I'm ready to see how it goes. I think my concern about Swarm Intelligence lies in its use domains. So, for example, let's say that we have 100 people, each of whom likes one of 25 different flavors of ice cream. In another scenario, we might have 100 oncologists with an opinion about 25 different ways to treat a cancer. (Let's stipulate that in both cases, we're looking for one decision.) In which of these scenarios is SI likely to be of use? If it's not obvious, the scenarios are both designed to unpack the idea of yielding a good outcome. In the first case, the good outcome is subjective, so you can't show me a better way based on measurable outcomes or whatever. In the second case, the outcome might have some measurable, objective measures, such as treatment duration, survival rate, probability of complications, etc. I can see SI being pretty good for the latter but not the former. I don't care how many people like chocolate if I really prefer pralines 'n' cream. Matters get more difficult when people have moral-political preferences, such as what sort of society they want to live in -- or worse, they are Conformists. Even smart people can succumb to groupthink or become victims of an error contagion. But, more importantly, some things are just subjective or have enough subjectivity in the mix to muddy the waters. How does SI help in such cases? True decentralization means people self-select into niches that are more bespoke relative to their preferences. So I'm trying to imagine how SI helps when it appears to synthesize opinions about a "best" course of action, when in innumerable cases, we want either no course of action at all or a preferred course of action in which "best" is in the eye of the beholder.
We need to see your mind experimenting with us in Human Swarm Intelligence.
https://joshketry.substack.com/p/human-swarm-intelligence-the-most
I'm ready to see how it goes. I think my concern about Swarm Intelligence lies in its use domains. So, for example, let's say that we have 100 people, each of whom likes one of 25 different flavors of ice cream. In another scenario, we might have 100 oncologists with an opinion about 25 different ways to treat a cancer. (Let's stipulate that in both cases, we're looking for one decision.) In which of these scenarios is SI likely to be of use? If it's not obvious, the scenarios are both designed to unpack the idea of yielding a good outcome. In the first case, the good outcome is subjective, so you can't show me a better way based on measurable outcomes or whatever. In the second case, the outcome might have some measurable, objective measures, such as treatment duration, survival rate, probability of complications, etc. I can see SI being pretty good for the latter but not the former. I don't care how many people like chocolate if I really prefer pralines 'n' cream. Matters get more difficult when people have moral-political preferences, such as what sort of society they want to live in -- or worse, they are Conformists. Even smart people can succumb to groupthink or become victims of an error contagion. But, more importantly, some things are just subjective or have enough subjectivity in the mix to muddy the waters. How does SI help in such cases? True decentralization means people self-select into niches that are more bespoke relative to their preferences. So I'm trying to imagine how SI helps when it appears to synthesize opinions about a "best" course of action, when in innumerable cases, we want either no course of action at all or a preferred course of action in which "best" is in the eye of the beholder.
Thought provoking post. Being a fan of thought experiments, Caplan’s quote prompted me to search for his full article.
Thanks, I should have linked Bryan's piece: https://www.econlib.org/archives/2013/05/virtue_conformi.html
Will do now.