Well, it certainly might be an interesting experiment. I think it would be a good idea to try it out on one test case, then follow up with a serious post-mortem to come up with groundrules so as to do it even better next time.
The difficulty is to choose a topic which grabs the attention and engages the intellects of a large fraction of the people here. But something which is fresh enough that no one already has quasi-religious commitments to a position on the subject.
One idea that comes to mind is space exploration, in particular the exploration and exploitation of our own solar system in the absence of an AI singularity. There are sub-issues related to history, economics, psychology, and engineering. What ought to be the next step? Moon, Mars, asteroids, earth orbit? Do we send men, or robots? Do we dare to build a space elevator? Should we terraform Mars? Can we terraform Mars? Should we instead focus on colonizing and exploring the oceans of earth? What kinds of propulsion systems should we develop? Tethers? Fusion? Was Project Orion insanity or a missed opportunity which we should not miss a second time?
This is one area where I suspect we can acquire information that is as good as exists anywhere. It is an area that many of us have some interest in. But I’m not quite sure how many of us have enough time and interest to dig deep enough into the available information to really come up with viewpoints that are new, interesting, and also correct.
One problem that springs to mind is the origin of life. There’s mountains of evidence relating to this topic—and I expect a rational agent could figure out many of the important details if given the existing public evidence. I expect LessWrong would do badly on this topic—due to lack of knowledge.
Another topic is how—in practice—to implement an oracle—a.k.a. a resource-limited Solomonov induction agent. That plays to LessWrong strengths—and is based on a “neat” pure maths problem.
Well, it certainly might be an interesting experiment. I think it would be a good idea to try it out on one test case, then follow up with a serious post-mortem to come up with groundrules so as to do it even better next time.
The difficulty is to choose a topic which grabs the attention and engages the intellects of a large fraction of the people here. But something which is fresh enough that no one already has quasi-religious commitments to a position on the subject.
One idea that comes to mind is space exploration, in particular the exploration and exploitation of our own solar system in the absence of an AI singularity. There are sub-issues related to history, economics, psychology, and engineering. What ought to be the next step? Moon, Mars, asteroids, earth orbit? Do we send men, or robots? Do we dare to build a space elevator? Should we terraform Mars? Can we terraform Mars? Should we instead focus on colonizing and exploring the oceans of earth? What kinds of propulsion systems should we develop? Tethers? Fusion? Was Project Orion insanity or a missed opportunity which we should not miss a second time?
This is one area where I suspect we can acquire information that is as good as exists anywhere. It is an area that many of us have some interest in. But I’m not quite sure how many of us have enough time and interest to dig deep enough into the available information to really come up with viewpoints that are new, interesting, and also correct.
Re: Should we instead focus on colonizing and exploring the oceans of earth?
That is step 2 - according to Marshall Savage—and I agree.
One problem that springs to mind is the origin of life. There’s mountains of evidence relating to this topic—and I expect a rational agent could figure out many of the important details if given the existing public evidence. I expect LessWrong would do badly on this topic—due to lack of knowledge.
Another topic is how—in practice—to implement an oracle—a.k.a. a resource-limited Solomonov induction agent. That plays to LessWrong strengths—and is based on a “neat” pure maths problem.