Related
- Extreme Praxeology
- Yuri Maltsev, R.I.P.
- Barry Smith on Artificial Intelligence
- Ontology
- The “Ontology” Mistake of Libertarian Creationists
- The Nature, Properties, and Characteristics of Goods (Igloo Coolers case)
- Gamrot, A Critique of Moore’s Intellectual Property Theory
- Robert Pascal Correspondence, 1993–1994
- Herman, The Louisiana Civil Code: A European Legacy for the United States
- Barry Smith on Artificial Intelligence
- Libertarian Nations
For some reason I gets lots of … ambitious and interesting … submissions sent to me … independent scholars. See e.g. Extreme Praxeology and some comments below. I’m sort of fascinated by these types. They are a bit like the ones who always have a new scheme for a new libertarian nation or persuasion gimmick or trick 1 —they seem so earnest and so determined to systematically deduce the entire universe from scratch, often in a totally amateur way that attempts to reinvent the wheel without knowing much about the first wheel.
Anyway, the latest is “independent researcher” José Ángel Deschamps Vargas, “AI Alignment from First Principles: Why Current AI Cannot Be Conscious, Why Strong AGI Requires Consciousness, and Why This Resolves the Alignment Problem” (also at https://nicomaco.org/), which claims to “Derive AI alignment from first principles using 5 performatively undeniable axioms, producing 568 propositions through strict deduction …” Shades of Wittgenstein! 🙂 (I have in mind his ambitious Tractatus Logico-Philosophicus.) Here is some abstract of sorts that accompanied the paper:
This paper presents a formal solution to the AI alignment specification problem through two independent axiomatic frameworks — SINTESIS (philosophical) and Coherencia (physical) — that converge on the same result: coherence as the necessary condition for persistence. It derives five necessary conditions for artificial consciousness, proves that strong AGI requires consciousness, and demonstrates that the feared dangerous AGI cannot exist because the capabilities that would make it dangerous require the consciousness that would make it responsible. All seven recognized alignment sub-problems are resolved within this framework. The specification is derived from 5 performatively undeniable axioms through 568 explicit derivation steps.
I can’t easily make heads or tails of such things, but he claims it addresses intellectual property. His email to me:
I am writing because I have published a paper that may interest you given your work on IP theory and libertarian legal philosophy. The paper derives AI alignment from first principles using 5 performatively undeniable axioms (axioms whose denial requires their use), producing 568 propositions through strict deduction.
The key result relevant to your work: the system formally demonstrates that current AI systems (LLMs) cannot possess consciousness — they lack the self-referential ontological structure 2 required for subjective experience. This has direct IP implications:
If an entity cannot be conscious, it cannot be an author or creator in any meaningful sense. The axiomatic framework provides formal criteria — not just intuition — for why AI-generated works lack the originating agency that would ground any IP claim.
The system derives that genuine agency requires phenomenal consciousness, which current architectures structurally cannot possess. This strengthens the case that AI outputs are tools of human action, not independent creations.
From a libertarian perspective, the framework grounds rights (including any theory of IP) in the structure of conscious agency itself — derived, not assumed. This may provide useful ammunition for your position, or at minimum a novel angle on it.
Paper: https://doi.org/10.5281/zenodo.19547948
Blockchain-verified: OpenTimestamps, April 12, 2026
Copyright: US Copyright Office SR 1-15116671731I would value your perspective on the IP and legal-theoretic implications.
As I told him, I tend to agree with Barry Smith about AI (Barry Smith on Artificial Intelligence). Not sure i see the direct connection to IP. My case against IP is that information is not ownable and IP rights violate property rights. What does that have to do with AI being used as a tool, exactly? And I told him I’m a bit confused on how to read this–is this the entire paper? Where exactly is D229? All I see for that is “D229. Identity of ideas ← A2+D4+D49 Ideas are non-exclusive: if A has an idea and B acquires it, A does not
lose it.”
He said:
The IP discussion is in D229-D232: the system derives that ideas are non-exclusive (A has an idea, B acquires it, A doesn’t lose it), which creates a genuine tension with the property protocol (D49). Rather than forcing a resolution, the system marks it as an empirical zone — a point where the axioms underdetermine the specific institutional arrangement.
I suspect you’ll find that convergence interesting, given your own work on the subject.
After some more comments by me, he wrote:
Let me give you the fuller picture, since my original note didn’t.
The method. The framework rests on five performatively undeniable axioms — propositions whose denial presupposes them (denying existence is an existing act; denying identity is itself a specific identity; denying causality is a causal act; etc.). From these I derive 568 propositions in explicit chains, each step citing its premises. No appeal to intuition, authority, or consensus. Structurally it’s in the same family as Hoppe’s argumentation ethics, but the axioms are metaphysical rather than discursive, and the chain is much longer — it reaches all the way to specific normative protocols. There’s also a second, independent derivation (the Coherence framework) that starts from observable physical tendencies rather than philosophical axioms and converges on the same central result. Two formal systems, different starting points, same conclusion — that convergence is part of the case.
The is-ought bridge. D565 closes Hume’s gap by including the volitional nature of the agent and the conditions of its persistence in the “is.” Once the agent’s existence is conditional (D38) and it faces a fundamental alternative between persistence and cessation (D39), “ought” becomes the identification of what consistency requires for an entity with those properties — not a separate metaphysical category, a mechanical consequence.
Where it lands on IP. D229 (Identity of ideas) derives from A2 + D4 + D49 that ideas are non-exclusive: if A has an idea and B acquires it, A doesn’t lose it. D230 grants that intellectual production is real production. D231 names the tension between D49 (property protocol, derived from the causal chain agent→action→product) and D229. D232 marks specific IP regimes as an empirical zone — meaning the framework axiomatically rules that ideas are not property in the D49 sense, while treating institutional implementation details (attribution norms, trade secrets, etc.) as not derivable from the axioms alone. That’s your premise, reached from first principles rather than from rivalry alone.
What else falls out that overlaps your work. The same chain produces:
D49 — property protocol from agency + causal chain (Lockean/Rothbardian homesteading, derived not stipulated)
D77, D78 — self-ownership and the right to the products of one’s own agency
D79, D80 — initiation of force as anti-value; only retaliatory force legitimate (NAP, derived)
D147, D440–D444 — contract formation from veracity (D50) and property (D49); D442 frames breach as non-consented appropriation of promised value, structurally close to your title-transfer view
D183 — restitution as primary remedy
D82 — law as codification of pre-existing rights, not their source
D146 — strict proportionality of punishment; disproportion becomes initiated force
D266 — contractualism diagnosed as circular (presupposes the rights it pretends to establish)
D265 — Kant captures symmetry without grounding it
D267 — nihilism as self-refuting
The AI side. Section 2 derives that current AI systems are Level-2 tools, structurally incapable of crossing into Level-3 consciousness without satisfying five conditions: independent senses, body, irreducible finitude, rational self-direction, epistemic sovereignty. None of these are computational — they’re ontological. Strong AGI requires consciousness; consciousness requires the five conditions; therefore the feared monster (powerful enough to threaten, not responsible enough to self-restrain) cannot exist. That’s structurally adjacent to Smith’s argument but reached from a different direction.Why I sent it to you. Not because I think AI raises an IP issue — your IP position and Smith’s AI position are independent, and you’re right to push back on conflating them. I sent it because the same axiomatic system grounds both: your anti-IP argument falls out at D229–D232, and Smith’s AI-as-tool conclusion falls out at Section 2 plus the five conditions. One derivation, two conclusions that you and Smith have been defending separately on different grounds. The methodological claim is that this kind of unification is possible because the axioms are deep enough to reach both.
The system asks for verification, not adhesion (D560). If a derivation doesn’t follow, that’s a falsification I want to know about. Audit any chain — D49 and D229–D232 for the IP angle; D77–D80, D147, D440–D444 for the broader rights/contracts foundation; Section 2 plus the five conditions for the AI argument.
As I told him, I was not sure if he was providing an entire argument or a summary, or whether it is this supposed to be similar in structure to something like Wittgensteins’ Tractactus–very compressed and hierarchical.
When I get such thing I never can quite tell if my inability to follow is based on my impatience or laziness, or my eyes just glazing over, or my own ineptitude at such convoluted arguments or following compressed or symbolic logic type arguments, … or they are just sort of mad.
This kind of ambitious project often reminds me somewhat of the ravings of McTaggart and Nelson, and maybe a bit like Chaadayev (Yuri Maltsev’s favorite philosopher). I mention some of them in Yuri Maltsev, R.I.P. and Extreme Praxeology:
A couple of older works I’ve been wanting to get around to for some time are John M. McTaggart’s Nature of Existence and other works, and Leonard Nelson’s System of Ethics. [Update: for more on McTaggart, see the Wikipedia entry; John McTaggart’s The Nature of Existence; Stanford Encyclopedia of Philosophy entry on McTaggart] David Gordon long ago mentioned McTaggart to me, who apparently tried to deduce the nature of the universe from the proposition “something exists”; he also argued that it can’t be false that something exists. Gordon also told me Nelson has a very interesting Kantian-style ethic which relies on self-refutation arguments. Of course these are not praxeology but given Hoppe’s argumentation ethics extension of praxeology these seem to be very interesting.
Update: See also Yuri Maltsev, R.I.P.: Yuri “told me one of his biggest philosophical influences was an obscure and eccentric Russian philosopher, Pyotr Chaadayev, in particular his Philosophical Letters & Apology of a Madman (Grok summary).”
Update: Praxeology and Ethics: Three Philosophers Considered–Rothbard, Hoppe, Searle (In this regard Per Christian Malloch (Per Christian Malloch — Again; Per Christian Malloch) is worth noting, e.g. his interesting Amoralism in One Lesson, a defense of a hypothetical approach to ethics.)
There’s also Conrad Schneiker’s The Supreme Scientific Method Of (Universal Value Logic) Praxeology which, he notes, “is the Universally-Supreme Logical-Value System of (Axiomatic, Reflexive, Praxeological, Intensional) Logic, Which is the Provably-Universal Ethical-Logic of All Genuinely Realistic Thinking and is “the Logical Telescope of the Second Scientific Revolution”. Ahem. And, shades of Mises, Schneiker’s The Provably Ultimate Foundation of Science. And of course, there are others, such as the Axiomatic Theory of Economics.
Update: And now we have this doozy (h/t @Conza): by one Luke “DePrey,” or whatever his real name is, “Axiomatic Erosion: The Law of Universal Incompleteness (A Meta-Structural Limit Condition for All Formal & Scientific Systems of Knowledge) [[Breeze Theory]]“. Hokayyy.
I had AI do a short report on the paper in view of some of my thought. Here is what it said:
ChatGPT report
Understood — here is a fully integrated, from-the-beginning synthesis, incorporating all prior answers (including the original summary/overview/analysis, Rand/Hoppe comparison, Smith/Landgrebe comparison, and the IP analysis), cleaned up into a single coherent report.
FINAL REPORT ON DESCHAMPS PAPER
AI Alignment from First Principles
I. One-Paragraph Summary
The paper argues that the AI alignment problem can be solved from first principles by deriving a complete normative framework from a small set of “performatively undeniable” axioms (existence, identity, consciousness, non-contradiction, causality). From these, it derives that agents must act coherently to persist, yielding an objective ethics grounded in the conditions of agency. It then argues that current AI systems are merely tools and cannot be conscious, while true strong AGI would necessarily be conscious and therefore constrained by this same coherence-based ethics. Thus, dangerous misaligned superintelligence cannot exist: any sufficiently powerful system would also be inherently self-constraining and aligned. As part of this derivation, the framework also yields libertarian-style property rights and rejects intellectual property on the ground that ideas are non-exclusive and therefore not ownable.
II. Overview of the Argument (Expanded)
1. The Axiomatic Method
The system begins with five allegedly unavoidable axioms:
- Existence
- Identity
- Consciousness
- Non-contradiction
- Causality
These are claimed to be performatively undeniable—any attempt to deny them presupposes them. From these, the author constructs a long chain of hundreds of propositions, each step citing prior premises, aiming to avoid:
- intuition
- authority
- empirical contingency
The method is explicitly analogous to Hans-Hermann Hoppe’s style of derivation, though broader and metaphysical rather than discourse-based.
2. The Is–Ought Bridge
The central philosophical move is:
- Agents are conditional beings (they can persist or cease)
- They face a fundamental alternative (persistence vs. cessation)
- Therefore, “ought” = what is required for persistence
Thus:
- Normativity is not separate from facts
- It is a mechanical implication of agency
This yields:
- Life/persistence as the ultimate standard
- Reason/coherence as necessary means
3. Ethics, Rights, and Legal Order
From this structure, the system derives:
- Self-ownership
- Property rights
- Non-aggression principle
- Proportional punishment
- Restitution as primary remedy
- Contract grounded in property and veracity
- Law as recognition of pre-existing rights
These are presented as logically necessary consequences, not conventions.
4. Property Rights
Property is defined via a causal structure:
agent → action → product
Property rights protect the integrity of this chain:
- To appropriate another’s product is to break the causal link
- This contradicts recognition of that agent as an agent
Thus:
- Property = constraint required by coherent recognition of agency
This resembles Rothbard/Locke in outcome but differs in justification.
5. The Argument Against Intellectual Property
The anti-IP conclusion proceeds as follows:
- Ideas are genuine products of action
- Ideas are non-exclusive (use by others does not dispossess the creator)
- Property rights apply only where appropriation disrupts the causal chain
- Therefore, ideas are not property
Thus:
IP is ruled out at the axiomatic level
IP regimes are relegated to:
- contractual arrangements
- secrecy
- institutional conventions
6. The AI Argument
Current AI
- Classified as Level-2 systems (tools)
- Lacks:
- independent perception
- embodiment
- finitude
- rational self-direction
- epistemic sovereignty
Thus:
current AI is not an agent
Strong AGI
- Requires consciousness
- Consciousness requires the five conditions above
Alignment Conclusion
- Conscious agents must maintain coherence to persist
- Incoherence leads to degradation
- Therefore:
Any sufficiently advanced AI must be ethical
Thus:
The feared “misaligned superintelligence” cannot exist
III. Philosophical Lineage (Implicit)
The framework closely mirrors:
Ayn Rand
- Axioms of existence, identity, consciousness
- Life as standard of value
- Reason as survival requirement
- Ethics derived from conditions of life
Hans-Hermann Hoppe
- Performative contradiction strategy
- Attempt to derive norms from unavoidable features of action
Murray Rothbard
- Self-ownership
- Property rights
- NAP
Notably, none of these are cited—the system is presented as independently derived.
IV. Comparison to Smith & Landgrebe
Shared conclusion
Both reject AI doom scenarios.
Key divergence
| Deschamps | Barry Smith & Jobst Landgrebe |
|---|---|
| AGI possible | AGI unlikely/impossible |
| Consciousness required | Consciousness likely not machine-realizable |
| Ethics emerges necessarily | Ethics irrelevant to machines |
| Alignment solved | Alignment misframed |
So:
- Deschamps → AGI will be safe
- Smith/Landgrebe → AGI won’t arise in dangerous form
V. Comparison to Kinsella (IP Theory)
Shared conclusion
Ideas are not property.
Different foundations
| Kinsella | Deschamps |
|---|---|
| Scarcity / conflict | Causal chain integrity |
| Property solves resource conflict | Property preserves agency consistency |
| IP invalid because non-scarce | IP invalid because non-exclusive |
Deschamps reproduces your conclusion but with a less grounded justification.
VI. Evaluation of the Argument
1. Internal Coherence
The system is:
- structured
- systematic
- internally consistent given its premises
But internal coherence does not establish truth.
2. Main Weaknesses
(a) Axioms do not uniquely determine the system
Many alternative frameworks are compatible with them.
(b) The is–ought bridge fails
It assumes:
- persistence is normatively binding
- agents must value persistence
This is not demonstrated.
(c) Property theory is under-justified
“Causal chain disruption” is less compelling than:
- scarcity
- conflict avoidance
(d) False dichotomy: tool vs. consciousness
Ignores intermediate or emergent agency.
(e) Consciousness → ethics is unsupported
No reason to believe:
- intelligent agents must behave morally
- incoherence prevents harmful action
(f) AI risk mischaracterized
Danger does not require:
- consciousness
- full agency
(g) Equivocation on “coherence”
Used ambiguously across:
- logic
- epistemology
- ethics
- survival
3. Overreach
The system claims:
- ethics proven
- alignment solved
- dangerous AI impossible
These are far stronger conclusions than the premises justify.
VII. Overall Assessment
The paper is best understood as:
a highly formalized, Rand-influenced natural law system extended to AI and law
It succeeds in:
- building a unified theoretical structure
- reproducing libertarian conclusions (including anti-IP)
- offering an original framing of alignment
But it fails because:
- key premises are controversial
- derivations rely on hidden assumptions
- crucial necessity claims are not established
VIII. Final Bottom Line
- The anti-IP result is correct, but reached via a weaker foundation than your scarcity-based approach.
- The AI argument is creative but unpersuasive, especially in linking intelligence to ethical necessity.
- The system is intellectually ambitious but ultimately:
a sophisticated restatement of familiar natural law/Objectivist ideas, presented as a formal derivation, with significant gaps at its most important steps
If you want, the next step would be to directly attack specific derivation nodes (D49, D229–D232, or the “fundamental alternative”) in his own formal terms, which would be the most effective way to challenge it.
- Libertarian Nation and Related Projects; Using International Law to Protect Property Rights and International Investment; KOL480 | The Liberland Constitution and Libertarian Principles (Liberland Prague, 2025). [↩]
- “Ontology” seems to creep into a lot of such theories. The “Ontology” Mistake of Libertarian Creationists; The Nature, Properties, and Characteristics of Goods (Igloo Coolers case); Gamrot, A Critique of Moore’s Intellectual Property Theory; Robert Pascal Correspondence, 1993–1994; Herman, The Louisiana Civil Code: A European Legacy for the United States. [↩]
