Here are my notes from the Optimal Philanthropy online meeting. Things in square brackets are my later additions and corrections. Let me know if there are factual errors or if I’ve incorrectly captured the drift of what you were saying.
Nick:
existential risk argument goes as follows:
small chance that people will be around for a really long time
if so, increasing probability of that happening by a small amount then really good
therefore, you should focus on this rather than things that have short-run impacts
focus only on things that reduce xrisk
Similar property: make things a little bit better for a really long time. Might be easier to do that.
name for these: trajectory change. Different from just economic growth – changes default trajectory along which things go
interesting for same reason as xrisk reduction but people haven’t talked much about them.
[sorry, didn’t note who said this]: education as an example. Particular kind of education?
Nick:
not really thought through this idea yet. [This idea is at the same stage] as if you’d just stumbled across xrisk concept
another example is just changing people’s values for a better
ordinary – e.g. campaigning for particular political party
Scott:
calls this “static” [I assume this means as in “noise” rather than as in “stationary”]
I’m doing something – tends to break down after a month or so, very difficult to predict long term results
should focus on things that can be predicted, build better models
Nick:
things that look short run could have long run impacts
e.g. thinking of buying bednets for people as having short-term impact – has short run impact today, no impact in 100 years time
but if you think about all downstream consequences, actually making small impact
Scott agrees.
both good and bad long-term consequences included in static
it’s a wash
Jeff:
what makes you think bednets have good long term consequences?
currently thinks of bednet as short term good, long term wash
benefits of having more people, having more healthy kids?
really hard to predict long term value
Nick:
Four classes of thing
things that reduce xrisk
things that speed up progress
things that make trajectory change
only short run effect
bednet not a trajectory change.
buying bednet increases economic growth – long term effects
economic growth has been compounding for a long time.
Plausible story for why this happens – people get richer, specialization, help each other out. Overall a positive process, and a process that ripples far into the future.
Jeff:
really hard to get good predictions. Maybe you could figure it out
even evaluating short-term effects of individual charity now is hard.
long term effects incredibly difficult
Nick:
when comparing within categories, makes sense to ignore these thing.
e.g. two things that contribute to economic growth – just ask which has better predictable impacts.
but if you’re comparing something that reduces xrisk with something that increases economic growth
Jeff:
funding education versus funding health.
wouldn’t surprise him if someone got handle on their long term effects and they were very different
Nick:
expects one of them to contribute to trajectory change more
hard to believe one of them has a very different effect on long term economic growth
e.g. education appeared on a par with health intervention, but expecting 1000x more economic growth seems wild.
Jeff:
[if optimizing for] economic growth – would do it very differently from doing health interventions
Nick:
can’t use economic growth argument to promote AMF: “AMF is best charity to
more plausible candidate: political or meta-research (some GiveWell thing)
At this point everyone on the call introduces themselves.
Giles Edkins: Toronto LW, THINK
Jeff Kaufman: married to Julia Wise. Both into earning to give stuff
Nick Beckstead: PHD in Philosophy at Rutgers. Population ethics and xrisks. Involved in Centre for Effective Altruism.
Nisan Stiennon: Berkeley. PHD in Math. Teaches for CFAR. Confused about whether to focus on xrisk or other kinds of philanthropy e.g. AMF
Peter Hurford: Ohio. Political Science & Psych. Joined GWWC – giving 10% of meager college income. New to smart giving stuff, on smart giving subreddit, on LW.
Raymond Arnold: does a lot of work for THINK. LW for past two years. General boat of xrisk & GiveWell stuff, only things he has a lot of info about
Scott Dickey: LWer, gives 25% of income to charity, trying to make that go as far as he can – SingInst, GiveWell. Following, reading them, watching where things are going
Boris Yakubchik: president of GWWC Rutgers. Giving 50% of income, high school math teacher. Don’t know what to expect from this hangout – needs to run off!
Ray:
assumes everyone has read GW, Nick Bostrom’s xrisk stuff
anyone read anything else?
Nick:
a lot of my dissertation is not empirical, instead it’s moral philosophy
has been thinking on the side on comparison between xrisk and other kinds of things, talking to people
feeling is there’s not huge amount of writing
e.g. economics of climate change – most similar to xrisk. Not a whole lot out there that’s helpful
Ray:
earlier this summer (interrupted by other things), wanted breadth-first look at potentially promising high impact stuff
3 books on global poverty, each with a different thesis
The End of Poverty – Jeffrey Sachs
White man’s burden – William Easterly
The Bottom Billion – [Paul Collier]
would be good for people in our community to start compiling summaries of this information
Giles:
GW prioritizes things that are easier to understand – targets beginners
Ray:
And where there’s good information
Jeff:
GW are reasonably honest in not just choosing charities that are accessible, but that the people involved are convinced about the best ones – has talked to Holden on Skype 8 months ago.
Ray:
Only recently with GW Labs have they made an effort to look at harder questions, not just looking at charities with easy-to-obtain information
Giles:
Any info on how GW Labs is going?
Ray:
100k to Cochrane. Not a whole lot of money! Still question on what else is out there
Peter:
Thought they were going to move away from GW Labs, were going to pursue this as main funding strategy
Scott:
Thought they were going to integrate things into one effort
Peter:
target big funders
away from health related issues
Scott:
[Should we] give to GiveWell itself?
Ray:
[If we give to GiveWell], a small amount to funding goes to themselves, the rest goes to their favorite charity
Peter:
can give to Clear Fund – fund for GW activities.
But they strongly discourage that – want to give to their top charity instead.
Giles:
can we talk to GW and find out why?
Jeff:
if you give to charity, tell them it’s because of GW
Ray:
has anyone talked to non-LWers about optimal philanthropy?
Jeff:
talked to college friends, not LWers. Some people think it makes sense – they are the people most similar to LWers, computer programmers and other engineers. Job makes them a bit more money than they know what to do with. “If I started giving away more than 20% of my money it wouldn’t hurt.” Helps people to think about it dispassionately.
Other people not inclined to this approach – doesn’t like quantifying. Shouldn’t be prioritising one thing over another
Scott:
talk to people at work, highly diverse set of backgrounds. Vast majority doesn’t like analytic side – they call me “cold”. Work fundraising drive – literally had picture of crying puppy. Very emotional, very community oriented, didn’t like analysis. Could be because I’m a bad arguer.
Jeff:
tries really hard, really hard to have discussions with people that people aren’t giving to best charity. People used to get mad at him. Now usually agrees to disagree but not losing any more friends
Ray:
requested charity Christmas gift – think about who you’re giving to. Conversation over Christmas dinner – worst idea ever! Grandmother most likely to donate, donated to local causes. She prioritizes giving locally not because she believes its the most effective but she prioritises her own community [did I get this right?]
nowadays asks what being a good person means to someone before having that conversation
has been having conversation at work not as “you should be doing it too” but “here’s this thing I’m doing”
has laid groundwork.
one person has said she now gives to more global causes [is this right?] as a result of what I was saying
long-term non-confrontational strategy
Scott:
another thing GW is doing right is targeting high-wealth individuals
spent a lot of time talking to my Dad about it. After 10 conversations he says “fine, I’ll give to what you think are more effective causes” – gave $100, and that was it forever.
Jeff:
convincing someone to give $10000 a year is not something to be sneezed at. Need to accept both the “give effectively” and “give substantially” ideas.
Jeff:
not many people [in effective altruism movement] older than I am. (late 30’s or higher)
Scott:
Bill Gates and Warren Buffett started own movement. Get rich people to give 50% of resources
Jaan Tallin – level 2 – looking at “malaria” instead of “human disease”
Jeff:
B&M Gates Foundation does a lot of stuff with malaria because it’s effective – same reason as GiveWell.
Scott: agrees
goes back to Nick’s thing of smaller things making larger impacts over time
not trying to change human condition
what inspired Scott: “what can you do today that people can remember your name 50000 years for now”
Ray:
if goal is to be remembered, won’t get there by donating. Founding agency or inventing something.
Scott:
Or Hitler, but would rather not be on that list.
Ray:
new ways of thinking – Darwin, Galileo.
optimizing giving is a different question
even “change the way governments work” – won’t be remembered for donating to that. But that’s OK.
agrees with sentiment though
Jeff:
certainly an inspiring sentiment
Ray:
what is the thing you’d want to remembered in 50000 years for, even if your name isn’t remembered?
Jeff:
some really important things don’t get remembered, e.g. eradicating Polio and Smallpox. Huge amounts of work, people don’t think about it because it’s no longer a problem.
Scott:
amend it to “should” be remembered for
Ray:
something I’ve been trying to do is put together a list of inspiring things that humans have done that we should care about.
evidence that things are getting better
ways of dealing with stress – global poverty is really depressing
“we should do more stuff like that”
polio & smallpox on list
Scott:
green revolution
one guy, saved or created billions of lives
Nick:
book: “millions saved” [Ruth Levine?], success stories in philanthropy in particular global health
Steven Pinker: “better angels of our nature, chapter 1”, things that went surprisingly well in last few decades
“Scientists greater than Einstein” [Billy Woodward?]
Ray:
80000hours blog
have read it occasionally
first few times, changed how I looked at that side of things
since then, haven’t found a whole lot of content
Scott:
stopped reading 80k
they deferred their choosing of charities to GW
[I had] already read up on income maximization
Jeff:
am subscribed to 80k blog
haven’t read things that are wholly new ideas that I’m excited to read
mostly been summaries of things
remaining questions are really hard. People won’t make good progress sitting down for an afternoon reading [writing?] a blog post
Scott:
waiting for Nick Bostrom to come up with magnum opus?
Jeff:
happy with what I’m doing so far
important not to get attached to really high leverage factors where it’s only worth it if impact is too huge to think about
good to realize how much good you can do with a GWWC-style 10% given to GW charities
and try and do better than that!
Ray:
Jeff: you and Julia are highest% giving couple that I know of. On the giving a lot side…
Jeff:
keeping a donations page, very numbersy
Bolder Giving talks about us as if it’s still 2009 and we have changes to make, titled Julia Wise and should be both of us…
Ray:
question is “are they living in Manhattan”?
Jeff:
rent in Boston is still high
studio apartment costing $1000 a month
even if we’d be spending twice as much, it wouldn’t have had a huge effect
actually, yes it would have a huge effect! [but could still give away large percentage]
Ray:
and do they have kids?
Jeff:
yes, planning to
yearly expenses will change by $10000 for each kid we have [note I may have written down Jeff’s numbers wrong]
living at parents house, pay them rent but it’s low
own house with kids: $40000 on ourselves, giving at least that much away, still pretty good
looking forward to say “yes, you can have kids and a house and still give away lots!”
people still might manage to find a reason why you’re weird and not want to emulate you?
have house & kids because you want them, not for signaling!
Jeff:
yes, of course!
more normal and conventional our life seems, convincing in a non-logical way
“oh, I could live a life like that. Not so different from the life I was thinking of living”.
Nick:
a few people had a question about how to compare xrisk reduction and GW charities
are there specific aspects of that that people want to know about?
Giles:
how do we know xrisk orgs are having any impact at all? How do we know they’re not making things worse?
Scott:
Yudkowksy’s scale of ambition. Went into hacker news post, restructured entire conversation where highest you could go was “bigger than apple”, he put that at 2. His 10 was “you can hack the computer universe is running on”
Ray:
expected value calculation – don’t need to know they [xrisk orgs] are doing something meaningful, 1% chance is good enough. But how can we even be sure of that?
without some calculation, running on intuition.
intuitively sounds plausible that donating to xrisk org is higher impact than giving to AMF
whole point of GW is that our intuitions aren’t necessarily reliable
what are the numbers for xrisk?
one thing that’s bothering him – people say xrisk, only think of SingInst, FHI
Scott:
Methuselah Foundation
Lifeboat Foundation (even they sound amateurish)
Ray:
SingInst is most reputable of the various xrisk orgs [what about FHI?], still has a lot of reasons to be skeptical of SingInst
changing a bit since Luke Muelhauser took the reins, moving in more reliable direction
Holden should target any other organization in the field [i.e. should address other xrisk orgs besides SingInst]
Jeff:
Holden’s post on Singinst was mostly organisation critique. Looking at what they’re doing, things they’re doing that seem to be working and not.
even if they passed all of those things, I would still be unsure it would make more sense [to donate] than an easily measurable intervention [such as AMF]
SingInst still doing incredibly hard to evaluate work
Ray:
asteroid impact – already solved/diminishing returns point
AI risk
nuclear proliferation
bioterrorism
CDC
[question came up on what other xrisk mitigation efforts might there be that we don’t know about, in particular AI related]
what secret but massively positive stuff would there be?
Scott:
Google cars, save 30000 lives a year, not too big. Bigger if roll out to world
at that point, Google is AI company and should be focusing on AI safety
Jeff:
people in Google mostly think that kind of thing is silly?
very far from machine learning systems that are scary
so hard to get them to do anything at all intelligent
hard for people to think about possibility of recursively self-improving anything.
not my impression that they are excited about AI safety.
Scott:
not saying that they are, but that they should be
keep a lot of it locked down
we don’t know what’s happening in China, Korea, Japan had a push for AI recently
Jeff:
wouldn’t be surprised if no-one at Google knows more about AI safety than what you can read on LW
Ray:
one thing that’s potentially high impact is donating to scientific research
there’s a thing called Petridish – Kickstarter for research projects
not optimized for “these are the projects that will help the world the most”
wonder if there’s a way we can push this in that direction?
Jeff:
amazing how animal-focused they are
specific to biology, or is that what people like to fund?
Giles:
find people good at communicating effective altruism message and put them in touch with orgs [such as Petridish] that need it?
Scott:
finding marketers
Ray:
some of that is in progress at Leverage. At pre-planning stage
Scott:
campaign Red or Pink, generate millions of dollars by having campaigns. Doesn’t seem to be having much impact, but can we co-opt or create something like that?
Ray:
am definitely looking for info on xrisk in a more concise form
“I don’t have a degree in computer science or AI research, wouldn’t be able to analyze at Eliezer level”
so layman can feel like they’re making an informed decision, has important information but not too much information
Notes from Online Optimal Philanthropy Meetup: 12-10-09
Here are my notes from the Optimal Philanthropy online meeting. Things in square brackets are my later additions and corrections. Let me know if there are factual errors or if I’ve incorrectly captured the drift of what you were saying.
Nick:
existential risk argument goes as follows:
small chance that people will be around for a really long time
if so, increasing probability of that happening by a small amount then really good
therefore, you should focus on this rather than things that have short-run impacts
focus only on things that reduce xrisk
Similar property: make things a little bit better for a really long time. Might be easier to do that.
name for these: trajectory change. Different from just economic growth – changes default trajectory along which things go
interesting for same reason as xrisk reduction but people haven’t talked much about them.
[sorry, didn’t note who said this]: education as an example. Particular kind of education?
Nick:
not really thought through this idea yet. [This idea is at the same stage] as if you’d just stumbled across xrisk concept
another example is just changing people’s values for a better
ordinary – e.g. campaigning for particular political party
Scott:
calls this “static” [I assume this means as in “noise” rather than as in “stationary”]
I’m doing something – tends to break down after a month or so, very difficult to predict long term results
should focus on things that can be predicted, build better models
Nick:
things that look short run could have long run impacts
e.g. thinking of buying bednets for people as having short-term impact – has short run impact today, no impact in 100 years time
but if you think about all downstream consequences, actually making small impact
Scott agrees.
both good and bad long-term consequences included in static
it’s a wash
Jeff:
what makes you think bednets have good long term consequences?
currently thinks of bednet as short term good, long term wash
benefits of having more people, having more healthy kids?
really hard to predict long term value
Nick:
Four classes of thing
things that reduce xrisk
things that speed up progress
things that make trajectory change
only short run effect
bednet not a trajectory change.
buying bednet increases economic growth – long term effects
economic growth has been compounding for a long time.
Plausible story for why this happens – people get richer, specialization, help each other out. Overall a positive process, and a process that ripples far into the future.
Jeff:
really hard to get good predictions. Maybe you could figure it out
even evaluating short-term effects of individual charity now is hard.
long term effects incredibly difficult
Nick:
when comparing within categories, makes sense to ignore these thing.
e.g. two things that contribute to economic growth – just ask which has better predictable impacts.
but if you’re comparing something that reduces xrisk with something that increases economic growth
Jeff:
funding education versus funding health.
wouldn’t surprise him if someone got handle on their long term effects and they were very different
Nick:
expects one of them to contribute to trajectory change more
hard to believe one of them has a very different effect on long term economic growth
e.g. education appeared on a par with health intervention, but expecting 1000x more economic growth seems wild.
Jeff:
[if optimizing for] economic growth – would do it very differently from doing health interventions
Nick:
can’t use economic growth argument to promote AMF: “AMF is best charity to
more plausible candidate: political or meta-research (some GiveWell thing)
At this point everyone on the call introduces themselves.
Giles Edkins: Toronto LW, THINK
Jeff Kaufman: married to Julia Wise. Both into earning to give stuff
Nick Beckstead: PHD in Philosophy at Rutgers. Population ethics and xrisks. Involved in Centre for Effective Altruism.
Nisan Stiennon: Berkeley. PHD in Math. Teaches for CFAR. Confused about whether to focus on xrisk or other kinds of philanthropy e.g. AMF
Peter Hurford: Ohio. Political Science & Psych. Joined GWWC – giving 10% of meager college income. New to smart giving stuff, on smart giving subreddit, on LW.
Raymond Arnold: does a lot of work for THINK. LW for past two years. General boat of xrisk & GiveWell stuff, only things he has a lot of info about
Scott Dickey: LWer, gives 25% of income to charity, trying to make that go as far as he can – SingInst, GiveWell. Following, reading them, watching where things are going
Boris Yakubchik: president of GWWC Rutgers. Giving 50% of income, high school math teacher. Don’t know what to expect from this hangout – needs to run off!
Ray:
assumes everyone has read GW, Nick Bostrom’s xrisk stuff
anyone read anything else?
Nick:
a lot of my dissertation is not empirical, instead it’s moral philosophy
has been thinking on the side on comparison between xrisk and other kinds of things, talking to people
feeling is there’s not huge amount of writing
e.g. economics of climate change – most similar to xrisk. Not a whole lot out there that’s helpful
Ray:
earlier this summer (interrupted by other things), wanted breadth-first look at potentially promising high impact stuff
3 books on global poverty, each with a different thesis
The End of Poverty – Jeffrey Sachs
White man’s burden – William Easterly
The Bottom Billion – [Paul Collier]
would be good for people in our community to start compiling summaries of this information
Giles:
GW prioritizes things that are easier to understand – targets beginners
Ray:
And where there’s good information
Jeff:
GW are reasonably honest in not just choosing charities that are accessible, but that the people involved are convinced about the best ones – has talked to Holden on Skype 8 months ago.
Ray:
Only recently with GW Labs have they made an effort to look at harder questions, not just looking at charities with easy-to-obtain information
Giles:
Any info on how GW Labs is going?
Ray:
100k to Cochrane. Not a whole lot of money! Still question on what else is out there
Peter:
Thought they were going to move away from GW Labs, were going to pursue this as main funding strategy
Scott:
Thought they were going to integrate things into one effort
Peter:
target big funders
away from health related issues
Scott:
[Should we] give to GiveWell itself?
Ray:
[If we give to GiveWell], a small amount to funding goes to themselves, the rest goes to their favorite charity
Peter:
can give to Clear Fund – fund for GW activities.
But they strongly discourage that – want to give to their top charity instead.
Giles:
can we talk to GW and find out why?
Jeff:
if you give to charity, tell them it’s because of GW
Ray:
has anyone talked to non-LWers about optimal philanthropy?
Jeff:
talked to college friends, not LWers. Some people think it makes sense – they are the people most similar to LWers, computer programmers and other engineers. Job makes them a bit more money than they know what to do with. “If I started giving away more than 20% of my money it wouldn’t hurt.” Helps people to think about it dispassionately.
Other people not inclined to this approach – doesn’t like quantifying. Shouldn’t be prioritising one thing over another
Scott:
talk to people at work, highly diverse set of backgrounds. Vast majority doesn’t like analytic side – they call me “cold”. Work fundraising drive – literally had picture of crying puppy. Very emotional, very community oriented, didn’t like analysis. Could be because I’m a bad arguer.
Jeff:
tries really hard, really hard to have discussions with people that people aren’t giving to best charity. People used to get mad at him. Now usually agrees to disagree but not losing any more friends
Ray:
requested charity Christmas gift – think about who you’re giving to. Conversation over Christmas dinner – worst idea ever! Grandmother most likely to donate, donated to local causes. She prioritizes giving locally not because she believes its the most effective but she prioritises her own community [did I get this right?]
nowadays asks what being a good person means to someone before having that conversation
has been having conversation at work not as “you should be doing it too” but “here’s this thing I’m doing”
has laid groundwork.
one person has said she now gives to more global causes [is this right?] as a result of what I was saying
long-term non-confrontational strategy
Scott:
another thing GW is doing right is targeting high-wealth individuals
spent a lot of time talking to my Dad about it. After 10 conversations he says “fine, I’ll give to what you think are more effective causes” – gave $100, and that was it forever.
Jeff:
convincing someone to give $10000 a year is not something to be sneezed at. Need to accept both the “give effectively” and “give substantially” ideas.
Jeff:
not many people [in effective altruism movement] older than I am. (late 30’s or higher)
Scott:
Bill Gates and Warren Buffett started own movement. Get rich people to give 50% of resources
Jaan Tallin – level 2 – looking at “malaria” instead of “human disease”
Jeff:
B&M Gates Foundation does a lot of stuff with malaria because it’s effective – same reason as GiveWell.
Scott: agrees
goes back to Nick’s thing of smaller things making larger impacts over time
not trying to change human condition
what inspired Scott: “what can you do today that people can remember your name 50000 years for now”
Ray:
if goal is to be remembered, won’t get there by donating. Founding agency or inventing something.
Scott:
Or Hitler, but would rather not be on that list.
Ray:
new ways of thinking – Darwin, Galileo.
optimizing giving is a different question
even “change the way governments work” – won’t be remembered for donating to that. But that’s OK.
agrees with sentiment though
Jeff:
certainly an inspiring sentiment
Ray:
what is the thing you’d want to remembered in 50000 years for, even if your name isn’t remembered?
Jeff:
some really important things don’t get remembered, e.g. eradicating Polio and Smallpox. Huge amounts of work, people don’t think about it because it’s no longer a problem.
Scott:
amend it to “should” be remembered for
Ray:
something I’ve been trying to do is put together a list of inspiring things that humans have done that we should care about.
evidence that things are getting better
ways of dealing with stress – global poverty is really depressing
“we should do more stuff like that”
polio & smallpox on list
Scott:
green revolution
one guy, saved or created billions of lives
Nick:
book: “millions saved” [Ruth Levine?], success stories in philanthropy in particular global health
Steven Pinker: “better angels of our nature, chapter 1”, things that went surprisingly well in last few decades
“Scientists greater than Einstein” [Billy Woodward?]
Ray:
80000hours blog
have read it occasionally
first few times, changed how I looked at that side of things
since then, haven’t found a whole lot of content
Scott:
stopped reading 80k
they deferred their choosing of charities to GW
[I had] already read up on income maximization
Jeff:
am subscribed to 80k blog
haven’t read things that are wholly new ideas that I’m excited to read
mostly been summaries of things
remaining questions are really hard. People won’t make good progress sitting down for an afternoon reading [writing?] a blog post
Scott:
waiting for Nick Bostrom to come up with magnum opus?
Jeff:
happy with what I’m doing so far
important not to get attached to really high leverage factors where it’s only worth it if impact is too huge to think about
good to realize how much good you can do with a GWWC-style 10% given to GW charities
and try and do better than that!
Ray:
Jeff: you and Julia are highest% giving couple that I know of. On the giving a lot side…
Jeff:
keeping a donations page, very numbersy
Bolder Giving talks about us as if it’s still 2009 and we have changes to make, titled Julia Wise and should be both of us…
Ray:
question is “are they living in Manhattan”?
Jeff:
rent in Boston is still high
studio apartment costing $1000 a month
even if we’d be spending twice as much, it wouldn’t have had a huge effect
actually, yes it would have a huge effect! [but could still give away large percentage]
Ray:
and do they have kids?
Jeff:
yes, planning to
yearly expenses will change by $10000 for each kid we have [note I may have written down Jeff’s numbers wrong]
living at parents house, pay them rent but it’s low
own house with kids: $40000 on ourselves, giving at least that much away, still pretty good
looking forward to say “yes, you can have kids and a house and still give away lots!”
Giles:
“is that your true reason for rejection”
people still might manage to find a reason why you’re weird and not want to emulate you?
have house & kids because you want them, not for signaling!
Jeff:
yes, of course!
more normal and conventional our life seems, convincing in a non-logical way
“oh, I could live a life like that. Not so different from the life I was thinking of living”.
Nick:
a few people had a question about how to compare xrisk reduction and GW charities
are there specific aspects of that that people want to know about?
Giles:
how do we know xrisk orgs are having any impact at all? How do we know they’re not making things worse?
Scott:
Yudkowksy’s scale of ambition. Went into hacker news post, restructured entire conversation where highest you could go was “bigger than apple”, he put that at 2. His 10 was “you can hack the computer universe is running on”
Ray:
expected value calculation – don’t need to know they [xrisk orgs] are doing something meaningful, 1% chance is good enough. But how can we even be sure of that?
without some calculation, running on intuition.
intuitively sounds plausible that donating to xrisk org is higher impact than giving to AMF
whole point of GW is that our intuitions aren’t necessarily reliable
what are the numbers for xrisk?
one thing that’s bothering him – people say xrisk, only think of SingInst, FHI
Scott:
Methuselah Foundation
Lifeboat Foundation (even they sound amateurish)
Ray:
SingInst is most reputable of the various xrisk orgs [what about FHI?], still has a lot of reasons to be skeptical of SingInst
changing a bit since Luke Muelhauser took the reins, moving in more reliable direction
[These didn’t all come up in the discussion, but I’ll give lukeprog’s list of x-risk orgs: FHI, FutureTech, the Singularity Institute, Leverage Research, the Global Catastrophic Risk Institute, CSER]
Scott:
Growing pains
Ray:
OK, I can see SingInst is improving. Not at point I’d be willing to donate to them yet
set up criteria: if they meet these criteria, I’d consider them reliable
Giles:
Holden done some of the groundwork for that?
Ray:
part of what influenced me
Nick:
do you wish there was more like the Holden post?
Scott:
Holden should target any other organization in the field [i.e. should address other xrisk orgs besides SingInst]
Jeff:
Holden’s post on Singinst was mostly organisation critique. Looking at what they’re doing, things they’re doing that seem to be working and not.
even if they passed all of those things, I would still be unsure it would make more sense [to donate] than an easily measurable intervention [such as AMF]
SingInst still doing incredibly hard to evaluate work
Ray:
asteroid impact – already solved/diminishing returns point
AI risk
nuclear proliferation
bioterrorism
CDC
[question came up on what other xrisk mitigation efforts might there be that we don’t know about, in particular AI related]
Scott:
Google, Microsoft Research
Jeff:
Google mostly public with google.org stuff
what secret but massively positive stuff would there be?
Scott:
Google cars, save 30000 lives a year, not too big. Bigger if roll out to world
at that point, Google is AI company and should be focusing on AI safety
Jeff:
people in Google mostly think that kind of thing is silly?
very far from machine learning systems that are scary
so hard to get them to do anything at all intelligent
hard for people to think about possibility of recursively self-improving anything.
not my impression that they are excited about AI safety.
Scott:
not saying that they are, but that they should be
keep a lot of it locked down
we don’t know what’s happening in China, Korea, Japan had a push for AI recently
Jeff:
wouldn’t be surprised if no-one at Google knows more about AI safety than what you can read on LW
Ray:
one thing that’s potentially high impact is donating to scientific research
there’s a thing called Petridish – Kickstarter for research projects
not optimized for “these are the projects that will help the world the most”
wonder if there’s a way we can push this in that direction?
Jeff:
amazing how animal-focused they are
specific to biology, or is that what people like to fund?
Giles:
find people good at communicating effective altruism message and put them in touch with orgs [such as Petridish] that need it?
Scott:
finding marketers
Ray:
some of that is in progress at Leverage. At pre-planning stage
Scott:
campaign Red or Pink, generate millions of dollars by having campaigns. Doesn’t seem to be having much impact, but can we co-opt or create something like that?
Ray:
am definitely looking for info on xrisk in a more concise form
“I don’t have a degree in computer science or AI research, wouldn’t be able to analyze at Eliezer level”
so layman can feel like they’re making an informed decision, has important information but not too much information