Right.  What's first up, then?

How would Pilar feel about going to the former Chelish city of Korvosa, which is still on good terms with Cheliax, and preventing its inhabitants from being horribly sacrificed in somebody's mad blood ritual seeking immortality?

Sure.  Can Snack Service tell Pilar who's going to do it, and then Pilar can sweep in with an army company and twenty high-level adventurers and kill them?

Snack Service both won't, and can't, be any more specific than what Snack Service has said already.

Cool.  Pilar's going to find whoever's in charge of the city, identify herself honestly to them, and tell them a self-fulfilling prophecy about how they're going to successfully prevent their city from being sacrificed.  Then, instead of trying to do everything herself, Pilar is going to solve the problem in full cooperation with the authorities, obeying any reasonable orders they give her along the way.

Pilar can totally try that and see what happens!

Carissa does feel terrible once she takes the headband off but she decides to ignore this; if she's might only have weeks to live it's some kind of unimaginable horror to spend a minute of them sulking. She looks over her to-do list, instead, to see how many of the items on it still seem doable. 

She wants to reach out to Osirion and see what they know about the current status of Cheliax and Project Lawful. She suspects it'll be less than she hopes, because probably they were spying directly through Abadar, but conceivably Abadar's still at it, or they have other channels. She wants to talk to Erecura and maybe Dispater. She wants to figure out if confidential negotiations with Abadar and Iomedae are possible. She wants to publish cult pamphlets for drop-off in promising locations. She wants to pressure-test her Abrogail-assassination plans and maybe run them by Ri-Dul, who seems like the kind of person who has contemplated how he'd assassinate Abrogail Thrune. 

She wants to write a long angry letter to dath ilan about how they are the worst, in case this is actually a work of dath ilani fiction and she can influence them that way. 

She wants to play with sixth circle spells, even though she's never, ever going to grow powerful enough to cast them herself. 

She wants to write a letter to the children she would someday have had if they'd gotten to exist, which they won't. 

After that, whatever, she'll make headbands or something. 

Golarion's conventional view on Intelligence and Wisdom:

Intelligence is useful for learning how to read and doing figures and learning more complicated work; at higher levels, you can learn complicated math and be a wizard.  It has little or no effect on personality; you can be good at math or bad at math and have pretty much the same underlying personality.

Wisdom is perceptiveness and self-control; at higher levels, closeness with the divine.  Wisdom has a more pronounced effect on personality than Intelligence; it's associated with maturity rather than cleverness.  Golarion languages with curse-words for 'stupid' tend to mean low-Wisdom rather than low-Intelligence, albeit in the end the term mostly means low-status, of course.  The actual abilities associated with Wisdom, if not always the measured abilitystat, tend to increase with age instead of go down as Intelligence does.  Sudden stress episodes can increase apparent wiseness some time afterwards, though usually not the measured abilitystat.

From the perspective of dath ilan, that uses mental calculation for things Golarion does not yet know how to do - well, mostly dath ilan's perspective could not easily be translated out of their worldview at all, and would make mention of notions like 'RAM and processing speed' or 'cognitive reflectivity' or 'frontal cognitomotor cortex'.  But to translate it back into Golarion terms:  Intelligence is calculation and all that dath ilan can do with mental calculation, but Wisdom is the ability to control that calculation-power and listen to the results of that calculation-power.

INT 27 and WIS 20 makes you a dath ilani who's exceptionally talented at math even by their standards.

INT 27 and WIS 26 makes you something else entirely -

"Here's two more artifact headbands.  The +6/+6/+4 is on indefinite loan.  The +4/+6/+6 has to go back after the trip to the City of Brass.  Somebody else will come with on that trip and get what you were planning to get for yourself... so full upgrades, I suppose.  They also want a couple of hundred Wish diamonds afterwards and asked for some other stuff I gave them."

"Right.  Well, take Carissa's artifact headband back to her, then, I don't need it anymore.  Presumably that's the point and the implicit bargain, They can't help me."

He quickly removes Carissa's headband and successfully ignores the resulting massive sense of loss and disorientation while he pops on the +4/+6/+6, because yes, it's obvious that if he tries to rebuild himself a personality it's better done at INT-25/WIS-26/CHA-25 than INT-27/WIS-26/CHA-23.

...and then he puts himself back together again.

His own opinion is that you couldn't really call the result 'Keltham'.  If you take a short lifetime of experience at INT-18/WIS-16/CHA-14 and pour the result into a mind of INT-25/WIS-26/CHA-25, it's a bit like pouring the contents of a four-year-old into an eighteen-year-old container.  The numbers skip over a lot of important detail.  Golarion's abilitystats may not really be thinkoomph but they sure are some major subdimensions of it.

He is putting the traumatized pieces of that four-year-old back together as best as he can, improvising psychiatry skills that he was never taught because you can just do that at 25/26/25 if you have enough of the underlying knowledge.  (He could have reinvented psychiatry eventually, as he was before, but not quickly.)

But it's still a four-year-old being poured into something much larger than a four-year-old, with rapidly cascading knowledge and insights that contradict a lot of what did go into that four-year-old.  At some point there's not that much difference between making the new person out of Keltham-who-was and just making them out of a generic dath ilani with the same abilitystats.

...he still tries as hard as he can to make the new person be Keltham+, not just the + of a generic dath ilani with the same abilitystats.

There's a lot of reasons to do it that way, and no particular metareason to single out any particular reason as the justification.

He should not try to really talk with Carissa like this, he doesn't think, even if she has her +6 Wisdom back.  Not until she has her own abilitystat increases to go with her headband.  It would damage what's left of their relationship, and Carissa is one of the only things left in Golarion that still felt real to Keltham.

He types Carissa a letter, and doesn't need to retype it afterwards.  It includes among other things the notes that:

- If Carissa wants to assassinate Abrogail, she needs to do that before his child with Abrogail is ensouled; which may interact in a complicated way with other plans to kidnap Abrogail and/or see if Carissa can don her crown as a way to impede Cheliax's invasion of Osirion.

- He will strongly push that Carissa wait until she's INT 29 to actually negotiate with Dispater or Erecura, but she can potentially plan that now.

- If Carissa has copious free time left over, he thinks she should help him develop his Magical Simulation of Magic because that'll be the key Skill for Wishcrafting.  If she can't help with that or speed that up, so there's more total time to do it and make it safe, then what was she buying time for anyways?

- He's getting better now, sort of.  But until Carissa gets her own abilitystat boosts, his new improvised psychiatric skill strongly suggests that they should only talk dispassionate strategy or engineering with him wearing the +6/+6/+4 headband, and not say anything about their relationship or their conflicting goals.

- having more time is obviously good under a wide variety of assumptions about where precisely from here - yeah. Not talking about that kind of stuff with Keltham right now. And she's happy to help with his Magic Simulation. 

Her scenarios for assassinating Abrogail, a dispassionate letter back to Keltham reads, are as follows:

- assassinate Abrogail, raise her immediately. Don't take her headband. Explain to her that Keltham knows about the children and that Carissa may have just prevented him from blowing up Cheliax, but if he ends up considering it too likely there are more children, he might still do it. Suggest that Abrogail sign a peace with Osirion, as would credibly indicate that Abrogail's was the only pregnancy and Cheliax doesn't anticipate acquiring leverage against Keltham soon anymore. Abrogail removing children/relocating them does not reduce incentive to destroy enough of Cheliax to prevent invasion of Osirion. Abrogail might be able to credibly promise there are no more children. Abrogail might call the possible-bluff and say that she won't sign a peace with Osirion and if Keltham decides to destroy Cheliax, so be it, in which case Carissa will regretfully depart and maybe go to step 2.

- assassinate Abrogail, take her crown, tell her security that Carissa owns options on their souls and they should consider reconsidering their loyalties. Aspexia will be suspicious. Carissa plans to show off the souls she's been buying to assuage Aspexia's suspicions but there aren't that many, and Carissa doubts she'd be able to stall the invasion of Osirion too far, and it means she'd be out of time dilation and busy during important stages of planning.

- assassinate Abrogail, take her body and her crown, hope the succession crisis in Cheliax delays the invasion of Osirion by months.

- kidnap Abrogail, possibly via briefly killing her but the intent in this scenario is to have her alive, take her to doombase, hope the succession crisis in Cheliax etc etc and that Abrogail can be persuaded to help overthrow Asmodeus and take His job. Unlikely but would be very valuable if it worked.

Abrogail might refuse an immediate resurrection, making it unsafe to kill her as part of an intended kidnapping plan.  She can still have the child aborted via Baleful Polymorph to male.

Kidnap Abrogail, via:

- Route 1, luring Abrogail outside the Palace Forbiddance not under a Mind Blank, via fake intelligence leak suggesting an assassination attempt, which Abrogail may be willing to play along with in hopes of luring Keltham into complacency;

- Route 2.1, hitting that Forbiddance with a Dispel from somebody as powerful as Rugatonn, or 2.2 successfully obtaining a scroll of Mage's Disjunction;

Then, have multiple noble Efreet in rapid succession cast Wishes to try to kidnap her until Abrogail fails a Will save.

This may plausibly not work even then, based on Keltham's research into levels of magical resistance apparently granted by the Crown of Infernal Majesty, in the light of purchased reports on past assassination attempts against Crown-wearers.

Ideally, Cheliax should be allowed to think Abrogail permadead and out of the game for good, as by statuing her or soul-trapping her, and having Carissa attempt to claim the throne herself.  This forces them to deal with the succession crisis rather than Rugatonn just appointing a regent.  If the successor appears to be getting organized enough for an invasion, this would be a fine time to send Abrogail back, possibly with visible gaps in her own memory just to complicate things.

Yeah, the reason Carissa thought they'd assassinate Abrogail even in the course of a kidnapping is that kidnapping an alive Abrogail seems very hard. Assassinating her isn't; Ri-Dul could presumably walk up to her with a Contingent Antimagic Field that triggers when in Abrogail-range, at which point nonmagical ilani weapons should suffice. Getting out is the complicated part, but you can always just not try to; kill yourself too, have a True Resurrection already reaching completion. 

If they were to try kidnapping an alive Abrogail Carissa would honestly be templed by just saying to Abrogail 'I swear I think this reduces the risk of Cheliax or Egorian being destroyed' and Plane Shifting with her. Any plan that involves displaying lots of implausible capabilities seems likely to make someone realize that Keltham is better armed than the typical sulking teenager Rovagug cultist aspiring mass murderer person they're modeling him as. 

It just has to look like something that an angry, vengeful teenager could pull off using way too much money, maybe with Carissa as the real mastermind nudging his choices.

Does this mean she's giving up on closing-the-Worldwound plans?  Though Keltham is somewhat doubtful that his Wishcraft is going to reach that level in time, regardless.

In her current estimation Golarion looks certain to be destroyed. That makes it not worth bothering. She's hoping she's missing something she'll see when she's thought about this more and Wished on more bonuses and gotten drunk on Nefreti Clepati's wine and tried literally anything else there is to try. 

(Then she should go on playing for log-odds of success, right now, as she waits for that missing insight... this should not be said to Carissa at the moment.)

Does Carissa have any suggestions for highly stable and predictive magical dynamics?  He has a computationally-complete set that he's trying to use for the current Magical Simulation of Magic; but he still has less total experience playing around with spells, and there's a proverb out of dath ilan: just because it's possible to compute anything in a computationally-complete-system doesn't make it easy.

She's mostly just tried setting aside Golarion and thinking about the rest of the universe, the destruction of which is less certain. She doesn't see how Rovagug will be contained in time to save Golarion even with Asmodeus interested in doing it, but that's still a lot of other planets, and all the afterlives, to try to save. 

If Carissa were trying to develop a structure for describing the behavior of magic she'd have approached it very differently; she'd have picked the spells that are the oddest-feeling in your hands, the ones that feel slippery or tense like they're barely holding together, the metamagics that feel the most like they strain the original spell, and try to figure out what tolerances, exactly, those are straining, what invisible dynamic they come close to breaking. She's not sure she has anything she can translate into Keltham's terms, yet, but that's what she's attempting. 

Hopefully without any offense, he can't figure out whether she just said something relevant whose complicated relevance he missed, or if she wholly missed the relevance of what he asked.

Restating background, in case a piece of it wasn't shared:  He's not trying to build magic structures whose visible parts mimic particular invisible structures.  Somebody would've already tried that, to build a visible model of the invisible.  Instead:  There's a bunch of mathematical theory describing, for very simple cases, or particular interactions, the parts of magic that people can't see.

He is trying to build a system that encodes those known or guessed mathematical laws, where you'd have to look at the visible magical part of what he built, and then apply transformations in your own mind to read off what it was describing; it doesn't need to look like the spell being modeled, as is a thought that might be more obvious to a computerprogrammer than to a Golarion wizard.

He's also not trying to find visible interactions that directly mimic invisible ones, even via isomorphism.  He's trying to build up pieces whose interactions can be compounded together in ways that would mimic arbitrary mathematical rules.  When a piece of magic gets sticky because it's interacting with 11 other pieces in 5 dimensions, he builds a visible thing with five number-levels and 11 other pieces affecting the five number-levels additively, to simulate additive forces in 5 dimensions.

He was asking Carissa what sorts of very predictable, very regular magical interactions she knows about, as he might be able to use as additional pieces in a process that mimics very regular mathematical rules.  Including things that wouldn't ever appear in a spell, but if you put them together in a weird way where one piece stabilizes another, or forces it to flow along a narrow channel, it then becomes highly regular and predictable.

Possibly they should not in fact attempt to collaborate on this either right now, but once you've tried accounting for the very obvious regularities in magical interaction that are apparent in any spellcasting, the thing she would try, if she were building the thing he's trying to build, is to look at apparent irregularities, places that feel like they violate the intuitive rules that every wizard learns for the behavior of magic, and figure out what's actually going on in those cases. Which is what she's trying.

...continuing communications breakdown; he still didn't see at all how that suggestion interacted with what he's trying to do.  He's not trying to figure out hidden dynamics for particular spells, or even complete/repair current theories of magical physics.  He's looking for very stable, predictable, and above all precisely interacting, pieces, with which to build a magical 'computer'.

She thought that his aim was modelling magic so that he understands what parameters need to be specified in a Wish for it not to go catastrophically badly.

Correct, but he's not trying to do that the way previous INT 27 casters would have in earlier centuries, by finding current magical anomalies and poking at them until he figures out what's going on.

He's building a thing that can simulate possible things that could be going on, so that he can, for example, feed in 30 different possibilities, and see if any of them reproduce the behavior of reality, and think analytically about how they fell short.

He's doing that by building a general toolset for simulating continuous processes, which he can then adapt to simulate possible laws of magic as a special case.  In principle, he could calculate out the possibilities himself by hand, given infinite time, but he doesn't have infinite time, so he's making a thing that does the calculations for him.

To do that, he needs pieces that interact in precise ways, so that they can model precise hypotheses; and they need to go on interacting in that precise way even when there's a lot of them, which is the hard part with magic.

Ideally, he needs small regular tiling parts which can be spun off by other stable magical structures, so that he can spin off a hundred thousand of them that will last for ten minutes apiece in a regular pattern whose resulting internal forces will allow it to be stable.