"Sounds like fun.  Five minutes."

"Keltham started one godwar in his first three days of being here so I don't see how we get through an entire year without another one," says Meritxell.

"Well, that was with Zon-Kuthon and there's not another god who almost wants to destroy the world. Except Rovagug, and someone'd have to let him out."

"Well, maybe someone'll do that," says Meritxell. And smugly: "though I wouldn't give four in five specifically for that."

"A god wouldn't have to almost want to destroy the world," says Pela. "We're changing it enough they could've been really in favor before and still turn out against now. Like....Urgathoa, if Keltham cures all diseases."

"Wouldn't she have figured that out and fought with Zon-Kuthon?"

"Do we know she didn't?"

"Well if she did then that's not another godwar in the future."

"Okay, actually, you know what, on second thought, everybody shut up and my apologies for not thinking ahead faster about what might be a dangerous line of thought.  I suspect that properly this conversation happens between myself and the Grand High Priestess in a heavily screened room.  Nobody repeat the name of that god you just mentioned until cleared to do so."

He wants to tell them not to think it, but has a dreadful suspicion that this would be a counterproductive instruction for non-dath-ilani.

- solemn nods and immediate silence.

"Going right back to Probability, one thing I'd be enthusiastic to test is whether devils make the same error.  If we try this on twelve of whatever Gorthoklek is, do they make the error?  Twelve Lrilathas?  Twelve of whatever devil said hi to Asmodia when she showed up?  Twelve people who've been in Hell ten years, fifty years, a hundred years?"

" - I bet not, for full devils. I'm less sure about new petitioners but we could, actually, ask Hell to check for us."

"Do you know how expensive that is to ask?  Also, doing this with 144 subjects is more reliable than 12 subjects, if it's not much more expensive to run a larger test on the other side."

"No idea how expensive but we could ask the High Priestess. I think most of the expense would be communicating the instructions so larger's probably fine if we can do it at all."

"If we can batch questions more cheaply then I have additional questions.  Such as, for example, whether devils already knew about where the balance comes from between men and women, and weren't allowed to say.  Or we could find that the younger devils don't know it until we tell them, but the older devils already know it, which would imply that information is stratified within Hell the way that it's stratified between gods and Golarion."

"If Lrilatha and Gorthoklek are past this fallacy, they presumably watch people committing it every 3 minutes, but are not allowed to say out loud what all the humans are doing wrong.  Whatever rules prevent them from showing you what I just showed you, those rules are a key part of the foundations of order for Golarion.  I want to know what those rules are.  I want to check if I can maybe snap them over my knee in five minutes if I come at them from the right angle."

"Like, say, maybe there are things that, say, Gorthoklek isn't allowed to tell younger devils, but Gothoklek is allowed to point out those pieces of knowledge to me because I already know, and then I can tell one younger devil that, and they can tell others."

"Of course, before doing that, it might be good to know why those rules existed in the first place."

"I don't know the deep secrets of Hell but my prediction would be that Contessa Lrilatha gets it right the way my father might get it right because he does a lot of guessing which ships are going to be profitable, and he doesn't know a rule he could tell me about, but he's pretty good at what he does. And Contessa Lrilatha would be better but still like that. Maybe not, though."

Broom is, without showing any outward sign, trying to pray to determine whether his goddess wants him to kill Keltham yet.  As always there's no response, leaving matters up to his own best judgment.

Keltham has no particular idea that he might be close to death!

"Well, that result would be interesting from a completely different angle.  Because if Asmodeus does not explicitly know that thing I just showed you, I will be really surprised.  Which would mean there are things I know that Asmodeus can't tell Lrilatha."

"...I'm actually just going to check that the next time I meet her, shouldn't be long at this rate."

Girls sit there nervously.

Keltham is aware that some people might consider the current topic of conversation to be portentous!  This is deliberate!  He has an infohazardous chain of thought he is trying to distract them from, and himself too for that matter.

"But let's return to the experiment we just ran.  One notable thing about it is that, just based on the experimental results themselves, we can't point to any one of you, and say, this person must've done something insane.   We can't point to Peranza," yesterday Keltham gave up and did actual memory exercises to try and remember people apart from their nametags, "and say, her estimate was unreasonably high, her estimate was unreasonably low.  We can tell that the two groups were collectively insane but not that any particular person in them was insane."

"Let's go back to considering the estimates for the chance of duck at lunch.  Even after we get to see lunch today, can we say anything about who was sane and who was crazy, based on the 12-point scale?"

"I mean, we can keep track and see if we notice a pattern over time, about who is good at guessing."

"What kind of pattern would look like being good at guessing?"

"....being right more often?"

"Well, on a scale of 1-12, suppose you say 4, and then there's duck for lunch.  You say 5 the next day, and there's no duck for lunch that day.  You say 4 the day after and there's no duck for lunch.  How are you doing?  Are you being right more often?"

"You can make some kind of scoring system where you don't really get any points for saying '5' every day and get points for higher numbers when there's duck and lower numbers when there isn't duck." 

And then she can do performance reviews. 

"And there's more rules like 'you can't have your number go up when you're predicting more things at once' and when you have four rules like that there's only one possible scoring system which is the Law of Scoring," Asmodia says out loud, a strange electric excitement running through her.

"Good, Asmodia.  That's not literally exactly correct, but you've seen the pattern."

"That's sufficiently far ahead of where I was going, that the next time you have a prediction that far ahead, I want you to say 'Prediction' and then write it down instead of... or actually just say 'Prediction' and then Message me, because we're all casters here."

Asmodia gets some envious glances. 

If they want their own envious glances they can go be that good.

"Now, if we could all just pretend to forget that Asmodia spoiled the book for us by telling everyone the ending while I was just getting started..."

"We've just heard a new rule proposed by Carissa that you should get more points for giving higher numbers on duck, when there's duck, and more points for putting lower numbers on duck, when there's no duck.  Well, I agree, that's pretty reasonable.  Any other rules come to mind?"

"...if a prediction breaks down into predictions for two parts, you shouldn't get more points for the combined predictions than from getting each of the parts right separately?"

"If - hmm - if two people both say duck, but one is more sure than the other, and it's duck, that one should get more points."

"And if two people say no duck and it's duck, but one was more wrong that one should lose more points. And they should get the same points for the same prediction."