this is exactly what epistemology tends to do today, which is, in my view, why it struggles with Gettier cases. This also offers a way to understand the Ship of Theseus problem: broadly speaking, Gettier cases are the epistemological counterpart to the ontological identity paradox. Both involve questions of change and time—identity over time versus knowledge over time.

In Platonic terms, some things are timeless and unchanging, while others are not. For Plato, there was no knowledge of the real, physical world—only of the eternal world of ideas, according to his special ontology. In my opinion, what Plato missed—and what we intuitively critique—is the conceptual dimension of knowledge.

Thus, we can even claim to have knowledge of things like current traffic conditions, which may be the best available to us so far, even if they don’t necessarily reflect reality. Since the world can change without our awareness, there can be no definitive knowledge about the future—only conceptual knowledge.

One might object: “But we constantly experience that we do have knowledge!” However, this is a mistaken intuition, as illustrated by Gettier cases. The problem is what Popper also pointed out in the sciences: verification is a poor guide and a pathway to pseudo-science—or in our case, pseudo-knowledge. In dynamic environments, we can only corroborate, not verify.

Cheshire

The problem with knowledge is that it is used in a paradoxical way; we need a knowledge dualism. On the one hand, we say “I know” in cases where we can be certain that something is true, such as in mathematics. On the other hand, we use the same expression in situations involving contingency, like traffic routes. The underlying distinction is that some domains, like mathematics, involve elements that do not change—numbers, for example. The number 1 is always 1, without exception. — DasGegenmittel

I see it as a problem in defining knowledge in a generalized way. Nothing paradoxical going on in the rational application of what is being said. One is a self-consistent model and the other a reference to the novelty of naming conventions.

Thus, we can even claim to have knowledge of things like current traffic conditions, which may be the best available to us so far, even if they don’t necessarily reflect reality. Since the world can change without our awareness, there can be no definitive knowledge about the future—only conceptual knowledge. — DasGegenmittel

Seems like a bit of a strawfolk argument to put foward that induction is an artifact of intuition alone. When assembling a puzzle and there's only one space remaining I can safely predict where the final piece is going to go?

One might object: “But we constantly experience that we do have knowledge!” However, this is a mistaken intuition, as illustrated by Gettier cases. The problem is what Popper also pointed out in the sciences: verification is a poor guide and a pathway to pseudo-science—or in our case, pseudo-knowledge. In dynamic environments, we can only corroborate, not verify. — DasGegenmittel

I disagree, I don't think Gettier breaks JTB once you qualify belief as having to persist. Which falls in line with Popper's recommendation of an attitude that treats knowledge as provisional. Again, knowledge in the context of dynamics implies a probablistic statement. Being willing to confuse it with a static declaration ignores the context for the sake of cobbling an argument.

Take Popper's gold standard for a scientific theory, general relativity, well when we do find a unified theory it will be tested by corroborating GR, not dismantling and rejecting it.

DasGegenmittel

@Cheshire thanks for the response — and for raising the strawman concern. Let me clarify and build on what you’ve said.

That’s why I refer to the puzzle metaphor myself — but in the way it’s developed in the essay: real-world knowledge, especially in dynamic environments (as in Gettier cases), resembles completing a puzzle with pieces from different sets. A piece may appear to fit structurally, and the picture might seem momentarily coherent — but it ultimately belongs to another image. The danger is that we stop questioning once the piece “fits,” even though the match was coincidental. The point isn’t to abandon inductive reasoning, but to recognize that in such situations, the fit alone doesn’t guarantee conceptual integrity. What looks like knowledge may only be a coincidental alignment — a Gettier-style conceptual coincidence rather than a robust connection between justification and truth.

The problem isn’t that belief fails to endure — it’s that the justification isn’t connected to the truth-maker in a reliable way. The broken clock case isn’t a fluke: it’s a stand-in for many real-world scenarios where we reason well, but for the wrong reasons — and get it right only by chance.

That applies even in scientific contexts. We might persistently believe in a well-tested model (like GR), and be right — but later discover that our reasons were incomplete or that better evidence recontextualizes our belief. This doesn’t mean GR was useless, but it highlights how fragile even strongly corroborated beliefs can be when new conditions emerge.

I fully agree with your reading of Popper: knowledge grows through conjecture and refutation, not final proof. But that’s exactly why pseudo-knowledge is a danger — especially in systems where new variables can arise without being noticed, and where verification is still socially or institutionally treated as conclusive. Financial crises, brittle AI models, or medical recommendations later overturned are real-world versions of Gettier: beliefs that seemed justified and true, until the structure shifted.

This is why I argue for epistemic dualism — not as semantics, but as a conceptual clarification. When we use the same word “knowledge” for both timeless, deductive certainty and context-sensitive, probabilistic belief, we risk flattening important distinctions. It’s not about discarding JTB, but about supplementing it with a model that tracks change, risk, and the potential for conceptual coincidence — a kind of epistemic humility built into the structure.

So in short: I don’t think we disagree on the value of provisional knowledge or the limits of verification. But I do think Gettier still shows something important — that being right isn’t always enough, especially when we don’t control or even perceive all the variables. That’s not a strawman — it’s a call to make our knowledge models better reflect the messy, shifting nature of the world we’re reasoning in.

And in this discussion — or thread — about the dialogue, I was originally trying to show that even Plato saw what the analytic tradition often overlooks: ontology matters, the thing in question matters – especially if it changes over time.

Cheshire

The problem isn’t that belief fails to endure — it’s that the justification isn’t connected to the truth-maker in a reliable way. The broken clock case isn’t a fluke: it’s a stand-in for many real-world scenarios where we reason well, but for the wrong reasons — and get it right only by chance. — DasGegenmittel

Well, then this is excellent news. If the broken clock is a stand-in for a real world scenario we could examine one of those? Because scientific justifcation isn't something in a wave function about an afternoon. Having measured things before I'm very aware of the difference between it is and it was measured to be. So, a moment of belief doesn't imply a persistent justified belief over some course of time. But, since it's a stand in I'm expecting the real world scenerios drawn from the pervasive use of accident-coincidental knowledge making events will demonstrate this error in the clearest possible way. Proceed. Or not.

DasGegenmittel

@Cheshire It’s not about undermining scientific measurement — it’s about illustrating how, even with methodical reasoning, truth and justification can drift apart in dynamic or information-limited settings.

Here are a few real-world cases that echo this:

- Medical misdiagnosis: A doctor makes a diagnosis based on symptoms that align with a common illness and prescribes treatment. The patient recovers — but later it’s discovered the symptoms were from a different, unrelated condition that would’ve resolved on its own. The outcome was “correct,” but the justification was misaligned with the actual cause — a textbook case of epistemic luck.

- Algorithmic bias: In machine learning, models often produce accurate predictions — but for spurious or hidden correlations in the data (e.g. ZIP code used as a proxy for race or income). The output may be right, but the reason why it’s right is structurally flawed. These systems perform well until the background conditions change, and then they collapse — just like the clock.

- Investing on false signals: A trader acts on a technical indicator that “predicts” market movement. It works that day — not because the indicator captured a real economic signal, but because of an unrelated geopolitical event that happened to push the market in the expected direction.

In each case, we’re not dealing with belief that simply “fails to endure.” We’re dealing with beliefs that appear justified, turn out to be true, but for the wrong reasons. And that’s precisely what Gettier — and the notion of conceptual coincidence — helps uncover.

Cheshire

— just like the clock — DasGegenmittel

This is my entire point. All of these examples fail to rise to something accepted into a functional system that would be called "knowledge". All of them dissolve under a short temporal arc. A diagnostic error is not medical knowledge, AI bias is in the context of how the computer believes complicates at best. Post rationalization of a stock trade isn't thought of as persistent justified belief.

I acknowledge they apply when we ignore the context of a 4th dimension. But, I can't see that insisting one exists to be creating a new conditon. Gettier found a hole in the formulation of the theory; not in the existence of knowledge as we find it. I maintain there isn't a documented case of incidental knowledge that we can rationally point to as an analog for the daily occurance of alt-knowledge type-B. Once, exteneded over a reasonably implied temporal dimension these Gettier cases disolve.

— just like the clock. — DasGegenmittel

Or not.

DasGegenmittel

This is my entire point. All of these examples fail to rise to something accepted into a functional system that would be called "knowledge". All of them dissolve under a short temporal arc. A diagnostic error is not medical knowledge, AI bias is in the context of how the computer believes complicates at best. Post rationalization of a stock trade isn't thought of as persistent justified belief. — Cheshire

No one has ever claimed that these cases constitute knowledge. The whole point of the Gettier problem is precisely that the definition is fulfilled, yet what we have still doesn’t count as knowledge in the way we expect it to. Ergo, JTB doesn’t work as a universal account of knowledge.

In dynamic, changing, or temporally bound environments, 100% certain knowledge is unattainable. The very nature of such contexts introduces variability that undermines the permanence and infallibility that traditional definitions of knowledge require.

What I’m getting at with this post – and the Gettier dialogue in general – is that Gettier misunderstood Plato. He ignores the ontology of the Forms, which are unchanging and timeless. As a result, he misapplies the JTB concept – which itself is more complex than the standard definition implies, as the dialogue shows – and applies it universally to everything: to the changeable and the unchangeable, to the temporal and the timeless. That this must fail is exactly what Gettier cases demonstrate. Our belief that it should work universally is mistaken. A related but secondary question is why and how we still talk about knowledge in uncertain scenarios.

All my essays and comments are meant to highlight the factors of adaptation and temporality. To put it plainly: current epistemology, as it deals with the Gettier problem, does not do this – which is why it continues to buzz around inside the fly bottle. It treats changeable and unchangeable, temporal and timeless, in the same way – and then wonders how to arrive at a 100% certain definition of knowledge. But that is impossible, because it’s trapped by its own self-imposed analytical limitations.

Anyone who incorporates temporality into the JTB definition—whether implicitly or explicitly—is necessarily moving toward a dynamic epistemology, which is indispensable for understanding knowledge in changing, context-dependent environments. In such a framework, Gettier cases do not simply disappear or get “solved”; rather, they emerge as structural indicators—features in the sense of epistemic phenomena that expose the instability of justification across time. They reveal how knowledge claims, even when seemingly justified and true at one moment, can lose their validity as conditions evolve. In this way, Gettier cases serve not as anomalies to be patched, but as diagnostic elements that signal the limitations of static definitions within dynamic systems of belief revision and information updating.

And yes: time indexing is essential. Without it, we treat epistemic states as if they were fixed, ignoring the fact that justification and truth conditions often shift over time. Temporal indexing allows us to distinguish between a belief that was justified at t₁ but no longer at t₂, and it provides the conceptual space to track when and how knowledge claims are valid. Without this, Gettier cases appear paradoxical; with it, they become predictable artifacts of dynamic belief systems.

But this only makes sense if we accept that what we’re dealing with is conceptual knowledge — knowledge understood as a temporally indexed, representational structure rather than as an unchanging metaphysical entity. It’s the subject’s best available model of a situation, justified at a specific point in time, and open to revision. In this light, time indexing isn’t a technical workaround; it’s a necessary consequence of treating knowledge as dualism. Knowledge can take a static form, but it can also be dynamic — evolving, time-bound, context-sensitive, and perspectival (as illustrated by the Rashomon effect).

Count Timothy von Icarus

↪RogueAI

Do you mean Borges' Library of Babel or the story from Genesis or something else? (The first is one of my favorite ways to think about this sort of thing).

RogueAI

↪Count Timothy von Icarus

I can't believe I said Tower of Babel! Yes, library.

Cheshire

No one has ever claimed that these cases constitute knowledge. The whole point of the Gettier problem is precisely that the definition is fulfilled, yet what we have still doesn’t count as knowledge in the way we expect it to. Ergo, JTB doesn’t work as a universal account of knowledge. — DasGegenmittel

It works just fine if we drop a few needless assumptions. That simply meeting a technical version of something believed being justified could be mistaken for knowledge long enough to base an argument on it. And then that knowledge was ever intended to be more than aspirational in it's certainty. Not really sure what Plato was going on about; always assumed the clergy found him useful for mystification of things people ought to already understand. I suppose we're in some type of aggressive agreement. Don't really see the advantage of non-generalizing something as broad as knowledge. Some of things people think is knowledge. Analytics I assume were the cure for people hiding behind words. I prefer people don't believe me if they think I'm wrong. No need to force it with symbol games.

DasGegenmittel

@Cheshire I agree with much of what you’ve said. Still, I think it’s essential to show why certain approaches fail—not just that they don’t work, but that their very structure produces the instability they aim to prevent. This is the paradox of epistemic security: the more analytic philosophy tries to define knowledge in order to eliminate luck and error, the more it constructs conceptual systems that are disconnected from actual epistemic life.

The analogy to the security dilemma in international relations is helpful here: just as states provoke greater instability by seeking more security, epistemologists often end up undermining the very concept of (dynamic) knowledge through overprecise constraints. Each definitional refinement is meant to protect, but collectively they risk sterilizing the phenomenon itself. The result is a brittle framework: too rigid to account for real-world knowing, and too narrow to be practically meaningful.

In this sense, analytic philosophy’s “curative” impulse—its desire to safeguard knowledge—can itself become pathological. What’s needed is not more insulation against epistemic risk, but a better understanding of how to navigate uncertainty without denying it. True epistemic maturity lies not in eliminating risk, but in learning to orient ourselves responsibly within it.

GETTIER – Why Plato Would Reject Justified True Belief (a Platonic dialogue)

Welcome to The Philosophy Forum!

Categories

More Discussions