Due to constant MWI branching, delayed observation produces more exact copies which don’t know yet the result of the coin toss.
Applying Self-Sampling Assumption (SSA) to all exact copies, we get that I am more likely to be in the wining branch.
This isn’t how you should be doing anthropics in MWI though. You need to weigh each copy by the Born probability of its branch, which will make this effect disappear. It all adds up to normality!
As another outside observer I also got the impression that the Duncan conflict was the most significant of the ones leading up to the ban, since he wrote a giant post advocating for banning Said, left the site in a huff shortly thereafter, and seems to be the main example of a top contributor by your lights who said they didn’t post due to Said.
I’ve always liked Adam Cadre’s website as a mix of blogging and longer-form content stuff.
Well that would be a rather unnatural conspiracy! IMO you can basically think of law, property rights etc. as being about people getting together to make agreements for their mutual benefit, which can be in the form of ganging up on some subgroup depending on how natural of a Schelling point it is to do that, how well the victims can coordinate, etc. “AIs ganging up on humans” does actually seem like a relatively natural Schelling point where the victims would be pretty unable to respond? Especially if there are systematic differences between the values of a typical human and typical AI, which would make ganging up more attractive. These Schelling points also can arise in periods of turbulence where one system is replaced by another, e.g. colonialism, the industrial revolution. It seems plausible that AIs coming to power will feature such changes(unless you think property rights and capitalism as devised by humans are the optimum of methods of coordination devisable by AIs?)
which is far, far safer and easier to coordinate than trying to completely disempower all non-lawyers and take everything from them
But it would probably be a lot less dangerous if lawyers outnumbered non-lawyers by several million, were much smarter, thought faster, had military supremacy, etc. etc. etc.
The truth is, capitalism and property rights has existed for 5000 years and has been fairly robust to about 5 orders of magnitude increase in population
During which time many less-powerful human and non-human populations were in fact destroyed or substantially harmed and disempowered by the people who did well at that system?
I haven’t fully read through your paper, but from the parts I have read sounds like it might be similar to the neural tangent kernel applied to the case of ReLU networks
OK I see, didn’t get the connection there.
humanity has a bad track record at that
People do devote some effort to things like preserving endangered species, things of historical significance that are no longer immediately useful, etc. If AIs devoted a similar fraction of their resources to humans that would be enough to preserve our existence.
So why would AIs be more willing to do that?
He spells out possible reasons in the paragraph immediately following your quote: “Pretraining of LLMs on human data or weakly successful efforts at value alignment might plausibly seed a level of value alignment that’s comparable to how humans likely wouldn’t hypothetically want to let an already existing sapient octopus civilization go extinct”. If you disagree you should respond to those. Most people on LW are already aware that ASIs would need some positive motivation to preserve human existence.
The default on Polymarket is no expiration. It is only after actively choosing to have an expiration date that it goes to 1 day
Hmmm you’re right, guess I misremembered.
Polymarket and manifold already have optional expiration dates, and Polymarket suggests a default expiration of “end of today”.
But these dynamics aren’t very scalable. Also I dunno if “prediction markets” is a very precise analogy, I think the dynamics are better than PMs in some ways and worse in others.
If you care about reality, that’s your prerogative.
Well, the notion of “objective reality” could itself be confused, an incoherent thing to care about. Analogously, you might initially care about carrying out the will of God but later become an atheist. You need to somehow refactor your goals to match your new ontology.
You could consider the totality of mathematical truths to be “reality” if you want. Or if you mean you care about “the real world, which I am actually in”, then this perspective tries to dissolve that notion.
This is one of the better sentences-that-sound-bizarre-without-context I’ve seen in a while.
You’d have a random mix of both (infinite?) alien simulations (both broken and unbroken. But you’ll be at their complete mercy, if they want to torture you for fun they can) and infinite disembodied reincarnations of a few yottoseconds
I think the probabilities will be heavily tilted towards the aliens(or really, humanity/human-descended AIs in the future or an alternate timeline). But yes, I agree with the overall point that continuations of your stream of consciousness might become very weird and unpredictable past the point at which you would normally die on Earth(and probably even weirder as time goes on into the distant future)
Then we have some margin for error in some unimportant parts and have absolutely zero margin for error in the other parts
But all of the parts would have some margin for error because bounded physical processes can’t store infinite information(Bekenstein bound). Adding in your guts, microbiome etc. doesn’t fundamentally change the argument, they still have a far smaller number of degrees of freedom than your past light cone.
If you think aliens could recreate you by using enough random coin tosses to get to you, then you’re only recreating a Boltzman brain but unlike the universe aliens do not have infinite ressources
Flipping a quantum coin a bunch of times doesn’t actually require that many resources though, assuming you already have enough computational resources to store the brain. And aliens have a crucial advantage over vacuum fluctuations in that they can model dependencies that would be found in actual brains and only sample from brain states that would likely occur. Like if (made up numbers) there are 2^(10^15) possible brain states but only 2^(10^9) of them would be likely to occur in reality, the aliens can sample from that narrower space and thus get a relative factor of ~2^(10^15) more probability. (Yes aliens can’t perfectly sample from the actual state space but it doesn’t matter, any improvement over pure randomness will give a large increase in probability)
They’d need not only to simulate quantum states, they’d need the whole list of all quantum events and their results in the spacetime they are simulating in order to reproduce you perfectly
Not really though. Your brain has ~10^15 synapses. It literally doesn’t have enough space to store records of its entire past light cone. So there is actually some equivalence class of possible past histories which could produce an identical brain state to the one you have now.
there is no conceptual way they could simulate you particularly
They definitely could, by just randomly sampling human brains they would have some low probability of picking out yours or mine in particular. With a randomized simulation they could pick us out with much higher probability.
In order to simulate a particular individual, it would require far more than merely simulating the species homo sapiens[...]
Yeah but the point is, for however many unlikely coincidences/”coin flips” the aliens would need to get right, the Boltzmann fluctuations would have to get at least that many coin flips right, plus more because they aren’t modeling any deterministic parts of the process.
I mean, sportsbooks can shut down your account if you’re too consistently successful so that seems like a major downside for “serious gamblers”. This is also a way that they’re pretty blatantly more predatory than PMs.