"I can tell you the result of that one, if a Chelish-Tian mating is a human one and the human matings work anything like they do in dath ilan.  Twenty-three package-pairs of heritable information, remember, with each child getting their two packages one from each parent package-pair, selected at random?  So most things in a human mating will be a mix, unless a trait is being determined all at once in a single package-pair location, the way that sex is determined by the sex package-pair.  Intelligence is not determined all in one spot, if it's determined here anything at all like it is in dath ilan.  There'll be bits of heritage all over the twenty-three package-pairs that affect it, positively, negatively, and subtly."

"Well, and now that I've told you that, is there still anything you could find out from observing Chelish-Tian matings?  What could you observe differently, where you're not already sure of what you might observe?  What could those possible observations say about the package-pairs and the heritable-information coding for Intelligence?  What could it be saying about what's going on inside of those package-pairs, in Cheliax, in Tian, and in your guesses extrapolated from distant Tian to the far more distant dath ilan?"

"...variance," says Meritxell. "If you mix two eighteens you probably get an eighteen, on average, but - but it's much more valuable to get a sixteen and a twenty than two eighteens, because the twenty can end up running the country - if Chelish people and Tian people have the same bits making them smarter, then they'll have the same variance mixed, but if they have different bits then they'll have higher variance."

"You have slightly impressed me.  Be justly proud of this."

She beams at him fiercely.

Oh, good.  After he said that, Keltham started to worry that some dath ilani flirting tropes wouldn't make it across the vast cultural gap, but at least that one seems to have landed.

"Now here's a harder question.  How valuable is higher variance in the intelligence of offspring, if the alien visitor has mostly different bits in his package-pairs that increase and decrease intelligence?  How much can the Chelish government gain from using that variance, how much should they be willing to pay?"

"Well, depends how valuable smart people are. But I think - I mean, one person with native-born INT 20 is itself something they'd pay a lot for, and the variance thing probably applies for a couple generations until it's all diluted..."

"Hm, yes, how much you're willing to pay for one Intelligence 20 offspring does depend on what use you can make of them, which in turn depends on how clever you are in thinking up potential uses.  This advisory panel of the greatest native experts in Cheliax on heritage optimization has been convened to make expert recommendations about heritage-optimization to the Chelish government, since that is what this advisory panel knows more about than anybody else in Cheliax, or Golarion for that matter.  This may now actually be true about you in real life, by the way.  Anyhow, the rest of Governance will decide for itself how valuable an Intelligence 20 person will be for purposes of fighting at the Worldwound and such.  What can this advisory panel say to Governance - or wildly guess with appropriate qualifiers - about the clever use of any Intelligence 20 children from the alien, for the purposes of heritage-optimization in particular?"

"- you probably shouldn't pair them off with their half-siblings, you get weird genetic defects that way, and there aren't any native twenties...that I've ever heard of -"

"And now I know you actually do have the package-pair system around here!  The pairs of packages carry duplicates of a lot of the same genetic information for constructing a person - that is, if a package is carrying some of the instructions for building, like, fingers, you've got the same information on both packages in the pair, usually.  If one of the package-pairs gets damaged, in that particular, the other element of the pair can often take over and make sure your fingers still get built.  If a brother and a sister mate, their offspring has a one-fourth probability of ending up with the same package twice, for each of the twenty-three pairs.  You are a lot more likely to end up with no information for building fingers that way.  In dath ilan, the corresponding equilibrium is that people usually aren't sexually attracted to other people they grew up with.  If people here weren't also built from package-pairs, they wouldn't get the same deleterious effects from mating with half-siblings."

"Anyways, yes, you shouldn't pair off my kids with each other in the first or second generations, unless you've already developed other magic or technology for telling who got which packages from me.  Still, what are they worth to the Chelish government?  Ignore the part for the moment about whether any of my kids are smarter than any other kids in this world - they may not be, for reasons I'll get into - that's a fighting-at-the-Worldwound issue anyways.  How would you guess the value of intelligence variance for the heritage-optimization program, or the value of having different bits of heritage that are increasing and lowering intelligence?"

They are pretty stumped by this. You could have smarter kids, but that's been said already. You can't directly try to use the best bits because they have no magic that refers to things on the level of bits.

It's not occurring to them that they could just... "Ah, what kind of - intelligence-training games, you don't have a word for them, that's not a good sign - what kind of complicated games do Chelish children play, if any, at all?"

People play Knights, which is a two player board game with pieces that move in different ways, and Spy, which is a group game where some children are spies and some are soldiers and they have to figure out who's sabotaging their operations, and it takes them a suggestive while to come up with a third, though someone eventually offers that there are spelling bees.

HOW DOES EVERYTHING MANAGE TO BE THIS BROKEN SIMULTANEOUSLY WOULDN'T THE KIDS THEMSELVES SPONTANEOUSLY INVENT BETTER GAMES THAN -

"Right, then, what we are going to do now is invent a new reality-mirroring game.  That is one of the things that a dath ilani child would do as a matter of course, faced with a question they had no idea how to answer, whether that question came from an older child or a Watcher or from life itself; they would try to invent a game, corresponding to the question, that they might learn something from playing."

"There is math that I know which describes in a quite simple way the value of variance for heritage-optimization, as it happens, or the result of having the variations come from different bits of heritage.  But it can be hard to correctly derive the math that will give you the fully general answer right away.  Inventing games can be straightforward in a way that doing math is not."

"I did tell you how the twenty-three package pairs work.  So why not invent a game with, say, four package-pairs, and four places in each package where heritages can potentially vary?  How would you make a game like that, and play it, to learn something about how heritage-optimization might play out?  As an advisory panel to the Chelish government, it may be worth your time to spend a whole day playing games like that, if that's where you're generating your ideas and advice - in fact, you might even hire other people to play them for you, and report back about the results.  Though here, you can't do that yet, even if you brought in somebody else to hire in real life, because you're still learning how to invent and play relevant games at all."

They set at this. Somewhat abashedly. ...everyone has four package-pairs and four places where heritages can potentially vary, and each spot can either be a "smart" or a "nothing" or a "dumb", and your score is the smarts minus the dumbs. Everyone can trade with other people, and you can drop the current 'person' you're playing for any of your offspring, and other people tell you their total score but not any components, and you try to have the smartest 'person' after twenty rounds.

Would that be the right kind of game.

"Play it once and see what happens!"

They try it? It's not clear anyone is having any fun but they work very effortfully at it. Mostly everyone tries to do all of their trades with the highest-scoring person around, who has no real incentive to trade with dumb people. This is probably reflective of life but it's not very fun.

This game is not even slightly reflective of how anything would plausibly work in reality, and nobody has any idea of how fun game design or accurate simulation-game design works, and they are playing a game that would be cooperative in real life as if it is a competitive one, and it is occurring to Keltham for the first time that they may not even consciously know the difference between positive-sum and zero-sum games, which sure would explain why simple words for that don't seem to exist in Taldane, or why no Chelish book or Chelish person has ever happened to use the terms around him yet.

Right.  This is fine.  Everything is fine.

"All right, let's stop there," Keltham says, trying to maintain the same calm demeanor that a Watcher would have in this situation.  "I think this game isn't playing out the same way it would play out if Chelish governance was running a heritage-optimization program; do you have any ideas what went wrong and how you should modify the game?"

They noticed some problems!

 - high intelligence people can't get much out of trading, whereas in real life people are pretty motivated to have sex even if they don't get smart children out of it

- you can only be one person, so you have to ditch the person you were if you want to be a new person, but maybe you should just get to add them to your stable? that'd incentivize having more children, though also the paperwork would get really annoying

- no one has any particular insight into how some people got better, it seems like mostly just luck

"I am genuinely not sure how to introduce this topic if you have never encountered it before, but sometimes in real life, a group of people have to work towards a common goal, and at the end get some payout in common that they have to divide up - according to previously agreed-upon rules, if they were smart.  Games can work like that too, and should, if they're trying to mirror that kind of reality.  Chelish Governance is going to be trying to make the game work like that, that's sort of the whole point of having a government in the first place."

"The other thing that's missing is that you don't - seem to have a current notion of what you're trying to find out from the game - you know, I think I may be doing this wrong and not providing you with enough direction literally the first time you're doing this.  You play games with older kids leading, before you get handed some problem that requires you to invent a game for younger kids to play with you.  Well, I may just be motivated to think that because of my personal interest in having this particular topic taught quickly; still, the rationalization seems valid after having been invented."

"So!  Things you could be testing in this game include alternate rules that Chelish governance could give the players about subsidies and rewards, which would lead players each trying to maximize their own score, to do useful things for heritage-optimization.  And now it's occurring to me that maybe you don't already have this concept, maybe people here just make up laws that sound like good ideas and don't play games to figure out what the laws will do in their effects on selfish people.  In which case I also have to, at some point, convey basic competence for figuring out what effects different legal systems would have and optimizing better laws, which, the complete wreckage of what my own people consider Lawfulness in Golarion, should maybe have suggested to me earlier was also going to end up a Project issue, but never mind one thing at a time."

"The other thing you were going to investigate was the effect of adding one person with different bits of heritage to the system.  So for now, just make up some fixed rules about points that players score by having smarter characters in their hand, let's say there's at most three characters you can keep in your hand at one time, and your objective is not to score higher than other players, it's just to score as many points as possible for yourself.  Characters can be male or female at random; men can have any number of children in a lifetime, women can have only three children per lifetime.  When any two players mate their characters, they have to agree in advance on which player will retain the resulting offspring.  Any time a mating occurs between two characters one of whom has over ten smarts, the Chelish government pays an extra point to whichever player doesn't end up owning the resulting offspring.  Any time a mating occurs between two characters both with over ten smarts, the Chelish government pays an extra three points to whichever player doesn't end up owning the resulting offspring.  At the end of the game, everybody gets a bonus equal to total smarts divided by ten, which mirrors the real-life fact that smart people don't capture all of the value they create for themselves and that smarter societies end up generally richer even for the nonsmart people in them; and remember, your in-game goal is to maximize your total points for yourself, not worry at all about how many points other players are scoring..."  Keltham goes on sketching some additional rules intended to make the game mirror real-life genetics, and real-life incentives under conditions of government subsidy.

Each player gets one free mating at the start of the game with the 'Thamkel' character, which is the only one that has any 'smart' or 'dumb' values at the fourth locus of each chromosome - all other characters have 'neutral' at the fourth locus.  'Thamkel' also has only neutral values at the first locus, where other characters can have 'smart' or 'dumb' there.  This reflects the way that Thamkel has some different varying bits.  You can't legally mate two Thamkel offspring during the first two generations afterwards.  There are a few other characters as smart as Thamkel, but only Thamkel has any variation in the fourth locus for each chromosome.

And then Keltham observes what happens under the new game conditions.  Do they - sort of get the point of the simulation, now, or the incentives?  At all?

Yes, once he spells it out they can follow how that's like the situation they're trying to produce, and they can try to fumble through it with all the new scoring rules in mind. They're...still not clear on how you learn from this how much Thamkel helps, aside from playing the game without Thamkel in it.

Well, yes, after they've played this game for a few generations, they're going to score themselves, and then play a new game without Thamkel in it.

Right. 

They're still evidently not having any fun but they do this very diligently, and get better at it over a couple of trials, and take down their scores. (They seem to find it hard not to track who has the highest score). 

He's seriously not trying to score them on their game scores!  First of all, people were given randomized characters with uneven qualities at the start of the game, just like life itself isn't fair in that way, and second, he's watching to see if people play according to their incentives, not whether they get higher scores than other players.

...he's not sure what's going wrong with the way that nobody's having fun.  Maybe just the sheer cognitive load associated with not really instinctively feeling what's going on and not having pre-learned brain patterns for this complicated game?  He should maybe back off of this soon, then, and only re-approach after playing simple games with shorter-term intrinsic real rewards associated with good play, like if they're doing it on a playground, or the adult equivalent of that.  He doesn't want to train his students that simulation games aren't fun.

"So this game wasn't a very realistic one - for example, you knew all of your characters' genetic information locus by locus, where, for a more realistic game, we should've had some Game Masters who only told you a character's total Intelligence score, didn't tell you anything about the specific loci, and generated new characters for everyone after each mating.  Nonetheless, for this unrealistic version, around how valuable was Thamkel?"

(Two games without Thamkel managed to take Intelligence from 10 to 12.2 and 12.8, over the prescribed number of generations.  Two games with Thamkel took Intelligence to 13.6 and 12.8.  It's frustrating how slowly the game seems to progress!)