This is our monthly place to discuss relevant topics that have not appeared in recent posts.
Robin, how did you first meet Eliezer? How did he end up writing on your blog?
Robin Hanson joined the Extropians email list at some time in or before 1994.
Eliezer Yudkowsky joined the extropians email list in 1996.
The extropians email list was largely populated by transhumanists and is where the cofounders of the Singularity Institute met.
My understanding is that much worry about a bad singularity was had in the emails and since then EY has focused on that while RH has come to think that a singularity is unlikely and that a civilization-ending evil intelligent AI originating from earth in the near-to-mid future is also unlikely.
If we can hold out for just over 8 more months, August of this year will mark 70 full years since the first (Aug. 6) and last (Aug. 9) applications of atomic weapons in combat. Amazing, no? Who thought that would happen?
You were inclined to think there would be nuclear war?
[Retrospectively, in light of what we now know the U.S. is capable of (unless it’s changed since then), what amazes me is the fact that the U.S. didn’t strike Russia pre-emptively. Once the Russians had nuclear weapons too, who would use them?]
Bertrand Russell & John von Neumann both advocated a first-strike, and the U.S had the capacity to win such a war, but the Kennedy & Johnson administrations deliberately allowed the Soviets to build a second-strike capability.
I guess it was a matter of pragmatism: a first strike would have killed millions of Soviets and probably lots of civilians in surrounding nations allied to the US as well and the image of the US would be forever damaged, no one would trust them ever again, a cold war with the British (who had access to most of the Manhattan project research) and French empires could have occurred among many other bad outcomes for the US.
Russia surrounded itself with a buffer of satellite states, so I don’t think there would have been much of a problem with allied states. And the U.S entered into a sort of “cold war” of dismantling those empires anyway.
The US did not have a vast nuclear arsenal in the days of its nuclear monopoly. A very limited number of bombs, in the low dozens, would have been available to wage such a war.
Given unreliable delivery (especially vs deep targets in Russia), and low yields, the nuclear weapons would have been supplemented by a sustained conventional bombing campaign prior to Soviet collapse, and US chiefs were not sanguine about ground operations in the intervening period.
The numbers obviously still greatly favoured the US, but it would have been the work of months or years. Difficult to justify in policy terms short of an act of outright Soviet aggression. Remember; the Cold War was not yet at its height and many elements of the western democracies remembered the USSR as an ally against fascism. Could the war effort have been reliably maintained for the necessary duration?
Actually, most historians and nuclear strategists point to the late 50’s, not 40’s, as the “years of maximum danger”; when both sides are incentivised to meet a crisis with a disarming first strike.
Also, the US was far too sanguine over Soviet bomb development; contemporary estimates thought that the Soviet A-bomb would be ~1960. Hence there was no 1940’s urgency to “deal” with the problem.
I was just re-reading your papers on hard steps in the development of intelligence, and had a comment and a question.
You state that the hard-step durations have an exponential distribution, but my analysis and simulations show that they are polynomial of order (#hard steps-1). Essentially, a single hard step occurrance is uniformly distributed over the available time range, as the chance of an earlier occurance preempting a later occurance is (by definition) very small. Thus, with multiple hard steps, you have a combinatorial situation where a small duration allows more options for the positioning of the other steps in the time period, hence a polynomial distribution. (This all ends up with the same expectations relative to the time period that you give in the paper).
Given that, I’m not sure how the model gives us any information. If we condition on all the major-looking events in the development of life & intelligence, the fact that any arrangement is equally likely just gives us back whatever original priors we had on what might or might not be hard steps. Likewise if we just start positing addition hard steps in the gaps.
We could condition on a smaller set of observations than the entire list, but other than using our current end boundary (our presence), and then stopping, I don’t see how one can do so in a consistent fashion. If we use the appearance of life and the appearance of intelligence, why not use the appearance of eukaryotes as well? If we use the minimum gap between hard steps, why not the maximum, the mean, the full arrangement?
I’m basically just asking how one gets unbiased and useful information out of the idea.
If/when you have a paper describing your analysis in detail, I’d be interested in seeing it. Hard to respond otherwise, as I don’t yet understand what you are doing.
Robin, go see Black Mirror Christmas episode if you haven’t already. It has ems in it.
That’s so low status
Prof Hanson, a few questions for you sir.
1. Regarding your “Great Filter” heuristic, should we expect to have received obvious alien broadcast media by now if there was not some filter? Any sense on if the odds are if the filter would be lack of broadcast technique development or development of hiding/erasure of broadcast ability before it reached modern earth?
2. Do you think the evidence (Rand Study) still points to that half of all medical spending is wasteful?
If the sky was full of aliens we shouldn’t expect to get hello messages, but we should expect the sky to look different. Yes I’d guess half of US medicine is wasteful.
No I mean broadcast noise. Where would the sphere of median broadcast noise carry if we were central tendency common? Should we see broadcast noise from 10 million alien civilizations? I’m asking that sort of probability question.
Boradcasts can be stopped by nebulas and such, but if the sky is full of aliens we could be visited by probes. Depending on your philosophy you either believe that aliens consider probes too expensive/uninteresting/they don’t want their probes to be visible to us or that the lack of visible/harmful probes means there are no aliens nearby.
I’d just like to thank this site for talking about signaling by way of misrepresenting one’s goals, and how this can lead to self-deception, because having this information allowed me to realize that the reason I wasn’t accomplishing a goal was because I didn’t want to.
Once I realized that, I changed the goal to a related goal that I did want to accomplish. Having no incentive to fail to accomplish the goal, I actually managed to accomplish it. This modification was beneficial to both myself and the people around me.
You were really useful to me, and I really appreciate it.
(Context is of low relevance but will be made available upon request.)
You are most welcome.
it might be interesting to hear your comments on this
Am reading it now.
Hi Robin. I know this is a January open thread, but it’s the last i’ve found. Sorry if I was not supposed to post here.
What do you think about growth, resource depletion and pessimistic views of the human future?
The more I read rational texts like the ones you post, the more I think humans are not going to make it.
… be a charity angel.