Watched Cronenberg's Dead Ringers and eXistenZ with @jessi_cata and @umbersorrow, and I understand much better how from a monotheistic perspective atheism can look like nihilism. π§΅
Speech - and reference more generally - emerges for and is learned in the context of intersubjective thinking, not private reason. benjaminrosshoffman.com/language-powerβ¦
Most people have profoundly limited independent ability to participate in a referential system - they mimic speech as languageless barbarians. benjaminrosshoffman.com/actors-and-scrβ¦
Religion is a way to preserve a system of shared reference under such conditions. Even if your parents past a point can't relate to you, you can relate to an abstract personality who is always present, and has full linguistic capacity.
The attachment patterns that constitute religion embed in language a strong prior that one lives in a world ruled by a god or gods, i.e. by a one or a few very powerful autonomous persons with intentions.
Under monotheism one can represent an integrated model of the world and human action by relating specific facts one knows to the imagined mind of God who knows and judges all, even in the absence of anyone live human to compare notes with.
Darren Aronofsky's and Woody Allen's movies show people with monotheistic assumptions trying to orient themselves in a world that empirically doesn't seem to contain the kind of God embedded in their priors.
If your orientation to descriptive language assumes a divine listener, becoming a true atheist can't mean representing God's failure to exist in your preexisting system of reference. You would have to reconfigure your whole system of reference not to assume God's existence.
Cronenberg's protagonists have no relationship to an abstract reference personality. They can relate to one or two other people with a comparatively full range; with everyone else, they can have stereotyped social interactions but cannot reconcile perspectives at all.
They live in an environment but not a world; there is no sense of an ordered cosmos in which they're located, but instead the environments the protagonists have shared knowledge of are structured and well-specified. The rest of their experience merely has recognizable patterns.
It was difficult for me to distinguish this from the nihilism of contemporary action movies in which we see characters as stereotyped personas going through a series of stereotyped situations about which we are asked to have stereotyped feelings.
The difference is that while action personas have no integrated perspective, can't be understood and predicted by empathizing with their beliefs and motivations, Cronenberg's characters do have local perspectives, just not cosmic ones.
Careful use of anthropic reasoning, Solomonoff induction, and the simulation hypothesis can provide a frame within which these hypotheses can be compared.
Solomonoff induction rigorously integrates Bayesian inference with a simplicity prior. lesswrong.com/posts/Kyc5dFDzβ¦
A Cronenberg protagonist who learned enough science, computer and natural, would update towards a synoptic perspective encompassing all accessible environments as part of a single order.
On the anthropic principle, an embedded reasoner ought to update in favor of hypotheses that explain their own existence. How do we explain this in terms of Solomonoff induction? True Solomonoff inductors are not computable but we can think about computable approximations.
The computable approximation of a Solomonoff inductor will only be able to compare a finite range of representations of reality, not a countably infinite generators of data streams that can be compared exactly with its experience. Let's call this computable approximation Solomon.
Naturalized hypotheses are programs that embed (a low-resolution representation of) Solomon as a computational module. All other programs under consideration would have to directly describe Solomon's experiences. These are Boltzmann Brain hypotheses.
Boltzmann Brain hypotheses can be discarded; they are decision-theoretically irrelevant, and if specified precisely would have much higher message lengths than hypotheses can compress Solomon's representations of reality.
But the philosophical and compression advantage of self-describing hypotheses assumes the existence of memory. If there's not a stream of past events to explain but only the present to describe, you might prefer pragmatic compressions that don't contain an agent compressing them.
Instantaneous induction may be seriously comparing naturalized and non-naturalized hypotheses, but reasoning over time about what is true implies memory, which overwhelmingly favors naturalized hypotheses.
This is the computational basis for anthropics, i.e. favoring naturalized over non-naturalized hypotheses.
The Self-Sampling Assumption discredits programs that do not represent Solomon within what is otherwise a naive approximation of Solomonoff induction.
The Self-Indication Assumption adjusts its probability distribution by multiplying the prior weight of each hypothesis by the number of times Solomon is represented within that hypothesis. The resulting probabilities are more convenient for some expected-value calculations.
Self-Indication Assumption (SIA) leads to the anthropic trilemma:
IF future humans will have control over a huge amount of machine computation, AND they will be interested even slightly in simulating perspectives that think they're animals, THEN most perspectives that think they're animals are simulated.
A naturalized hypothesis should explain not just why there's an optimizer, but why that optimizer can represent its environment. Who is there for Solomon to talk with?
If you're starting with the strong assumption that everything is represented to a single strong ally who holds you accountable, you might get far enough to learn practically advantageous information from texts written by people with unusual insight, even surrounded by barbarians.
Woody Allen's movies take the perspective that I worked hard to represent the world, under the assumption that the world was just and keeping commitments was rewarded. But in that representation of the world, the world is frequently unjust and keeping commitments is punished!
Crimes and Misdemeanors in particular follows two characters facing this realization, one of whom desperately looks for cleverer ways to fool himself, the other of whom uses his capacity for lookahead to sociopathically outmaneuver others for narrowly selfish ends.
But there's no alternative theory of education; parents avoid informing their children so their children orient towards participating in a just world their parents know does not exist.
There is an interesting mind-control analogue to the simulation hypothesis; quite plausibly most people who think they have someone to talk with are embedded in a simulation specifically designed to create that false impression.
At the peak of American power in 1999, The Matrix, Columbine, Fight Club, Office Space, and American Beauty, which seem to be different versions of the same movie, addressed the breakdown of the simulation directly.
They all agreed that we are contained by a simulation in ways that prevent us from authentic contact with reality and lead to a lot of fake work / bullshit jobs, and proposed different responses.
Office Space depicted an office worker constantly harassed by partially imaginary threats, until rescued by a wizard. It proposed a basically Taoist solution of non-neurotic nonacquisitive local opportunism, flowing around conflict using Tai Chi.
The protagonist makes some important mistakes because he is legitimately confused about which barriers are real. He ends up with a better life but fails to rescue his friends.
The Matrix proposed psychedelics + maintaining a physically secure location + violent conflict (kung fu + guerilla special-forces combat) against suits to jailbreak people who are starting to disbelieve.
Fight Club proposed the bombing of the simulation's record-keeping apparatus. The Columbine Massacre was basically a scream of despair by people who wanted to die and take down their abusers. I haven't seen American Beauty.
eXistenZ came out the same year, and addresses simulations, but differs from these other movies because despite handwaving to the contrary it doesn't posit a canonical simulation, just specific environments simulated in specific contexts for specific targets.
One plausible response to the simulation hypothesis is that large simulations with neutral substrate are computationally equivalent to large worlds (so it doesn't matter much if we're in one).
Our behavior in small, narrowly constrained simulations are mostly decision-theoretically relevant as ways our behavior might be predicted in larger worlds, not for their own sake. This is analogous to writing off Boltzmann brain hypotheses.
The analogous solution with respect to simulations like monotheism is to notice that something structurally similar to us must have engineered the simulation in the first place, and try and take that thing's perspective.
This is why I have special interest in Jaynesian Bronze-Age mind control, the Bible and similar cultural source documentation, cybernetics, ideology, and integrated ecological and computational interpretations of civilization.
@threadreaderapp unroll
Simple example of using theism to save a more complete expressive range than might otherwise exist
https://twitter.com/wordpaley/status/1399504402562621443
β’ β’ β’
Missing some Tweet in this thread? You can try to
force a refresh
γ
