"Well, there's two components, I think, to my answer to that."

"The first answer is that outputs aren't always legible, and then you have to appropriately incentivize people's fairness on valuing the outputs.  In the version of the training game that the kids got, how much effort they had to put in wasn't fully legible, but the outcome of the game being won was visible and unmistakeable.  But suppose somebody is making a shoe; how good of a shoe is it exactly?  Maybe you could pay a trained third-party shoe-evaluator to come in and say exactly how many gold pieces they thought it would be worth, but measuring your output objectively like that is expensive.  What we have instead is the partially legible output of a shoe, where the quality of shoeparts or the evenness of the make or whatever it is that people value in Golarion shoes, might not be clear and objective to the point where the shoemaker and shoebuyer couldn't possibly agree on it.  So then they need to both reason in a way that incentivizes fairness from the other, without everything shattering with probability 1 in the presence of a small disagreement."

"Well, there's two components, I think, to my answer to that."

"The first answer is that outputs aren't always legible, and then you have to appropriately incentivize people's fairness on valuing the outputs.  In the version of the training game that the kids got, how much effort they had to put in wasn't fully legible, but the outcome of the game being won was visible and unmistakeable.  But suppose somebody is making a shoe; how good of a shoe is it exactly?  Maybe you could pay a trained third-party shoe-evaluator to come in and say exactly what fraction of a gold piece they thought it would be worth, but measuring your output objectively like that is expensive.  What we have instead is the partially legible output of a shoe, where the quality of shoeparts or the evenness of the make or whatever it is that people value in Golarion shoes, might not be clear and objective to the point where the shoemaker and shoebuyer couldn't possibly agree on it.  So then they need to both reason in a way that incentivizes fairness from the other, without everything shattering with probability 1 in the presence of a small disagreement."

"Well, there's two components, I think, to my answer to that."

"The first answer is that outputs aren't always legible, and then you have to appropriately incentivize people's fairness on valuing the outputs.  In the version of the training game that the kids got, how much effort they had to put in wasn't fully legible, but the outcome of the game being won was visible and unmistakeable.  But suppose somebody is making a shoe; how good of a shoe is it exactly?  Maybe you could pay a trained third-party shoe-evaluator to come in and say exactly what they thought it would be worth, but measuring your output objectively like that is expensive.  What we have instead is the partially legible output of a shoe, where the quality of shoeparts or the evenness of the make or whatever it is that people value in Golarion shoes, might not be clear and objective to the point where the shoemaker and shoebuyer couldn't possibly agree on it.  So then they need to both reason in a way that incentivizes fairness from the other, without everything shattering with probability 1 in the presence of a small disagreement."

"Well, there's two components, I think, to my answer to that."

"The first answer is that outputs aren't always legible, and then you have to appropriately incentivize people's fairness on valuing the outputs.  In the version of the training game that the kids got, how much effort they had to put in wasn't fully legible, but the outcome of the game being won was visible and unmistakeable.  But suppose somebody is making a shoe; how good of a shoe is it exactly?  Maybe you could pay a trained third-party shoe-evaluator to come in and say exactly what they thought it would be worth, but measuring your output objectively like that is expensive.  What we have instead is the partially legible output of a shoe, where the quality of shoeparts or the evenness of the make or whatever it is that people value in Golarion shoes, might not be clear and objective to the point where the shoemaker and shoebuyer couldn't possibly disagree on it.  So then they need to both reason in a way that incentivizes fairness from the other, without everything shattering with probability 1 in the presence of a small disagreement."