Interests:
Phenomenology, subjective experience, and qualia;
Anthropics, cosmology, and quantum foundations;
The physics of intelligence, and more broadly of optimality and inference;
Metaethics and metaphilosophy;
Animal and digital welfare, and how we can secure and improve them;
Too many other things to enumerate here...
Contact:
Substack: https://mariven.substack.com
Website: https://n.cohomology.group
Oh, sorry about that… Thanks for the feedback, though, I’ll consider this in the future
There’s nothing special about humans, save for their capability to cause a technological singularity, which is what is being selected for here. This does rule out the overwhelming majority of non-human life forms, though: turtles, for instance, are clearly not going to construct an artificial superintelligence any time soon. They can’t, because they don’t have the neurological means to construct and transmit the encyclopedias of knowledge that are required to build one, or the physiological means to achieve the industrial capacity required to manufacture the programmable computational substrates that are required to build one. They aren’t capable of technological evolution.
(The human form in particular is especially suitable for technological evolution—we are social animals with a fully general communications channel in the form of speech and brains gigantic enough to make use of this generality, which allows us to form, encode, and transmit abstract ideas across societies and generations; we stand upright and have incredible visual acuity, which makes it incredibly easy to construct and use tools; we have very long lifespans and few natural predators, which allows individual humans to undergo profound and prolonged intellectual development; and so on. So conditioning on the presence of a singularity dramatically amplifies human experiences in particular. It could well be that there are other possible histories in which turtles evolved into weird reptoids with similar features, that were simply not sampled).
And, well, maybe you are only important insofar as your experience is required to compute something that’ll have a causal effect on a more important distant descendant of yours. If the computing Thing is not so aggressive in its compression, these are the kinds of things that would be worth following backwards in order to figure out their generative mechanisms and general statistics and so on. If it’s more aggressive, it might not consider this effect to be worth trying to generate in detail, in which case it just asserts (as in Fn. 3) that the specific details of its history are such that the outcome on the distant descendant ends up being xyz.
Yes, oops—whether speech is a ‘lie’ in this context should definitely depend on the speaker’s world model, not the facts of the world itself. I probably accidentally cut this dependence out during a rewrite and then missed it b/c of editorial blindness.