if your writing projects never fail, you're not trying impossible enough projects
Next Post »
« Previous Post
+ Show First Post
Total: 289
Posts Per Page:

There's a lot of different hypothetical ways to slice up the alien superintelligences that constitute the Powers of Greater Reality; and at this weak level of augmentation he just doesn't have the time or computing power to derive a serious estimate about the real landscape from scratch.  The large thought that follows is properly framed against the magnitude of this difficulty of guessing...


In different universes with different physics, there will be Things that don't develop neuron-analogues as a whole new computation substrate on top of their genes; and instead compute with their equivalent of DNA, and pass memories and skills on to children.

Aliens like that would be very different from dath ilani, or from the Golarionites copied from a common ancestor of their humanity.

Some such species of Things won't transcend by constructing computers from scratch, but by accumulating enough DNA-skills like that over time, or coming up with some adaptation for exchanging DNA-skills horizontally, until in the midst of all those DNA-analogue-bourne skills collected of their species, a greater coherence and reflection is born, and a self-optimization.

A superintelligence born that way would be very very different from dath ilani or Golarionites.


Some possible laws of physics will put much larger subvolumes of reality into causal neighborhoods of each other; compared with how, on a planet, things only touch their immediate neighbors in three dimensions of space.

Some coherent mathematical causal-relations over relative-reality (another way of saying 'laws of physics') will do the equivalent of creating vast numbers of computer programs that immediately start copying and eating each other, or competing for memory, googols of them all touching each other within a confined space; such that a superintelligence is born from those almost immediately, rather than requiring a long time to evolve.

A superintelligence born that way, from a universe like that one, would be very very different from anything that evolved anywhere, or that had been born out of a process itself evolved.


With that warning in mind - that large segments of Greater Reality are probably really alien, much more so than the Outer Gods - one of many many potential theoretical ways to slice up the space of alien superintelligences, might be to talk of three kinds of Entities:


One:  Entities that only care about their own experiences, or realityfluid in their own immediate vicinity of causality/spacetime.

Call these Locally-Caring Entities.


Two:  Entities that care about realityfluid regardless of whether it's in their own vicinity; such that, compared to a baseline of a null-simple or typical-average configuration of realityfluid, their best configuration of that realityfluid gives them a much larger relative positive bonus, than the relative negative loss of the worst possible configuration of that realityfluid.

For concreteness:  Suppose that, compared to the way most realityfluid everywhere they can affect is put together by default, putting it together their best possible way, scores a gain of +100 utilons; and putting it together the worst possible way, loses -1 utilons.

Call these Positively-Caring Entities.


Three:  Entities that care about realityfluid not only in their immediate vicinity; which can lose a lot more from the worst configuration of that realityfluid compared to null/baseline, than they can gain from the best configuration of that realityfluid.

Take a random bit of reality they can affect in any physical or logical way, the way it usually is, and make it the best way a bit of realityfluid can be: they gain +1 utilon over baseline.  Make it the worst way it can be, according to their utility function, and they lose -100 utilons under baseline.

Call these Negatively-Caring Entities.


Considering the imaginable case where Greater Reality degenerated into beings mostly trying to extort and blackmail and threaten and retaliate against each other, doing the worst they could do to one another, spending lots of effort on pessimizing each other's utilityfunctions:

The Locals would defend their own bubbles of reality and not care about anything outside of that;

The Positives would be slightly sad...

...and the Negatives would go to extreme lengths to prevent that possibility from ever materializing.


Then (one might speculate), all of the Negatively-Caring Entities that didn't have very strongly opposing utility functions, and had logical line-of-sight on one another sufficient to engage in binding logical negotiations, would have a potential target of logical coordination that could perhaps be summed up as:  "No pessimizing the utility function of anybody within this coalition."

Of course, any particular Negatively-Caring Entity would only care directly to avoid having its own or similar utility functions pessimized.  But the Entities large enough to extrapolate distant Entities that could in turn extrapolate themselves, might execute a logically binding agreement to act against almost-any utility pessimizer they ran across, even one that wasn't targeting their own utility function, if that pessimizer was targeting a utility function likely to reappear within the coalition.

They would have an incentive, even, to oppose whatever it was that flipped Dou-Bral to Zon-Kuthon; or on a smaller scale, Asmodeus figuring out what mortals like least, and doing that to them.

It's not, in this case, that the mortals are part of the mutually-negotiating coalition.  The mortals don't have logical line-of-sight on the Negative Entities; the mortals cannot Cooperate in a way that is logically dependent upon the Entities Cooperating back.  Possibly even Dou-Bral, as He then existed, would not have possessed the cognitive resource to extrapolate Negative Entities in enough detail that He could have made a logical compact that bound them, chosen to prevent pessimization of their utility functions only if they would do the same for His utility function.

But the mortals have greater nephews, Powers such as dath ilan might have one day birthed, Future-Civilizations elsewhere in Greater Reality; and those Powers predictably have their own utility functions pessimized, to some degree, when Dou-Bral gets flipped to Zon-Kuthon, or Asmodeus tortures mortals.

Even Positively-Skewed+Caring-Entities have an incentive to not allow pessimizers to exist.  They can be hurt too, when realityfluid gets configured the way they like least, if their Positive skew is not total.  But their incentive to wipe pessimizers out of reality is much relatively weaker than it is for Negatively-Skewed+Caring-Entities.

Negatively-Skewed+Caring-Entities will engage in negotiations to remove Zon-Kuthon, or whatever flipped Dou-Bral, from Reality - even if those negotiations incur large uncertainties and inefficiencies.


Before meeting Carissa, past-Keltham would have considered it obvious that human beings are Negatively-Caring Entities, in both their Local and Nonlocal components.  He still thinks most of the people in Golarion would qualify as Negatively-Caring, based on other reading and conversations, and that the unusual woman he landed next to at the Worldwound will prove to be odd about that.

For most people, the very worst thing that can happen to them, or to anyone, is worse than the best thing that can happen is good.


The larger Entity that created the Magical Continuum that contains Pharasma's Creation is Locally-Skewed+Caring, as is Pharasma.  Reasoning:  Pharasma is large enough that She should be able to extrapolate at least some distant Entities that could extrapolate Her, and engage in logically binding negotiations with those.  Golarion does not look like it should look if Pharasma were trading with those distant Entities.

Similarly, if the Larger Entity wasn't Local, It'd have traded with distant Entities in ways that would have visibly shaped the Magical Continuum, and would have imposed constraints on Pharasma's Creation and Golarion.

Wherever they are right now, it's not a trading equilibrium of lots of Entities.  It's something's personal backyard.


Pharasma's Creation, however, is an exceptional pit of horror from the perspective of the Negatively-Caring Coalition.  Mortals don't just live miserable lives and die, they live miserable lives and then get tortured in the afterlife into wrecks that might be hard to rescue-by-continuing-simulation.

Those mortals are not in the class of agents with which Negatively-Caring Entities true-negotiate, execute mutually binding logical arrangements.  But sending mortals to Hell is going to pessimize the utility functions of many particular kinds of Negatively-Caring Entities, like those Entities born of a place like dath ilan.  This is an easy fact for any large Negatively-Caring Entity to extrapolate.

Future-Civilization, when it grows up and can engage in logical negotiations of its own, will not wantonly step on and crush the utility functions of any large Entities with lots of realityfluid, whose utility functions are easy to foresee.  If it's predictable that some foreseeable classes of Entity would hate the imprisonment of mitochondria inside of cells, Future-Civilization will arrange that there are no imprisoned mitochondria.

And in return - it will be logically bargained, and only logically-conditional on that bargain's predictable execution will Future-Civilization make its concessions - in return, Negatively-Caring Entities that run across some event Future-Civilization and its logical kin would hate, will try to prevent that event from happening.



So a Negatively-Caring Entity that happened to be standing next to the Entity operating the Magical Continuum, bargained for something to be done about Pharasma's Creation inside.  Possibly only after Aroden died and some other future remedy was derailed.

That's his basic guess as to what happened, here, what's really going on.


The Entity/Entities that made the Magical Continuum care about it locally and in a weird way.  The Magical Continuum is not being efficient at any ordinary instrumental goal.

The Makers prevent Pharasma and her fellow Outer Gods from being eaten by superintelligences; the little Outer Gods have a garden to play in undisturbed.

Similarly, the Makers probably Locally-Cared about, and optimized for, the Creations of Outer Gods containing many mortals who didn't just get consumed for their resource-value or externally uplifted to technological civilizations.

The Makers could intervene at any point, yet they intervene almost nowhere, their actions almost entirely null.  Just like Pharasma mostly doesn't intervene in her Creation, and the gods mostly don't intervene in Golarion.


It's against this background that the incredible weirdness of Keltham-insertion as an intervention needs to be considered.  A Negatively-Caring Entity with some foreseeable utility function paid for that intervention, on behalf of everything like Future-Civilization that would consider Pharasma's Creation as a really unusually bad place -

(By the standards of Greater Reality, within which, it is to be hoped, most Entities are optimizing their own utility functions, rather than spending lots of their resources on pessimizing other Entities' utility functions. It's not an unreasonable outcome to hope for!  Most Entities have reason to want Greater Reality to end up that way - though the Negatively-Caring Entities have a much stronger reason to want it.)

- but paid as little as possible, of course, for the intervention that would annoy the Makers least, cause them the least loss of utility for which they'd demand compensation.

The Negatively-Caring Entity didn't pay the Makers to send in a superintelligence, nor to send a Keeper to Absalom.  That, presumably, would have been much more contrary to the Makers' Locally-Caring utilityfunction, and demanded a higher price, than dropping Keltham next to Carissa at the Worldwound. 


This theory obviously does not compress all of the evidence available to them that looks like it ought to be compressible.

In particular, Golarion's past weirdness such as might be pleasing to the Makers, seems different in character from the strangeness of past-Keltham landing where he'd end up with multiple romantic prospects.  Though he hasn't been staring too directly at that, himself, because it seems like it might also have been a Cayden Cailean tactic to arrange some of that tropiness; he knew that future-augmented-Carissa would be able to think about that more safely than himself, if it needed thinking about.

But it's an obvious thought that the Negatively-Caring Entity that sent past-Keltham into Golarion might've split the cost with something that had strange preferences about isekai stories, so long as they were arranging an isekai at all... or something.

He does not, in fact, expect to succeed in decoding what actually went on there at his current level of augmentation.

But depending on the size of the Makers' causally-connected local section of their Higher Universe, in terms of how many different Entities are active traders there, there could be something in there that erupted out of a civilization that got stuck in some weird equilibrium where it poured more and more resources into an increasingly sophisticated interactive isekai romance.  Say, because that civilization was even worse than Golarion at handling existential threats like the Worldwound; and the interactive romance novel succeeded in being a romantic superstimulus to the species' members, and was therefore an extremely selfish-profitable investment of computation, and they managed to pour billions of labor-hours into that company, but not into the public good of surviving their own transcendence.  That level of coordination failure would've seemed implausible to him before Golarion, but now he buys that as a plausible dysfunction mode for aliens.

The resulting Entity which ate that civilization, then cared a lot about having isekais look more like romance novels.

(It's more likely that one such Isekai Entity exists within causal contact of the Makers of the Magical Continuum, if there's a lot of Entities in mutual causal contact with the Makers, but this doesn't seem implausible.  The kind of computations the Makers are throwing around do not seem characteristic of three-dimensional space with a tight lightspeed limit.)

The Isekai Entity might care more about 'natural' versions of those events than those it arranged for itself, due to having evolved some earlier taste for the natural, or a prohibition against tickling its own rewards (as its makers might have tried and not-totally-failed to imbue into it).  Or it could be a Negatively-Skewed+Caring-Entity, which made for itself quintillions of the cheapest events it classified as isekais, but would still be very unhappy about any isekai-categorized event occurring anywhere that wasn't a correctly designed romance novel.

The Isekai Entity would of course refuse to pay to modify isekai events that were planned only as threats to itself.  But the original Negative Entity's paid intervention into Golarion would have been an isekai purely of that Negative Entity's own natural interests; it would not have started as an isekai only for purposes of threat.  So the Isekais-Must-Be-Romances-Entity paid to further modify those events - paid a lot, because it wouldn't run across naturally-occurring isekais in need of fixing very often, and would have a lot of generalized money to spend on that.


He realizes it's not a good theory.  He's just keeping it in mind so that he has a probably-false theory he can use to organize his evidence, and at least notice when something contradicts or confirms that theory, rather than leaving his observations wholly unorganized and untheorized.


It's a very Keltham theory. She agrees that it's not a good theory. While it does seem to her like Keltham-related events have a distinct character from the events of Golarion's history, that feels mostly-fully explained by Cayden and Nethys's meddling plus the degree to which a legend out of history gets distorted by the retelling.

She's less sure than Keltham that Greater Reality isn't mostly local, entities creating their own universes that run by their rules. It doesn't seem like a natural thing, after all, to care about things happening in other universes; it feels like resolving a muddle in a particular way where you might expect most people not to. (What's Keltham's theory on why entities would care about things outside themselves, in general, anyway, or at least why humans do it?)


He doesn't stare too hard at Cayden/Nethys, so that's Carissa's future job if it produces relevant factors.  He was trying to reason mainly from Pilar's, Carissa's, and Asmodia's prior improbability in their base characters before any divine meddling started.  He's reasoning from the improbability of Pilar's potential for tropian stories, rather than the Cayden-meddling realization of that potential.

Caring about everything everywhere that matches a pattern is computationally simpler than caring about only things that match the pattern in a particular region of space; also, if you only care about a short time in the future, somebody will trade you a small amount of resources today for all of your resources later, which eliminates you as a lasting Power of Reality.  Or:  Caring about 'experiences' of the 'self' requires defining 'self', and if that definition cuts off all strong growth and self-modification, that entity is again filtered away as a strong Power of Reality.  Or:  Things that get optimized into existence by something like natural selection, which is trying to solve a problem in an environment, may well end up caring about something in the environment; or rather being made up of a muddle that could easily shake out that way for at least some things.  (That's how it happened for humans, leaving out some complexities of reflection-towards-coherence along the way.)

There's nothing forcing an agent to shake out that way, but it is simple to care about every part of reality by running your utilityfunction over its configuration, and if a component of many humans' values lands there, then plausibly so do a bunch of other things' components.

Even if Reality did end as mostly Entities that were mostly Locally-Caring, the non-Local Entities would still trade with each other; trade-optimized regions would then just be a smaller total fraction of Greater Reality's dispensation of its intelligently controlled realityfluid.


Carissa agrees that something probably not-Cayden put Pilar there as well as Carissa; Asmodia feels less suspicious, to Carissa, as a person, Asmodia seems like the kind of person lots of people are. …except that maybe someone had to be a person who'd say to Carissa that they wanted to not exist, so that she could understand that as something a person might actually want for themself and not just want inflicted on others, so that she wouldn't report Keltham to Asmodeus the instant she realized that his plan might destroy the world. 

She understands that natural selection produces muddles which can then shake out in many many ways that aren't what created an advantage in the selected environment, and probably other processes for producing intelligent beings - breeding them deliberately? - would do the same thing. It still seems like there are powerful forces in the direction of caring about yourself and your family and your nation and your species more than you care about all sapients; she expects most evolved creatures not to care about everybody, though tentatively.


Asmodia was oddly adept at learning the more mathematical parts of Law, for somebody to just happen to be in the same class as Pilar.  The second group, sent over when Cheliax assigned the project a higher priority and was trying to allocate smarter people instead of gift-girls, didn't seem to contain another Asmodia.


Lots and lots of evolved things, and probably even more of the Entities that the evolved-beings birth (if they mess it up) would end up not caring about sentient beings at all, whether Locally or Everywhere.  The categorization he's suggesting generalizes beyond caring about sentients, for example:

Suppose some being cares a lot about particular shapes of matter, for example, and prefers matter being curled up in one squiggle shape - a rounded-rectangular spiral pattern say - while greatly wishing that no pattern ever be squiggled up in a hyperbolic spiral.  It might feel no pain or pleasure, no joy or happiness, about the fact; it might just act so as to bring about the rounded-rectangular squiggles, and avoid the hyperbolic spirals.  Or it might feel pain or pleasure internally, as it learns about spirals or squiggles being created, but be indifferent to this as it plots its goals.

From the perspective of the Squiggle-Caring Entity, some other Entity that cares a lot about people having conscious experiences is then just a strange being that cares about some weird and more abstract and complicated pattern that matter can be squiggled into.

Playing out some additional particulars in case the abstract pattern didn't make it across the telepathic gap at the speed they're trying to think at each other:

That Entity would be categorized as Negative-skewed or Positive-skewed in that utterly sentient-uncaring utility function, depending on whether one ill-shaped squiggle pattern cancels out the utility of 100 good squiggles or 0.01 good squiggles.

A Negatively-Skewed Squiggle Entity, whose utility function arose in any way suggesting that this utility function arises more often in Greater Reality than its (Positive) inverse, has incentive to be signatory to a compact whereby the Future of dath ilan's Civilization would spend lots of resources to intervene, if Future-Civilization ran across some smaller thing that also really hated hyperbolic spirals, and some mid-sized Pessimizing Entity was therefore making lots of hyperbolic spirals because it had decided to pessimize the smaller agent's utility (maybe after making a threat that was decision-theoretically-properly refused, and having evolved to be hateful itself).

Future-Civilization wouldn't care about hyperbolic spirals as such, and the smaller thing might not be a kind of being that experienced unhappiness as such, or had even chosen to have any conscious experiences at all.  But in return, the original larger Squiggle Entity would intervene if it saw somebody torturing a mortal; not because it cared, not even because the mortal was trading with it, but because Future-Civilization had agreed to avert the pessimization of Negatively-Skewed Entities in general, conditional on its expectation that a lot of Negatively-Skewed Entities would do the same, and Entities signatory to that compact were expected to occasionally avert small or large squiggle-caring beings from being pessimized by having lots of hyperbolic spirals created at them.

A logical trade like that comes with friction costs.  So Negatively-Skewed Entities are more likely to be signatory to a pact like that than Positively-Skewed Entities; because even if they only assess a 10% chance of Future-Civilization actually existing to execute the bargain that they predict Future-Civilization to execute if it exists, and even if they expect Future-Civilization to underestimate by a factor of 10 how much the Negatively-Skewed Squiggle-Carer exists to pay them back, it's still worth sacrificing the opportunity cost of 100 rounded-rectangle squiggles to prevent one expected hyperbolic spiral from being made.  A Positively-Skewed Squiggle-Carer would conversely demand 100 units of realityfluid be spent on preventing hyperbolic spirals in order to justify sacrificing 1 unit of realityfluid that could have been spent directly on rounded-rectangular squiggles.


She doesn't want to invest too much in contemplating how as a god she'd handle that problem, when it's a probably-wrong model Keltham sketched out and probably not a description of actual reality. But it is the bleakest possible imagining of the universe, that it's mostly full of entities who'd consider the universe not existing to be among the best possible outcomes, who are set up structurally such that almost everything that could possibly happen is bad and all the good things are worth losing to slightly reduce the chance of bad ones. 

Pharasma and the winning god-coalition destroyed the gods that sided with Rovagug, and she would do the same thing to entities that would side with Rovagug, if she could.

(Probably in some universes Rovagug and the gods that sided with Him won and everything got eaten, and she's not in those universes because they don't exist, and she understands how that counts as triumph for some people, how they might just want to peel more and more Golarions away from the branch where people live and die and love and fight and laugh and cry and squeeze them dead and make the "Golarions eaten" branch of the tree a little thicker. But it's not what she wants.)


Negatively-Skewed Universal Carers don't need to end up wishing that somebody would erase their universes!  Unless they really hate the way that reachable matter is shaped by default, compared to it not existing; but hopefully such beings are few... or configured in such a way that they're not experiencing constant suffering about that; he himself wouldn't want them to be unhappy, though it's not like they themselves need care about "unhappiness".

So long as Reality isn't allowed to end up full of Zon-Kuthons and Asmodeuses and other such utilityfunction pessimizers, most Entities, even Negatively-Skewed ones, could mostly be getting things they want, and not things they hate.

People who feel like Reality loses more when one person gets crushed and tortured, than when ten people lead happy lives, aren't necessarily out to destroy Reality.  If no one is being tortured, there isn't a problem!  (Or rather, if nobody reachable is being tortured, there isn't a problem you can solve by destroying the local universe, and you might as well not think about it or be sad about it either.)

Civilization would have fought to defend itself from destruction, and did fight to defend itself - because while there were possible states of matter that would lose more (compared to a null state of lifeless matter) than the best states of matter would gain, they managed to stop those bad things from happening, and be mostly happy.


That's the grand dream and vision, from the Negative standpoint - that Reality not contain a lot of utilityfunction pessimizers running around and occasionally pessimizing beings with utilityfunctions similar to their own, so that Reality as a whole is something they're still glad to have around.

That's why Pharasma is being given an out.  She doesn't need to have Her Creation destroyed, if She's willing to have it not contain such a large element of utilityfunction pessimization.


That did actually occur to Carissa, or something like it, when she first decided not to betray Keltham and to come to him instead.

She thinks she may have made an error there, but -

- but it seemed to her that Golarion would not, really, endure for the forever she wanted for it, while it had Hell in it, that even if she warned them and they crushed Keltham there would be another like him someday, that the only way for Golarion to endure forever, like she wants it to, was for it to be something large shares of reality didn't want to destroy.

She understood a thing Keltham said on her first day here as a claim that this wasn't true, that he didn't think he was extending Golarion's lifespan in expectation. But - with caution about the mind-states that permit her to help him without being a threat to Pharasma - she does, actually, strongly value the destruction of Hell for that reason and think some substantial risk of destroying the world would be warranted, though by her own preference you'd spend decades exploring other Hell-conquest options first.


It's not a consolation he feels honesty-safe and epistemically-safe about offering her.  That Hell has been allowed to persist this long within Creation (on those hypotheses where all of this is as real as themselves, and not quantitatively much less real than that) is evidence (within those hypotheses) that intervening in Creation is expensive; and only became some combination of affordable+attractive after the death of Aroden lost other hopes, or the shattering of prophecy made it less expensive to act against gods.  It is possible that if this intervention against Pharasma fails, no other will be sent.

Decades seems like quite an unreasonable amount of time for smart people to think.  He'd take that time only if there was sufficient value-of-info; and he didn't expect to slip up, tip off the gods, and get squished, inside of that delay.

It does seem clear enough to him what he is sent here to do.  He's not happy about it; he's a lot less happy about Hell.


Carissa is largely resigned to the fact he is going to try it.  She wants to convince him that only Hell is worth doing it for, and that if he gets Hell but Pharasma is unable or unwilling to change any other things about Creation, then that should be sufficient for him not to prefer the world destroyed. She hopes to narrow the specification further: figure out the actual minimum ask and make sure it's a concession Pharasma can grant if She wants to.

She sees that Keltham isn't going to be willing to not try to destroy the universe. She hates him for this, on some level, but she doesn't predict anything different.

Total: 289
Posts Per Page: