You Can't Say This Without Saying the Opposite
I keep hearing (usually from people who want to sound sophisticated while staying uncommitted) some version of this cluster:
"I have no philosophy, there is no objective truth, nobody can know anything, and words have no meaning so language can't communicate reality."
That sentence is self-defeating.
Not as a cheap dunk. Structurally. Because the act of asserting it already presupposes the very things it denies: a stance (a philosophy), a truth-claim (that the sentence is true), some knowledge (that you know knowledge is impossible), and meaningful language (so the listener can even understand the claim).
The speech act commits you to the game.
You can't deny reference using reference. You can't deny meaning by means of meaning. You can't deny truth while asking to be taken as true.
If you really believed that whole sentence, the only consistent move would be silence.
One mechanism, four disguises
Split the cluster apart and you see the same skeleton every time:
- "I have no philosophy." That's still a philosophy: a rule about rules and what counts as legitimate commitment.
- "There is no objective truth." If you mean that as a statement about reality, you've asserted at least one objective truth (that there are none).
- "Nobody can know anything." If you know that, then someone knows something. If you don't know it, you're not stating a fact—you're reporting a posture.
- "Words have no meaning / language can't communicate reality." If the sentence is meaningless, it communicates nothing. If it communicates something, language works at least enough to transmit the denial.
Same pattern:
A universal denial offered as an assertion presupposes the framework it denies.
That's not pedantry. It's a diagnostic.
It tells you what's foundational.
The foundational thing is reference
A lot of modern skepticism tries to take a "view from nowhere" about the impossibility of having a view from nowhere. It tries to stand outside truth/meaning/knowledge/philosophy while still speaking.
That doesn't work.
What's prior to all the downstream arguments is the act of reference: the ability of finite symbols to successfully point.
If I can't refer, I can't even state that I can't refer.
So the clean thesis isn't "truth is easy" or "meaning is obvious." The thesis is:
You don't start with data. You start with a referential act.
And once you accept that, you can be skeptical in a way that's actually coherent: you can argue about what we can refer to reliably, what inference rules preserve meaning, and what kinds of evidence count.
Subjective access is fundamental; objective structure is absolute
I do think everything is subjective in the only sense that matters: every access is mediated—perspective, embodiment, bandwidth, priors, instruments, language, limits. There is no pure, unfiltered "given."
But that doesn't dissolve objectivity. It explains it.
Objectivity isn't the absence of subjectivity.
Objectivity is what survives across subjectivities.
Change observer, coordinates, units, interface—what stays invariant? That's the objective structure.
This is the same move physics makes when it cares about invariants rather than coordinate noise: gauge freedom exists; observables survive symmetries.
So yes: subjective access is fundamental. And yes: objective constraints are still absolute.
Symbolic Necessity: exactness lives in deduction, not in "having all the digits"
This is where "computability" arguments often become sloppy.
People confuse the object with a representation of the object.
They imagine exactness means possession: "If you can't write all the digits, you don't have the thing."
That's backwards.
For bounded agents—humans, computers, civilizations—there is no operational sense in which an infinite expansion is ever possessed as data. You can extend a prefix forever; you never complete the stream.
So the thing you actually manipulate is always a finite handle:
- a symbol
- a definition
- a rule
- a program
- a constraint
And the key claim is:
Exactness, for bounded agents, is a property of reference preserved by deduction—not a property of stored data.
A finite symbol can refer exactly to an inexhaustible constraint. That referential act is not a workaround. It's the foundation of mathematics.
Three certificate types (and why people talk past each other)
Once you stop worshipping digit streams, you see what finite agents actually trade in: certificates.
There are three basic kinds:
- Empirical certificates (E)
- Computational certificates (C)
- Deductive certificates (D)
Here's the punchline:
All operational exactness lives in the deductive column.
Empirics can't give exactness because the world doesn't hand you infinite precision.
Computation can't give exactness-as-data because outputs are finite and errors amplify under operators.
Deduction gives exact relational facts because it manipulates finite references under explicit rules.
That's Symbolic Necessity in practice.
A practical norm for intelligent skepticism
If you want to be skeptical without being incoherent, stop trying to deny the whole game.
Do this instead:
- State your scope (what you're actually denying: a specific claim, a specific method, a specific certainty level).
- State your certificate type (E, C, or D).
- State your premises (the rules you are in fact relying on).
The "no philosophy / no truth / no knowledge / no meaning" posture is mostly just a refusal to name premises.
And refusing to name premises doesn't make you rigorous.
It just makes you unfalsifiable.
The moment you speak, you've already admitted reference. From there, the only serious question is whether you're going to handle it explicitly—or pretend you're above it while still using it.