I. Better SETI through qualia
My general thesis for SETI (Search for Extraterrestrial Intelligence — looking for alien signals in the sky) has been that anything we can infer about the likely telos of alien civilizations will greatly help us search for them. If we understand what intelligent civilizations are likely to do, we can specifically look for evidence of them doing this.
I’ve long thought qualia research can help here:
Premise 1: Eventually, civilizations progress until they can engage in megascale engineering: Dyson spheres, etc.
Premise 2: Consciousness is the home of value: Disneyland with no children is valueless.
Premise 2.1: Over the long term we should expect at least some civilizations to fall into the attractor of treating consciousness as their intrinsic optimization target.
Premise 3: There will be convergence that some qualia are intrinsically valuable, and what sorts of qualia are such.
Conjecture: A key heuristic for discerning the presence of advanced alien civilizations will be searching for megascale objects which optimize the production of intrinsically valuable qualia.
What could such “megascale objects which optimize the production of intrinsically valuable qualia” be? Dyson spheres are a good generic bet, but we’re already looking for them. Originally, based on a confluence of factors including the Symmetry Theory of Valence, the scales of energy, and the likely physical homogeneities involved, I suspected black holes, quasars, and pulsars might generate large amounts of intrinsically valuable qualia. I still do. But maybe we can add massive proof-of-work (PoW) blockchains.
II. The blockchain-as-universal-megastructure argument
My friend Dhruv Bansal of Unchained Capital has a lovely series on how something like Bitcoin might evolve when forced to integrate the constraints of interplanetary civilizations:
Part I / Law of Hash Horizons discusses issues around blocktime and the speed of light: PoW blockchains like Bitcoin will have a physical “hash horizon”, outside of which it will be possible to spend Bitcoin, but not mine it. A core prediction is that Mars will have its own cryptocurrency (“Muskcoin”), because Mars is usually outside of this horizon, and wouldn’t want to cede its financial sovereignty or the economic rewards from mining cryptocurrencies to Earth.
Part II / Hash Exclusion Principle discusses different temporal niches for PoW blockchains, in particular how quick settlement chains and slow settlement chains will coexist when dealing with interplanetary distances. Quick settlement chains preserve local autonomy; slow settlement chains allow larger coalitions. A core prediction is the rise of a very-slow-settlement chain (“Solcoin”) which offers neutral ground for miners across the entire solar system. Another significant prediction is that PoW chains incentivize energy harvesting on a massive scale, and may sometimes be a significant factor in civilizations successfully bootstrapping to Kardashev II & III.
Part III / Law of Hash Universality discusses blockchains as a universal in any non-hive mind civilization, something that neatly solves certain classes of coordination problems and will be as common among alien civilizations as joint-stock corporations, maps, and double-entry accounting. A core prediction is the first signal we receive from aliens could plausibly be an invitation to join their blockchain.
I unironically believe Dhruv’s work may be the most significant development in SETI in the last decade. It’s also really fun to read. I’m not fully convinced future blockchains will be PoW, though there are serious arguments to this effect and PoW being a Kardashev bootstrapping mechanism is compelling.
But if we take Dhruv’s arguments seriously, I think we can push further and say something interesting about the particular PoW algorithms likely used by alien blockchains.
III. Computational theories of consciousness: impossible in theory, possible in practice
I believe consciousness lives in the physical — if we wish to understand whether something is conscious we need to look at what its atoms are doing, not its bits. People talking about AI consciousness seem confused for (at least) two reasons: (1) defining AIs in terms of atoms is really messy and anything that doesn’t resolve to atoms isn’t ‘real’ in some interesting sense- “Is this LaMDA in the room with us right now, anon?”; and (2) there’s no objective fact of the matter about which computations any physical system is performing, though there are sometimes useful conventions. This argument applies to something like a waterfall, but it also applies to that A15 chip in your phone. Given one physical system, there are an infinite number of computations it could be performing; given one computation, there are an infinite number of physical systems that could implement it. I believe this is a fatal objection to (Turing-level) computational theories of consciousness, as discussed here.
But the underdetermination (many-to-many mapping) that stops computational approaches to consciousness breaks down when we consider extreme performance or adversarial computing. Essentially: there will always be a single most efficient physical way to compute something — for any given computing task, there will always exist some arrangements of atoms that is the optimal solution for this task. (I’ll claim that this is true for both classical and quantum computing). In the performant limit case, the desired computation is sufficient to specify the optimal physical system; the physical system plus the assumption of optimality is sufficient to infer the computation.
Let’s call this the “optimal atomic configuration thesis”. If we assume OACT holds with top-tier PoW blockchains such as that used by alien civilizations (Dhruv helpfully offers “Xenocoin”), this lets us claim that, at the limit, all economically competitive miners of Xenocoin will be using the same core hardware, and we could infer the molecular logic of this hardware if we knew the PoW algorithm.
IV. Proof-of-work has degrees of freedom
I think one *incomplete* piece in today’s proof-of-work paradigm is determining what the work ‘should’ be. Presently, the assumption is that for proper security, the ‘work’ should have no external value, otherwise ‘double dipping’ could lead to weird incentives and bodies of cached work and game theory that would sabotage the security of the chain. This seems right to me. There have been a few attempts at doing something useful with mining (e.g. factoring primes) but all mainstream PoW algorithms are intentionally arbitrary. “Your PoW algorithm can be anything, as long as it’s sufficiently and predictably difficult and provably useless.” Mining Bitcoin involves endlessly computing SHA256 hashes.
But PoW only has to be provably useless in a relatively narrow technical sense. If PoW happens to have some positive externality, like generating heat to warm your home, that can be a feature. Presumably we should search for PoW algorithms with positive externalities, as long as they don’t compromise security.
I’m going to suggest that *generation of qualia* could be such an externality. The theories of consciousness that I think could grow into a solution (e.g. strong monism) hold that all physical processes have corresponding qualia processes. When you do something in the physical domain, something happens in the qualia domain (though I’d note Aaronson’s caveats about quantum computers, qualia, and reversibility). Loosely speaking and with a nod to qualia fraggers, we can assume matching complexity between domains. Rocks probably don’t do anything interesting in the qualia domain; ‘what it feels like to be a rock’ is probably simple. I believe the qualia of black holes, pulsars, and quasars are more interesting. But mining crypto at the scale of solar systems using nanoscale quantum computers — we can be pretty sure something *very* interesting is going on in the qualia domain. This is no formal argument, but it feels it would require some intellectual contortions to claim that megascale crypto mining *wouldn’t* generate a lot of qualia.
And if you’re a Kardashev II-III civilization you’re going to understand the OACT, and you’re probably going to understand consciousness. You might even care about consciousness as a domain of optimization; if you do, you’ll probably chain these understandings together. And so if you’re going to be creating a PoW algorithm and recruiting your galactic neighborhood to terraform their star systems to create complex nanostructures that mine your coin *anyway*, you might choose a PoW algorithm that will create positive qualia when implemented by its molecular-optimal ASIC. I.e.: in sufficiently large-stakes PoW, by defining the class of work to be done, you define the class of qualia to be made.
The optimal class of qualia we can punt for another time; the current observation is simply that qualia will vary by class of computation, and some classes are more intrinsically valuable than others. This may introduce new constraints and vulnerabilities into what could be classified as a “safe” PoW scheme, but I believe also allows us some real choice over what qualia a proof-of-work scheme ‘mines’.
Are qualia aesthetics convergent? STV would loosely suggest yes. But just because something is convergent at the limit doesn’t mean it has to be at any specific point; I could see certain paths where humans develop a galactic PoW algorithm that implements the phenomenology of being rickrolled when implemented on its optimal molecular substrate.
An observation I made to Dhruv on the distance between Earth and Mars oscillating between being inside vs outside Bitcoin’s hash horizon:
While reading I was thinking it might be possible for Mars to (permanently?) steal the Bitcoin center of hash. Something like: build a ton of Bitcoin ASICs and park them in orbit around Mars. When Mars and Earth are closest (3 light minutes), turn them on and mine like crazy. Hopefully you can muscle the center of hash away from Earth; as Mars pulls away, your defender advantage should become bigger and bigger. You’ve successfully moved the center of hash (hashdragging?). Earth can spend but not mine.
Now what to do? What goes around comes around and Earth can just “hashdrag” you next cycle. But maybe you take your orbital fleet of Bitcoin ASICs (still broadcasting its solutions) and move them out of Mars orbit. […]
Our solar system is interesting in that the two habitable planets, Earth and Mars, oscillate between being fairly close (relative to Bitcoin’s blocktime) to not being close. Might incentivize shenanigans.
From What’s out there?:
>I’d offer there are four main classes of qualia in the universe:
>I. Evolved Qualia – e.g., humans and other free-energy-minimizing-evolved-systems. These will be characterized by intentional content, predictable dynamics, stable-ish boundaries, often with the behavioral hallmarks of agency and the qualia of free will. ‘Qualia agents’.
>II. Primordial Qualia – e.g., quantum fuzz. The small-scale, primordial ‘soup’ of mostly-not-bound-together flashes of simple qualia-information. ‘Qualia dust’.
>III. Megascale Qualia – e.g., black holes, quasars, stars, planetary cores. These will be characterized by stable-ish boundaries, highly predictable dynamics, likely no intentional content, but possibly significant binding. ‘Qualia (mega)crystals’.
>IV. Technological Qualia –
- >IVa: Qualia Fragments, aka ‘qualia fraggers’ – technological artifacts created for some instrumental functional purpose, e.g. digital computers. A key lens I would offer is that the functional boundary of our brain and the phenomenological boundary of our mind overlap fairly tightly, and this may not be the case with artificial technological artifacts. And so artifacts created for functional purposes seem likely to result in unstable phenomenological boundaries, unpredictable qualia dynamics and likely no intentional content or phenomenology of agency, but also ‘flashes’ or ‘peaks’ of high order, unlike primordial qualia. We might think of these as producing ‘qualia gravel’ of very uneven size (mostly small, sometimes large, odd contents very unlike human qualia).
- >IVb: Engineered Qualia – technological artifacts created for the production, optimization, or computation of qualia,
These ideas partly based on some ~July 2017 unpublished notes on how future quantum computing compilers could optimize algorithms for phenomenological valence, much as e.g. the LLVM compiler can optimize for memory usage.
A friend jpt4 comments:
> Given one physical system, there are an infinite number of computations it could be performing; given one computation, there are an infinite number of physical systems that could implement it.
> for any given computing task, there will always exist some arrangements of atoms that is the optimal solution for this task
These both revolve around the question of the realizability of a normal form . Regarding the waterfall, what Aaronson  elucidates is that it is not the capacity for representation which bears causal weight, but that for reduction, or equivalently, compression. Thus, viewing reduction as a resource, or reciprocally, realizability as a cost, the representation of any particular computation can be trapped inutile within an instantiation, absent any means of its extraction/interaction.
This is independent of whether a normal form is guaranteed to exist, or whether it is guaranteed to be confluent, because  1) if something extends the realization of the normal form from the immanent to the removed, then until it is realized one is in a different domain of optimization 2) it is not in general known a priori whether a normal form exists, or how to realize it (or any properties of its realizability, e.g. the bulk tally of reduction resources required).
Nevertheless, approximately optimal normal forms are often sufficient for the spacetime scales under discussion, and we can proceed to the latter section of the musings while bracketing the above.
To which, while I agree with the general principle that there should be convergence in megastructures, with regards to crypto in particular, any side effect of a cryptographic process is a sidechannel vulnerability. If dyson miners have qualia, then those qualia become targets for hostage taking. Isentropic/reversible computing  is the best model for maximal security.
If the universe is sufficiently predatory, qualia-tative megastructures will exist only as passive ruins, or during the brief hegemonically active periods of creators. This is the same issue which we encounter on Earth-scales now, with our occulted elites, who have learned that stealth is the mortal’s shield against death (when the kinder eras of the pre-missile past might have supported grander delusions that overawing majesty was sufficient).
Either we need grandeur in stealth, or an actual up-to-epsilon-omnipotent hegemon, if the trilemma of 1. grand 2. conscious 3. constructs  is to be resolved.
 The following two criteria definitely hold for Turing Complete phenomena, but I think analogues also apply for most of the sub-Turing space as well. Willard’s SJAS is a very narrow sliver where some of this is bypassed.
 I.e., dependent on creators, cannot regenerate/defend themselves; subpolitical.