EA Forum Reader

Today is India's 76th Independence Day! I thought it would be nice to look at some of India's progress over the last three-fourths of a century:

Infant mortality has gone down by 13.48% since 1960.

The death rate from malaria has gone down from 15.78 deaths per 100,000 people in 1990 to 2.61 deaths per 100,000 in 2019.

The number of children whose development has been stunted (as a consequence of poor nutrition or repeated infections) has gone down from 66.04 million in 1990 to 49.26 million, a ~25% reduction.

The share of Indians living in extreme poverty has drastically reduced from 58.96% in 1981 to 10.55% in 2017.

We still have a long, long way to go in many domains (as one EA-aligned example, lead in paint and lead concentration in children is still a huge issue in India).

Source: Our World in Data

I think the relevance of Starlink here is not for serving individuals in LMICs, but for providing backhaul for towers in places too remote to easily backhaul via fiber or line of sight microwave links.

Considering formalizations of actualism, Jack Spencer, 2021, The procreative asymmetry and the impossibility of elusive permission (pdf here) discusses problems with actualism (especially "strong actualism" but also "weak actualism") and proposes "stable actualism" as a solution.

Is promotion to the Frontpage automated? Do mods take an approve by default and then demote stuff they think is not worth promoting?

I'm used to the LessWrong approach of blogpost by default, and mods manually approve promotion to the front page. I'm not sure I want to make the decision for the community if my post should be on the front page. But I do want it to be promoted to the front page if the mods approve it.

I feel like promotion to front page by default will disincentivise demotion to personal blog post if a post has high engagement irrespective of whether it's otherwise frontpage material.

(I feel a bit dense in asking this, but where are you getting ~200M in the first 8 months from? In the sheet it's 139.5M; the ~200M is a projected estimate.)

I was just eyeballing the graph! Thanks ,your notes made sense. Cool stuff.
 

Great post! I agree that standard setting could be useful. I think it could be especially important to set standards on how AI systems interact with animals and the natural environment, in addition to humans.

Idk, what are you trying to do with your illegible message?

If you're trying to get people to do technical research, then you probably just got them to work on a different version of the problem that isn't the one that actually mattered. You'd probably be better off targeting a smaller number of people with a legible message.

If you're trying to get public support for some specific regulation, then yes by all means go ahead with the illegible message (though I'd probably say the same thing even given longer timelines; you just don't get enough attention to convey the legible message).

TL;DR: Seems to depend on the action / theory of change more than timelines.

I think it's plausible that some interests matter in relative terms between possible outcomes, rather than only in terms that can be described absolutely. I think it can be the case that it's neither better nor worse in itself to have a specific preference at all, no matter how satisfied or frustrated, even though it's better for it to be more satisfied between two outcomes both in which it exists. Say a child's dream to go to the moon, or the wish of a specific person who can't walk to walk, or the wish to be with a loved one (e.g. grief over loss). I don't think taking away a frustrated preference makes someone better off in itself, except for certain kinds of preferences. I don't think adding a (satisfied) preference is ever good in itself.

Part of the reason might be that there's no natural unique 0 or neutral point, i.e. a single degree of preference satisfaction/frustration where we should be indifferent about having that preference at all. Or, at least, you can imagine degrees between perfectly satisfied and perfectly frustrated, but no natural way to set some partial satisfaction/frustration states on either side of 0.

Other common intuitions may violate IIA. We might say you're not obligated to make a great sacrifice for others, but if you are going to, it could be obligatory to do the most good with the same level of sacrifice (see this example [EA(p) · GW(p)] and the discussion in that thread). Similarly for having a child: you have no obligation to have one at all, and it may be permissible to have a child as long as they at least have a good life, but if you do have a child, and you could easily guarantee a much better life for them than just good, you may be obligated to do so. Frick discusses these as "conditional reasons".

I guess these reasons could apply similarly to transitivity. An important issue with intransitivity is that it's not clear what act to choose if each available option is beaten by another, but intransitive views can be turned into transitive views that violate IIA through voting methods, especially beatpath/Schulze, like in this paper.

Thinking back on this, rather than violating IIA, couldn't this just mean your order is not complete? Option 3 > Option 2, but neither is comparable to Option 1.

I suppose this violates a (permissibility) choice function version of IIA, but not an order-based definition.

I have a comment on specifically the part about AGI discourse.

Imagine there's a community that seems unusually concerned about the risk that some asteroid will soon hit the Earth and kill every person (you, your parents, your friends, your country members, members of every other country, etc.).

If I became aware of that community, the question I'd be most interested in answering is "are they correct about the asteroid?"

It wouldn't occur to me to make confident statements about how the community is biased or ill-intentioned in various ways before having assessed their object-level arguments.

Maybe if I could persuasively refute the arguments and it would be so easy to do that and I couldn't help but wonder how someone could think something so deeply flawed, maybe then would I engage in speculation about what might be going wrong with people's thinking. Even so, I'd still bother to put up a thorough refutation of their arguments at the start of my critique.