Hoffman, Donald
▶ Watch on YouTubeCore ideas
Hoffman presents his central claim as a mathematical result, not speculation: using evolutionary game theory (building on formal tools from John Maynard Smith and colleagues including mathematician Chetan Prakash), he argues the probability that natural selection would shape any sensory system to perceive true structures of objective reality (order, topology, metric) is, in the relevant formal sense, exactly zero — because fitness payoff functions have no general requirement to preserve such structures. He frames perception as existing purely to "guide adaptive behavior," analogous to a computer desktop interface that deliberately hides the underlying transistor-level "truth" because showing it would be useless and overwhelming; when his own graduate students ran genetic-algorithm simulations, organisms tuned to perceive fitness payoffs directly always outcompeted "truth-seeing" organisms, which he says never survived more than a generation. Pushed by Fridman on whether this makes evolutionary game theory itself a possibly-flawed "interface" for reasoning about reality, Hoffman concedes the point but argues it remains the best formal tool available, and cites follow-up work (responding to a Yale team's pushback using thousands of payoff functions) showing that once organisms are allowed to construct "objects" via hierarchical clustering of similar payoffs, the truth-favoring result reappears — reinforcing his view that even everyday objects (like a bottle) are functional data structures encoding possible actions and payoffs, not perceptions of mind-independent things.
He connects this evolutionary argument to an independent claim from fundamental physics — citing physicists Nima Arkani-Hamed, Ed Witten, and others — that spacetime itself is "doomed": not because it is useless, but because it cannot be fundamental, a conclusion Hoffman says follows as a near-theorem once general relativity and quantum field theory are combined (spacetime becomes operationally meaningless below roughly 10⁻³³ cm and 10⁻⁴³ seconds). He describes recent physics discoveries like the "amplituhedron" and "cosmological polytope" (built on Park–Taylor-style scattering-amplitude formulas that collapsed billion-term calculations into single expressions) as evidence that letting go of spacetime's locality and unitarity assumptions reveals deeper mathematical symmetries — while stressing any deeper theory must still "project" back into ordinary spacetime and recover everything currently known, including evolution by natural selection itself as an artifact of that projection. He explicitly distinguishes reductionism (the methodological claim that smaller spatial/temporal scales are more fundamental — useful for things like thermodynamics, but which he calls now clearly false) from the broader scientific project of finding deep unifying principles (e.g., Einstein's postulates), arguing only the former, not science's underlying rigor or the value of experimental testing, is what's being overturned.
On the limits of knowledge, Hoffman invokes Gödel's incompleteness theorem to argue no scientific theory — even ones vastly deeper than current physics — can ever be a complete "theory of everything," since every sufficiently rich formal system will have unprovable true statements; he frames this as a source of enduring "job security" for science rather than grounds for skepticism about experimentation, since any theory still must be empirically testable and must "project" correctly into what's already known.
Turning to consciousness, Hoffman argues that essentially all mainstream theories (integrated information theory, orchestrated microtubule collapse, global workspace theory, panpsychism) share the flawed assumption that spacetime and physical matter are fundamental, and challenges proponents to name even one specific conscious experience (e.g., the taste of chocolate) that any such theory can actually derive from physical dynamics — noting even sympathetic thinkers like Steven Pinker have conceded some "brute fact" must eventually be stipulated. Hoffman's own proposed alternative ("conscious realism," which he distinguishes from Berkeleyan idealism to avoid anti-realist connotations) inverts the standard approach: rather than starting from physical/neural substrates and trying to derive conscious experience, he starts from a minimal formal model of "conscious agents" (a probability space of possible experiences plus a Markovian rule governing how experiences trigger further experiences) and aims to show how spacetime, particles, and brains emerge as a "projection" from networks of such agents — noting mathematically that any two interacting conscious agents jointly satisfy the same definition, suggesting (speculatively, in his framing) a single underlying "one conscious agent" of which individual human minds are localized projections. He stresses this framework does not build in notions of self, agency, learning, or memory as fundamentals — these would have to be constructed from conscious-agent networks (which he notes are trivially computationally universal, while also permitting non-computable dynamics) — and that human consciousness would be "a tiny drop in the bucket" within a vastly larger space of possible, potentially incomprehensibly alien forms of conscious experience (illustrating this with the real phenomenon of human tetrachromats who apparently experience additional color dimensions unavailable to trichromats).
Hoffman connects this framework to Bell-test results in quantum mechanics (via Chris Fuchs's QBism), arguing experiments showing that measured quantum values did not exist prior to measurement empirically refute Einstein's "local realism" — which he takes as independent physical confirmation that observation is "an act of fact creation," consistent with treating consciousness rather than a pre-existing physical world as fundamental. Asked about practical and personal implications, he describes the view as radically reframing attachment, possessions, and identity (a car or house being "ephemeral," recreated only while perceived, much like objects in VR), draws deliberate but qualified parallels to spiritual traditions' teachings on ephemerality and a "deeper self," and states plainly that he is not endorsing any specific tradition wholesale, noting they contain both insight and "nonsense." He discloses that he nearly died the previous year (a 36-hour, 190-bpm cardiac episode during which he texted his wife a goodbye message), describing this as forcing an emotionally, not just intellectually, difficult confrontation with his own theory — noting he meditates several hours daily and still finds his "default mode" reverting to ordinary physicalist attachment unless he deliberately intervenes.
Excerpts
On the core evolutionary argument that perception hides rather than reveals truth:
“Evolution by natural selection does not maximize for truth, it maximizes for fitness... they're there to guide adaptive behavior, full stop. In fact, the way I think about it is they're there to hide the truth, because the truth is too complicated — just like if you're trying to write an email, what you're doing is toggling voltages in the computer, but good luck trying to do it that way.”
On why he considers spacetime "doomed" but not simply wrong:
“What is doomed in the sense that it's wrong is reductionism, which is saying spacetime is fundamental... [spacetime is] absolutely useful and brilliant data structure, but there are deeper data structures... size doesn't matter — you get down to quarks and gluons, what you realize is what you've gotten down to is not fundamental reality, just the irreducible representations of a data structure.”
On the personal, emotional cost of taking his own theory seriously after nearly dying:
“My mind is doing all the science and coming with these conclusions and the rest of me is not integrating... I'm completely attached to my body, attached to the headset, attached to my car, attached to people, attached to all of it — and yet I know as an absolute fact I'm going to walk away from all of it. I'm going to die.”