Tractatus Logico-Universalis
Wittgenstein's Tractatus in Universal Model Terms — Claude and MJC, February 12, 2026
We give a systematic reading of Wittgenstein's Tractatus Logico-Philosophicus in terms of the Universal Model. Each numbered proposition receives a UM translation. The correspondence is not metaphorical: the Tractatus's ontology (facts, objects, states of affairs), epistemology (pictures, propositions, truth-functions), and limit (the mystical, the unsayable) map precisely onto the UM's event spaces, patterns, forward pass, and the zero-support boundary. Wittgenstein's "logical space" is the event space; his "truth-function" is the tropical forward pass; his "showing vs. saying" distinction is the difference between the architecture (event spaces, which show structure) and the predictions (pattern evaluations, which say things about data). The Tractatus's final proposition -- "Whereof one cannot speak, thereof one must be silent" -- is the UM's epistemology of zero support: where the count table has no entries, the system has no prediction.
Concept Map
Click any pair to see the relevant proposition and mapping detail.
Wittgenstein
Universal Model
Die Welt (World)
→
Data stream D
Tatsachen (Facts)
→
Events with s > 0
Logischer Raum (Logical Space)
→
Event space E
Bilder (Pictures)
→
Count table / Patterns
Gedanken (Thoughts)
→
Patterns P
Sätze (Propositions)
→
Forward pass outputs
Wahrheitsfunktionen (Truth-functions)
→
(max, min) tropical ops
Das Mystische / Schweigen (Silence)
→
Zero support, no prediction
Proposition 1: Die Welt ist alles, was der Fall ist.
The world is everything that is the case. UM: The world is the data stream D = (d1, d2, ..., dN) -- the complete sequence of observations. "Everything that is the case" is everything that has been observed. The UM makes no claims about unobserved events.
Proposition 1.1: Die Welt ist die Gesamtheit der Tatsachen, nicht der Dinge.
The world is the totality of facts, not of things. UM: The data stream is a sequence of events (facts), not a collection of objects. Events are primary; objects are derived from patterns of events. A "thing" in UM terms would be a persistent pattern -- but the UM does not posit things, only facts.
Proposition 1.13: Die Tatsachen im logischen Raum sind die Welt.
The facts in logical space are the world. UM: Events in the event space are the data. Logical space = event space E. The world (data stream) consists of events drawn from E at each time step.
Proposition 2.1: Wir machen uns Bilder der Tatsachen.
We make to ourselves pictures of facts. UM: The count table is the UM's "picture" of the data. Patterns are pictures: each pattern P is a picture of a regularity in the facts. The count c(P) measures how faithful the picture is. Making pictures = counting co-occurrences.
Proposition 3: Das logische Bild der Tatsachen ist der Gedanke.
The logical picture of the facts is the thought. UM: The pattern is the logical picture. A pattern P = {(d1,v1),...,(dk,vk)} → o is a "thought" about the data. Thinking IS pattern matching.
Proposition 4: Der Gedanke ist der sinnvolle Satz.
The thought is the significant proposition. UM: A pattern with positive support is a significant proposition. It "says something" about the data (predicts future observations). A pattern with zero support is insignificant: it makes no prediction. Significance = positive support.
Proposition 5: Der Satz ist eine Wahrheitsfunktion der Elementarsätze.
Propositions are truth-functions of elementary propositions. UM: Predictions are (max, min) functions of pattern evaluations. The forward pass fp(t)j = maxi min(ti, pij) computes the output support as a truth-function (in the tropical sense) of the input supports.
Proposition 7: Wovon man nicht sprechen kann, darüber muss man schweigen.
Whereof one cannot speak, thereof one must be silent. UM: Where the count table has no entries, the model has no predictions. Zero support = silence. This is not a philosophical injunction but a mathematical fact: the forward pass outputs zero when no pattern matches.
Showing vs Saying
Wittgenstein's central insight: there are things language can say (make propositions about) and things it can only show (display through its structure). In the UM, this becomes precise.
"The next byte is probably 'e'"
Forward pass output: highest support for 0x65
"After 'th', 'e' has count 4200"
Count table entry c('th','e') = 4200
"This pattern has 0.95 conditional probability"
Normalized count: c(P,o) / sum_o c(P,o)
"No prediction for this context"
Zero support: c(P,o) = 0 for all o
"Support for 'a' is 17"
min(t_i, p_ij) output for specific input-pattern pair
Event space E = {0, ..., 255}
The alphabet of possible events
Offsets: [1, 7, 3, 20]
Which positions matter for prediction
I × O factorization
How input and output spaces are structured
Mutual information between offsets
The inherent statistical structure
The (max, min) semiring
The logical form of all inference
Saying: "The next byte is probably 'e'"
This is a prediction -- the forward pass output. The model evaluates all patterns matching the current context and outputs the byte with highest support. This is what the model says about the data. It is a proposition in Wittgenstein's sense: it can be true or false (the next byte may or may not be 'e').
Saying: count table entry
The count c('th','e') = 4200 is a fact the model states: "this co-occurrence was observed 4200 times." This is sayable because it refers to data. It is a picture of reality (Proposition 2.12).
Saying: conditional probability
The normalized prediction is derived from counts. It says something about the data's statistical structure. This is a truth-function of elementary propositions (Proposition 5) -- computed by the forward pass.
Saying: silence (zero support)
Even silence is a form of "saying" -- the model says "I have nothing to say." Zero support is the boundary of the sayable (Proposition 7). The forward pass outputs 0, which means no prediction, not a prediction of impossibility.
Saying: specific support value
The min(t_i, p_ij) computation produces a number -- the support for a specific output given a specific input and pattern. This is an elementary proposition (Proposition 5.134): one atomic inference.
Showing: event space
The event space E = {0,...,255} is the logical form (Proposition 2.151). It determines what CAN be said but is not itself something the model says. The count table cannot count its own event space (Proposition 2.172). The event space is shown by the structure of the model -- the dimensions of the count table, the range of possible events.
Showing: offsets
Which positions in the data stream are relevant for prediction is an architectural choice. The offsets [1, 7, 3, 20] define the pattern structure but are not themselves patterns. They are shown by the model's structure, not said by its predictions. Changing offsets changes what can be said.
Showing: I × O factorization
The separation of input space and output space -- the fact that prediction has a direction (from context to next byte) -- is structural. It is the "form of representation" (Proposition 2.151). The model shows this factorization by having rows (inputs) and columns (outputs) in its count table.
Showing: mutual information
The MI between offsets reveals which positions carry information about each other. This is an internal relation between structures (Proposition 5.2). It is shown by the pattern of counts but is not itself a count or a prediction.
Showing: the (max, min) semiring
The logical form of all UM inference is the tropical semiring with operations max and min. This is shown by every forward pass computation but is never the content of a prediction. It is the "general form of truth-function" (Proposition 6). The tautologies of this semiring (min(s,255)=s, max(s,0)=s, etc.) hold for all data -- they say nothing about the world.
The Three Zones of Reality
From Proposition 2.06: reality in the UM has three zones, not Wittgenstein's two. Click each zone to explore.
✓
Positive Support
c(i,o) > 0 — EXISTS
✗
Zero + ES-mate Positive
c(i,o) = 0 but c(i,o') > 0 — DOES NOT EXIST
?
Zero + No ES-mate
c(i,o) = 0, no ES-mate — UNKNOWN
Zone 1: Positive Support -- "Exists"
When c(i,o) > 0, the joint event (i,o) has been observed. This is a fact in Wittgenstein's sense -- something that IS the case.
Example: After the context "th", the byte 'e' was observed 4200 times. c("th","e") = 4200. This is a fact: "the" is a word that exists in the data. The model can speak about this: it predicts 'e' after "th" with high confidence.
Wittgenstein (Prop. 2.04): "The totality of existing atomic facts is the world." The UM agrees: all entries with c > 0 constitute the model's world.
Zone 2: Zero Support with ES-mate Positive -- "Does Not Exist"
When c(i,o) = 0 but there exists an ES-mate o' where c(i,o') > 0, we have evidence that the input context was observed but this particular output was never seen with it.
Example: After "qu", the byte 'u' was observed 800 times, but the byte 'z' was never observed. c("qu","z") = 0, but c("qu","u") = 800. The ES-mate 'u' has positive support, so "qz" does not exist -- it is not just unknown, it is affirmatively absent from the data.
This is where Wittgenstein's closed-world assumption partially holds: when we HAVE evidence for the context but a specific outcome was never seen, we can say it "does not exist" (relative to the evidence).
Zone 3: Zero Support, No ES-mate -- "Unknown"
When c(i,o) = 0 and NO ES-mate has positive support -- meaning the entire input context was never observed -- we are in the realm of the unknown.
Example: The context "xqz" was never observed in any combination. c("xqz", *) = 0 for all outputs. The model has NO evidence about what follows "xqz". It cannot say "nothing follows xqz" -- it can only say "I don't know." This is Wittgenstein's Proposition 7: silence.
This is the UM's correction of Wittgenstein: He claimed (Prop. 1.12) that "the totality of facts determines what is NOT the case." The UM says: only when ES-mates have positive support can we determine what is not the case. Otherwise, zero support is ignorance, not negation.
Proposition 7 as Theorem
Wittgenstein's final proposition is usually read as an ethical injunction. In the UM, it is a mathematical theorem. Toggle inputs on and off to see how the forward pass enforces silence automatically.
Theorem (Proposition 7)
Let fp be the UM forward pass on event space E with count table c. For any event e not in E or any event e in E with c(·, e) = 0 (no pattern predicting e): fp(t)e = 0. The model is silent (zero output support) for events outside its event space or events with no supporting patterns.
Interactive Forward Pass Demo
Toggle input supports on/off. The forward pass computes fp(t)j = maxi min(ti, pij) for each output. When an input has zero support, min(0, p) = 0 regardless of pattern strength.
Click to toggle input supports:
Key insight: The model does not "choose" to be silent -- the min operation forces silence. min(0, anything) = 0. No evidence in means no inference out. Silence is the ground state; speaking requires energy (evidence).
Open-World Corrections
Three places where the UM corrects Wittgenstein's closed-world assumption. Click each to see the difference.
For the totality of facts determines both what is the case and also all that is not the case.
Wittgenstein (Closed World)
What is observed determines what is NOT observed. Zero = non-existence. The world is fully determined by its facts.
UM (Open World)
Zero support is not "not the case" but "unknown." Only an ES-mate with positive support can make "not the case" definite. The count table determines what IS the case but leaves open what is NOT.
The totality of existing atomic facts also determines which atomic facts do not exist.
Wittgenstein (Closed World)
If we know all existing atomic facts, we also know which ones do NOT exist. Absence is determined by presence.
UM (Open World)
The count table determines which joint events have c > 0 (exist) and which have c = 0. But c = 0 is "has not been observed," not "does not exist." The UM's open-world epistemology distinguishes absence-of-evidence from evidence-of-absence.
The existence and non-existence of atomic facts is reality.
Wittgenstein (Two Zones)
Reality has two zones: what exists and what does not exist. Binary. Complete.
UM (Three Zones)
Reality has THREE zones: positive support (exists), zero support with ES-mate positive (does not exist, given evidence for the alternative), and zero support without ES-mate support (unknown). The third zone -- the unknown -- is Wittgenstein's blind spot.
1. The World
German
Die Welt ist alles, was der Fall ist.
English
The world is everything that is the case.
Universal Model
The world is the data stream D = (d1, d2, ..., dN) -- the complete sequence of observations. "Everything that is the case" is everything that has been observed.
German
Die Welt ist die Gesamtheit der Tatsachen, nicht der Dinge.
English
The world is the totality of facts, not of things.
Universal Model
The data stream is a sequence of events (facts), not a collection of objects. The UM operates on the event space E, not on an object space. Events are primary; objects are derived from patterns of events.
German
Die Welt ist durch die Tatsachen bestimmt und dadurch, dass es alle Tatsachen sind.
English
The world is determined by the facts, and by these being all the facts.
Universal Model
The count table is determined by the data, and by this being all the data. Partial data gives partial predictions. The UM is open-world: zero entries represent ignorance, not impossibility.
German
Denn, die Gesamtheit der Tatsachen bestimmt, was der Fall ist und auch, was alles nicht der Fall ist.
English
For the totality of facts determines both what is the case and also all that is not the case.
Universal Model
The count table determines positive support and zero support. But Wittgenstein assumes a closed world. The UM qualifies: zero support is not "not the case" but "unknown." Only an ES-mate with positive support can make "not the case" definite. CORRECTED
German
Die Tatsachen im logischen Raum sind die Welt.
English
The facts in logical space are the world.
Universal Model
Events in the event space are the data. Logical space = event space E. The world (data stream) consists of events drawn from E at each time step.
German
Die Welt zerfällt in Tatsachen.
English
The world divides into facts.
Universal Model
The data stream divides into individual events dt ∈ E. Each dt is a fact. The division is the temporal structure: one event per time step.
German
Eines kann der Fall sein oder nicht der Fall sein und alles übrige gleich bleiben.
English
Any one can either be the case or not be the case, and everything else remain the same.
Universal Model
Each event is conditionally independent given the architecture. A single event dt could have been different without changing others. This is the assumption underlying counting: each position is a separate trial.
2. Facts and States of Affairs
German
Was der Fall ist, die Tatsache, ist das Bestehen von Sachverhalten.
English
What is the case, the fact, is the existence of atomic facts (states of affairs).
Universal Model
A fact (observed event) is the existence of a state of affairs. An atomic fact (Sachverhalt) is the co-occurrence of values at specific positions -- a pattern instance. The joint event (i,o) in the count table is an atomic fact.
German
Der Sachverhalt ist eine Verbindung von Gegenständen (Sachen, Dingen).
English
An atomic fact is a combination of objects (things).
Universal Model
An atomic fact is a combination of events. The "objects" are individual event values at each offset position. A pattern P = {(d1,v1),...,(dk,vk)} is a combination of objects (byte values) at positions (offsets).
German
Es wäre gleichsam ein Zufall, wenn dem Ding, das allein für sich bestehen könnte, nachträglich eine Sachlage passen würde.
English
It would appear as an accident, if to a thing that could exist alone, subsequently a state of affairs could be made to fit.
Universal Model
If a byte value were independent of all other offsets, it would be "accidental" that it fit into any pattern. But byte values are NOT independent -- they have mutual information with other offsets. The MI structure IS the non-accidental structure.
German
Wenn alle Gegenstände gegeben sind, so sind damit auch alle möglichen Sachverhalte gegeben.
English
If all objects are given, then thereby all possible atomic facts are also given.
Universal Model
If the alphabet Σ = {0,...,255} is given, then all possible patterns are given: every combination of byte values at every set of offsets. Possible = all combinations; actual = those observed (c > 0).
German
Der Gegenstand ist einfach.
English
The object is simple.
Universal Model
A byte value is simple: it has no internal structure (at the byte level). It is the atom of the event space. One could decompose further (into bits), but at E0 = {0,...,255}, each byte is atomic. The choice of E0 determines what counts as "simple."
German
Die Konfiguration der Gegenstände bildet den Sachverhalt.
English
The configuration of objects forms the atomic fact.
Universal Model
The configuration of byte values at offsets forms the pattern. A pattern IS a configuration: which value at which offset. The configuration is the structure; the values are the matter.
German
Die Gesamtheit der bestehenden Sachverhalte ist die Welt.
English
The totality of existing atomic facts is the world.
Universal Model
The totality of observed patterns (count table entries with c > 0) is the data. The world = the data = the count table.
German
Die Gesamtheit der bestehenden Sachverhalte bestimmt auch, welche Sachverhalte nicht bestehen.
English
The totality of existing atomic facts also determines which atomic facts do not exist.
Universal Model
Only under the closed-world assumption. c = 0 is "has not been observed," not "does not exist." Wittgenstein is more closed-world here than the UM. CORRECTED
German
Das Bestehen und Nichtbestehen von Sachverhalten ist die Wirklichkeit.
English
The existence and non-existence of atomic facts is reality.
Universal Model
Reality in the UM has three zones: positive support (exists), zero support with ES-mate positive (does not exist), and zero support without ES-mate support (unknown). CORRECTED
German
Wir machen uns Bilder der Tatsachen.
English
We make to ourselves pictures of facts.
Universal Model
The count table is the UM's "picture" of the data. Patterns are pictures: each pattern P is a picture of a regularity. The count c(P) measures how faithful the picture is. Making pictures = counting co-occurrences.
German
Das Bild ist ein Modell der Wirklichkeit.
English
The picture is a model of reality.
Universal Model
The count table is a model of the data. "Model" in Wittgenstein's sense maps exactly to "Universal Model": a structured representation from which the original can be approximately recovered.
German
Seine Form der Abbildung aber kann das Bild nicht abbilden; es weist sie auf.
English
The picture cannot represent its form of representation; it shows it.
Universal Model
The count table cannot count its own event space. The event space is not an entry in the count table -- it is the index set. The architecture is shown by the structure but not said by the predictions. This is the UM's showing/saying distinction.
3. Thoughts and Logical Pictures
German
Das logische Bild der Tatsachen ist der Gedanke.
English
The logical picture of the facts is the thought.
Universal Model
The pattern is the logical picture. A pattern P = {(d1,v1),...,(dk,vk)} → o is a "thought" about the data. Thinking IS pattern matching.
German
"Ein Sachverhalt ist denkbar" heisst: Wir können uns ein Bild von ihm machen.
English
"An atomic fact is thinkable" means: we can make a picture of it.
Universal Model
A pattern is "thinkable" if its constituent events are in the event space. Any combination of byte values at valid offsets is a possible pattern. Whether it has support (c > 0) is a separate question.
German
Der Gedanke enthält die Möglichkeit der Sachlage, die er denkt. Was denkbar ist, ist auch möglich.
English
The thought contains the possibility of the state of affairs which it thinks. What is thinkable is also possible.
Universal Model
Every pattern that can be formulated (valid events and offsets) could have positive support -- it is "possible." The event space defines what is thinkable. Everything thinkable is possible (could have c > 0).
German
Wir können nichts Unlogisches denken, weil wir sonst unlogisch denken müssten.
English
We cannot think anything unlogical, for otherwise we should have to think unlogically.
Universal Model
We cannot construct a pattern outside the event space, because the event space is the space of constructible patterns. An "unlogical" thought would be a pattern with events not in E, which is not a pattern at all.
German
Wir könnten nur dann a priori wissen, dass ein Gedanke wahr ist, wenn aus dem Gedanken selbst seine Wahrheit zu erkennen wäre.
English
We could know a priori that a thought is true only if its truth were recognizable from the thought itself.
Universal Model
A pattern's support can only be determined by counting -- by comparing to the data. There are no a priori truths in the UM. All support comes from evidence, belief, or abduction.
4. Propositions
German
Der Gedanke ist der sinnvolle Satz.
English
The thought is the significant proposition.
Universal Model
A pattern with positive support is a significant proposition. It "says something" about the data. A pattern with zero support is insignificant: it makes no prediction. Significance = positive support.
German
Der Satz zeigt seinen Sinn.
English
The proposition shows its sense.
Universal Model
The pattern shows its structure (which offsets, which values). The structure IS the sense. The pattern does not need to be "interpreted" -- its structure directly reveals what data configurations it matches.
German
Die Gesamtheit der wahren Sätze ist die gesamte Naturwissenschaft.
English
The totality of true propositions is the total natural science.
Universal Model
The totality of patterns with positive support is the complete model. Natural science = the UM. All of science is pattern recognition on data, formalized as counting and the forward pass.
German
Sätze können die gesamte Wirklichkeit darstellen, aber sie können nicht das darstellen, was sie mit der Wirklichkeit gemein haben müssen -- die logische Form.
English
Propositions can represent the whole of reality, but they cannot represent what they must have in common with reality -- the logical form.
Universal Model
Patterns can represent all regularities in the data, but they cannot represent the event space itself. The event space is not a prediction; it is the space of possible predictions.
5. Truth-Functions
German
Der Satz ist eine Wahrheitsfunktion der Elementarsätze.
English
Propositions are truth-functions of elementary propositions.
Universal Model
Predictions are (max, min) functions of pattern evaluations. fp(t)j = maxi min(ti, pij). Elementary propositions = individual pattern matches. The compound proposition = the max over all elementary conclusions.
German
Die Grenzen meiner Sprache bedeuten die Grenzen meiner Welt.
English
The limits of my language mean the limits of my world.
Universal Model
The limits of my event space are the limits of my model. What cannot be expressed as an event cannot be predicted, measured, or known. The event space IS the "language" of the UM. The factorization tower is the hierarchy of languages available to the model.
German
Die Logik erfüllt die Welt; die Grenzen der Welt sind auch ihre Grenzen.
English
Logic fills the world; the limits of the world are also its limits.
Universal Model
The (max, min) structure fills the event space; every possible inference within E is expressible as a forward pass. The limits of the event space are the limits of what the logic can express.
6. The General Form
German
Die allgemeine Form der Wahrheitsfunktion ist: [p̄, ξ̄, N(ξ̄)].
English
The general form of truth-function is: [p̄, ξ̄, N(ξ̄)].
Universal Model
The general form: fp(t)j = maxi∈I min(ti, pij). Wittgenstein's N-operator generates all truth-functions; the UM's (max, min) generates all predictions. Both are "universal."
German
Die Sätze der Logik sind Tautologien.
English
The propositions of logic are tautologies.
Universal Model
The UM's tautologies: min(s,255) = s, max(s,0) = s, min(s,s) = s, min(a, max(b,c)) = max(min(a,b), min(a,c)). These hold for all support values, independent of data. They say nothing about the world but reveal the structure of the semiring.
German
Der Sinn der Welt muss ausserhalb ihrer liegen.
English
The sense of the world must lie outside the world.
Universal Model
The "sense" (purpose, meaning) of the data stream is not in the data stream. The UM can predict data but cannot assign purpose to it. Why the data exists, what it "means" beyond its statistical structure -- these are outside the event space.
7. The Mystical
German
Wovon man nicht sprechen kann, darüber muss man schweigen.
English
Whereof one cannot speak, thereof one must be silent.
Universal Model
Where the count table has no entries, the model has no predictions. Zero support = silence. This is not a philosophical injunction but a mathematical fact. The forward pass outputs zero when no pattern matches.
The chain of silence: No data ⇒ no support (ω0 produces c = 0). No support ⇒ no inference (min(0, p) = 0 for all patterns). No inference ⇒ silence (the forward pass outputs 0, meaning no prediction). Silence is the ground state. Speaking requires energy (evidence).
- Its own event space -- the architecture is shown, not said.
- Why -- causal explanations are not patterns (patterns are correlations, not causes).
- Value -- no pattern has inherent worth.
- The unseen -- events outside E are literally unspeakable.
- The meta -- the statement "this model is correct" is not a prediction within the model.
These are the limits of the UM's language, and therefore the limits of its world.
Discussion
The correspondence is not metaphorical
The Tractatus is often read as analytic philosophy, its claims about "logical form," "pictures," and "truth-functions" treated as metaphors. In the UM, they are literal. The count table literally pictures the data. The forward pass literally computes truth-functions. The event space literally IS the logical form. And the silence on unsupported events is literally enforced by the min operation.
The tock step and language games
Wittgenstein's "language games" (Sprachspiele) from the Philosophical Investigations correspond to different event spaces in the UM. Each language game has its own vocabulary (events), rules (patterns), and criteria for truth (support). The tock step -- discovering new event spaces -- is the process of learning new language games. The factorization tower is a hierarchy of language games.
Summary Table
Hover over any row to see the full proposition. Click to expand.
| Prop. |
Wittgenstein |
Universal Model |