We humans have remarkable minds, minds more capable in many ways that in any other animal, or any artificial system so far created. Many give a lot of thought to the more capable artificial “super-intelligences” that we will likely create someday. But I’m more interested now in the “super-intelligences” that we already have: group minds.
"The IQ of a mob is equal to the average IQ of the people in the mob divided by the number of people in the mob." - My father, quoting something he once heard somewhere
It is hard to distinguish between the influence of formal organizations and that of mobs in determining social outcomes. A certain policy that formally was instituted by a hierarchically organized body—the national government—may, in fact, have been put in place only because it was the consensus of a mob. In the first instance, the relevant mob will likely be the citizens as a whole (the voters, if the nation is a democracy); but this mob consensus may have been formed under the influence of another mob consensus—that of some global elite. Because of these different levels of influence on outcomes, the process of collective decision-making is hard to grasp as a whole; and it is unclear whether society is now relying more on mob decision-making than in the past.
Robin seems to be a fan of concise terminology. I put together this brief guide to Hanson-speak: * Collective intelligence -> mob mind * Expansionist -> grabby * Senescence -> rot
A long time ago I speculated about the IQ of organisations and states and how to measure it: https://web.archive.org/web...
The ideas involved letting an agent of the group mind take a test (e.g. an IQ test). I didn't touch on speed but I agree speed degrades with group size. Or at least it trades off against some capabilities. For example while an agent could score perfectly on usual IQ test question when having access to a whole state this wouldn't work out if the time to answer questions were still limited.
Would be interesting to explore the degree to which the superiority of prediction markets over mobs into is explained by the availability of measurable metrics for the former. One could do this by creating a prediction market that mimics a mob by adjudication via Keynesian beauty contest as recently proposed by Scott Alexander (for other reasons). Any superiority left of the simulation over the real thing would be due to the prediction market itself, instead of metric availability.
"The IQ of a mob is equal to the average IQ of the people in the mob divided by the number of people in the mob." - My father, quoting something he once heard somewhere
It is hard to distinguish between the influence of formal organizations and that of mobs in determining social outcomes. A certain policy that formally was instituted by a hierarchically organized body—the national government—may, in fact, have been put in place only because it was the consensus of a mob. In the first instance, the relevant mob will likely be the citizens as a whole (the voters, if the nation is a democracy); but this mob consensus may have been formed under the influence of another mob consensus—that of some global elite. Because of these different levels of influence on outcomes, the process of collective decision-making is hard to grasp as a whole; and it is unclear whether society is now relying more on mob decision-making than in the past.
I am indeed a fan of the latter words/phrases over the former.
Robin seems to be a fan of concise terminology. I put together this brief guide to Hanson-speak: * Collective intelligence -> mob mind * Expansionist -> grabby * Senescence -> rot
Even when metrics are available, mobs don't tend to tie individual incentives to accuracy relative to those metrics. That is a key difference.
A long time ago I speculated about the IQ of organisations and states and how to measure it: https://web.archive.org/web...
The ideas involved letting an agent of the group mind take a test (e.g. an IQ test). I didn't touch on speed but I agree speed degrades with group size. Or at least it trades off against some capabilities. For example while an agent could score perfectly on usual IQ test question when having access to a whole state this wouldn't work out if the time to answer questions were still limited.
Would be interesting to explore the degree to which the superiority of prediction markets over mobs into is explained by the availability of measurable metrics for the former. One could do this by creating a prediction market that mimics a mob by adjudication via Keynesian beauty contest as recently proposed by Scott Alexander (for other reasons). Any superiority left of the simulation over the real thing would be due to the prediction market itself, instead of metric availability.
Yes; fixed.
Our world is thisI think you meant to write "thus".
Yes the space of possible minds is vast.
The recent move in the US towards the legalization of sports betting is a hopeful sign re prediction markets.
But artificial minds need not be closely modeled on existing group (or individual) minds: very different designs can be tried.