Alexei
Karma: 2,481
NewTop
Of course, truly effective tea plus a well-conveyed story about its great properties will generate more sales than effective tea or a good story alone.
Sometimes not, I think. It’s almost like a measure of the efficiency / effectiveness of the given market. If the market is really good at recognizing reality, then you don’t need to tell a story. (Basic software libraries are like that: do they give the compute the right thing? If yes, then it’s good.) If the market is not at recognizing reality, then creating stories is often way cheaper than then doing the real thing. (And also transfers better across domains.)
Is there anything in the world that we know of that does alignment for something else? Can we say that humans are doing “coherent extrapolated volition” for evolution? Keeping in mind that under this view, evolution itself would evolve and change into something more complex and may be better.
I think there’s a bias when we consider optimizing for X’s values to consider only X without its environment. But the environment gave rise to X and actually much of X doesn’t make sense without it. So I think to some extent we also would need to find ways to do alignment on the environment itself. And that means to some extent helping evolution.
First off, great thought experiment! I like it, and it was a nice way to view the problem.
The most obvious answer is: “Wow, we sure don’t know how to help. Let’s design a smarter intelligence that’ll know how to help better.”
At that point I think we’re running the risk of passing the buck forever. (Unless we can prove that process terminates.) So we should probably do at least something. Instead of trying to optimize, I’d focus on doing things that are most obvious. Like helping it not to die. And making sure it has food.
Just wanted to say I’ve been reading all your recent posts. And I really like all of them, the ideas you lay out, and how you talk about them. You really make LW worth coming to almost every day. Thank you! FWIW, if you continue down this path, I could see you having your own SSC-sized community.
I love that game too. Thanks for the write up, it’s been fun to read your take on it. I haven’t played it recently, so didn’t know they added the ending. I’ll come back and play it again.
This will take a long time to load, but it’s comprehensive: https://arbital.com/explore/math/
I’m not sure when you tried. It works right now.
Hmm, I’m skeptical a barter system would work. I don’t think I’ve seen a successful implementation of it anywhere, though I do hear about people trying.
Yes, we’ve considered paying people, but that’s not scalable. (A good 3-5 page explanation might take 10 hours to write.)
It’s not open sourced.
The pages might take a while to load (up to 30 seconds).
I can imagine a neural-activation-like effect coming out of that, where frequently co-active posts naturally rise to the top of each other’s links and become threads or topics.
Not sure what you mean by this.
My guess for Wikipedia’s success is that they were one of the first; and there was more of a sense of an online community back then. Also it’s easier to create Wikipedia content than, say, a good explanation. StackOverflow succeeded because asking and answering questions is pretty easy, you get instant feedback, and they got community management right. (They solved exactly one problem well!) The founders were also really well known so it was easy for them to seed the platform.
I can’t open-source the platform as long as I’m doing the for-profit venture, since the platforms are too similar. However, if at some point I have to stop, then I’ll be happy to open source everything at that point.
I’m not into persuading people. :) If you want to write, go for it. I still think Arbital is a really good platform for writing up math explanations.
Not someone with sufficient authority, just the blog owner. That seems fair though. You can create you own blog and then you would be in charge of which comments to approve.
Yes, but that’s not “invite-only”.
How is it invite only? Are you talking about the comment section?
Originally the plan was to do exactly that if we couldn’t figure how to build a “joyful maze”: just throw open the doors and see what people do with it. Unfortunately there is still a significant amount of work left to do that well, and right now I’m more optimistic about the new platform than I am about scavenging the current version.
Here are like 5 pages explaining all the visions: https://arbital.com/p/more_about_arbital/
Basically what we tried is: “let’s figure out how people are supposed to have truth-seeking conversations, build a platform that facilities that, and then grow it.” Step 1 is very hard. Step 3 is made harder because your platform only attracts truth-seeking people.
New approach: “build a platform that facilities communication, grow it, then shape the ongoing discussion to be more truth-seeking.” Step 1 is still hard, but not made harder. Step 3 sounds a lot more doable.
Don’t use the outside view. Use your brain. If Arbital was confusing, but you didn’t look closer at it, then you didn’t use your brain. If MIRI seems confusing and you don’t look closer at it, then you aren’t using your brain. The whole concept of “smart people” whom you can trust with anything is just wrong. There are only niche experts.
My two cents: Arbital had very little to do with MIRI, aside from Arbital being Eliezer’s idea. But this was definitely out of his realm of expertise. MIRI/AI stuff is not.
It’s a complicated answer and also somewhat outside of the topic I want to discuss here (which is Arbital 1.0). For part of the answer see: http://lesswrong.com/lw/otq/whats_up_with_arbital/dqbc
Thanks for writing your thoughts out loud. Curious to follow your progress.