To summarize, you came up with Wolpert's law first in your "Theory of the Universe and Mind". Or at least some of the ideas.
Wolpert's law says that two computers cannot mutually predict each other's output, even when each has complete knowledge of the other's state and source code. As a corollary, a computer cannot predict its own output (as the special case where the two computers are identical). Wolpert's law applies to all levels of computing power, whether finite state machines, Turing machines with infinite memory, or even more powerful machines with oracles. It applies to human brains and to physical systems, such as a universe in which a computer is implemented being unable to make perfect predictions about the universe that contains it. We use prediction to measure intelligence. Therefore we cannot predict (and thus cannot control) machines that are smarter than us. Does your TUM discuss this? Wolpert's law also has consequences for physics. Since a universe perfectly predicts the state of any observers or computers within it, those observers necessarily see the universe as random, where Laplace's demon can't exist. Observers are described by a deterministic wave function as seeing particles with a probability distribution. Particles don't actually exist, even though they are a useful model most of the time. Wolpert's law is easy to prove. If you assume it is false, it leads to contradictions like Newcomb's paradox. But the rock-scissors-paper proof is easier. Suppose that player A simulates a copy of player B and knows that B will play Rock. So A plays Paper and wins. But B also runs a copy of A and knows that A will play Paper, so B plays Scissors and wins. So who actually wins? -- Matt Mahoney, [email protected] On Wed, Apr 15, 2026, 10:00 AM twenkid <[email protected]> wrote: > In a discussion from 8.2025, a part of which I archived at SIGI-2025 as a > paper, titled "Power overrides intelligence", Matt mentioned some > "Wolpert's law" of which I didn't know at the time. Later that year I > discovered it without searching for it from other publications, and I found > out that in fact this is *"Arnaudov's law"*, published and explained 4-5 > years before the first paper of Wolpert (2007-2008) and up to 16 years > before a consequent paper from 2018*. There is a match even on "Liar's > paradox" - it is addressed in my work as well, however I "scaled" it to > even more absurd form, ridiculed it and explained why it was nonsense and > it didn't prove what the "cheaters" intended. It is related to the concept > in TUM "resolution of causality-control and perception"; comparison of > representation with different and incompatible RCCP. "What's *the color* > of the rainbow?" (one color) These are ill-posed problems, presented as > well-posed by "cheaters". In this circumstances, any higher-resolution > answer is correct, as the lower-resolution causality-control doesn't have > capacity to distinguish the answers. > > The original "law" from TUM also reflects the hierarchical structure of > the predictions and CCUs, which work as and are created by *hierarchical > universal > simulators of virtual universes, *i.e. predictors and causers, which are > multi-resolution, multi-range, multi-scale, multi-domain, multi-precision, > ... multi- ... Wolpert has a corresponding concept "*a general-purpose > prediction device, capable of correctly predicting different aspects of the > universe’s future", * however his "devices", at least so long as I > interpreted the paper, are *flat *and they predict/not predict, > true/false", "A/not A" - 1-bit nonsense in the real Universe, as explained > in TUM. > > This paper is a chapter from the book *"Reflections on Everything",* or > *"Listove"*, the second-largest volume from "*The Prophets of the > Thinking Machines: AGI & Transhumanism: History, Theory and Pioneers; Past, > Present and Future"*, SIGI-2025, 10.2025. > > > <https://www.researchgate.net/publication/403842042_Wolpert's_Theorem_about_Mutual_Unpredictability_and_the_Impossibility_of_Subuniverses_to_Predict_with_Highest_Resolution_of_Causality-Control_are_Rediscoveries_of_Concepts_from_Theory_of_Universe_and_> > *Wolpert’s Theorem about Mutual Unpredictability and the Impossibility of > Subuniverses to Predict with Highest Resolution of Causality-Control are > Rediscoveries of Concepts from Theory of Universe and Mind – Entangled with > Unnecessary Mathematical Notation and Unsatisfiable Premises* > > > https://www.researchgate.net/publication/403842042_Wolpert's_Theorem_about_Mutual_Unpredictability_and_the_Impossibility_of_Subuniverses_to_Predict_with_Highest_Resolution_of_Causality-Control_are_Rediscoveries_of_Concepts_from_Theory_of_Universe_and > <https://www.researchgate.net/publication/403842042_Wolpert's_Theorem_about_Mutual_Unpredictability_and_the_Impossibility_of_Subuniverses_to_Predict_with_Highest_Resolution_of_Causality-Control_are_Rediscoveries_of_Concepts_from_Theory_of_Universe_and_> > > *Abstract* > This work presents a critical and comparative analysis of the theoretical > limits of prediction, inference, and control in physical systems, focusing > on the framework introduced by David H. Wolpert and its relationship to the > Theory of Universe and Mind (TUM). Wolpert’s theorems establish that no > inference device embedded within a universe can achieve complete and > error-free prediction, observation, or control of other devices, and that > mutual perfect predictability between independent agents is impossible. > These results are often interpreted as formal limits on knowledge, > extending earlier ideas such as the impossibility of Laplace’s demon. > The present paper argues that these conclusions correspond to principles > previously articulated within TUM, where the universe is modeled as a > hierarchical computational structure composed of interacting > causality-control units (CCUs). In this framework, predictive and causal > capacities are determined by the resolution of causality-control and > perception (RCCP), with higher-level subsystems operating on compressed > representations of lower-level dynamics. As a consequence, all subsystems > exhibit bounded predictive power, limited memory, and partial control, > while only the universe as a whole achieves maximal resolution and > completeness. > The analysis further examines the assumptions underlying formal inference > models, particularly the treatment of devices as independent entities. It > is argued that, in physically realized systems, all subsystems are > inherently correlated due to shared origin, continuous interaction, and > embedding within a common dynamical structure. This challenges the > applicability of certain formal premises and suggests that observed limits > on prediction arise from structural and hierarchical constraints rather > than solely from logical or computational restrictions. > Additionally, the paper critiques the use of highly abstract logical > formulations – such as binary query models and paradox-based arguments – as > insufficient for capturing the multi-scale, continuous, and physically > grounded nature of real-world systems. Instead, it advocates for models > that incorporate hierarchical organization, varying resolutions, and the > interplay between compression and prediction. > The conclusion is that while formal results on the limits of inference are > valid and significant, they can be more comprehensively interpreted within > a broader framework that accounts for the hierarchical and embedded nature > of cognition and physical processes. > ... > This work is a chapter from the book “Reflections on > Everything”, or “Listove” (Листове по всичко), which is an appendix and the > second-biggest volume from the hyperbook “The Prophets of The Thinking > Machines: Artificial General Intelligence and Transhumanism: History, > Theory and Pioneers; Past, Present and Future”, T.Arnaudov, 2025-1.2026 – > all published at the yearlong virtual conference “Self-Improving General > Intelligence/Thinking Machines” 2025, organized by The Sacred Computer: > Thinking Machines, Creativity and Human Development, a virtual multi- and > interdisciplinary AGI and Transhumanism research and development > laboratory, created in 2000. The first classic works of Theory of Universe > and Mind were published between 2001 and 2004. Core ideas from the theory > were presented in a lecture at Technical University of Sofia in September > 2009 and in the world’s first university course in AGI at the University of > Plovdiv “Paisii Hilendarski”, Bulgaria in 2010 and 2011. > > *The Sacred Computer: Thinking Machines, Creativity and Human > Development: *2000-2026 > https://github.com/twenkid > > * See references to my work in the paper and in the referred books > * Physical limits of inference, David H. Wolpert, MS 269-1, NASA Ames > Research Center, Moffett Field, CA 94035, USA, > https://arxiv.org/pdf/0708.1362 [Submitted on 10 Aug 2007 (v1), last > revised 23 Oct 2008 (this version, v2)] > * Theories of Knowledge and Theories of Everything, D.Wolpert, February > 2018 > > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + > delivery options <https://agi.topicbox.com/groups/agi/subscription> > Permalink > <https://agi.topicbox.com/groups/agi/T141084b5625cb2c2-M85e35d546935b9d061145a3e> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T141084b5625cb2c2-Maf9aeb50eb0203c373a6f518 Delivery options: https://agi.topicbox.com/groups/agi/subscription
