I expect of course that no good justification shall be forthcoming of why GF didn’t fund the Against Malaria Foundation with a casual wave of their hands,
There is a good justification: they have good empirical evidence that the Gates Foundation stamp of approval attracts third-party funding (governments, other foundations, and sometimes small donors), which causes diminishing returns (in addition to Gates funding being large enough to produce diminishing returns in general, and being most useful for thick concentrated start-up funding, which AMF does not need as a provider of funding to other organizations doing bednet distribution). GiveWell also seems to agree that the GF gives to many opportunities better than AMF that are not available or easily parsed by small donors, and that other GF public health projects are not radically worse on average in expectation than AMF.
which puts them far, far ahead of a Gates Foundation which AFAICT picks a measurable cause at emotional whim and then optimizes within that cause.
Internally, they do use DALY-like measures at the high levels, and use them in thinking about different kinds of projects for ballpark estimates and setting thresholds for action (junior-level employees work to implement). Also like GiveWell they take uncertainty about cause sign and magnitude into account, which boosts the relative virtue of picking out projects that are clearly promising within their field.
Why keep the reasoning of your effective altruism a secret?
Some possibilities:
The ‘seal of approval’ does almost as much to attract funding to target charities as more detailed explanations;
The press releases explain the core case
The Gates Foundation funds and participates in roundtables, conferences, and other academic and nonprofit venues to convey its thinking, and commissions a lot of external work making information available (e.g. they fund the DCPP cost-effectiveness estimates)
The Gates Foundation has many employees working on small projects, and doesn’t want to constantly produce public-facing substantive claims piecemeal which might draw it into controversy
The full case for even good interventions may depend in part on sensitive issues like judgments about particular people
You haven’t been reading what the GF does put out itself (or the public info of the initiatives it supports, or the research they sponsor for public consumption); have you been reading the detailed GiveWell reports, so as to be able to make a comparison?
More generally, what would make you update towards “People are crazy, the world is mad”? When in many cases such as this I see no evidence that the world is sane, I update towards madness.
Yet I see you surprised by the higher-than-expected competence of elites more often and severely than I see the reverse, e.g.:
That there are good reasons of legal predictability for the role of precedent in common law
The ability of good math and computer science students to grasp something you identified as an FAI issue at first glance
The judgments of physicists about nuclear chain reactions
The capabilities of math, AI, and hard science elites
The ability of venture capitalists and scientists to tell in 1999 that Drexlerian nanotechnology was not likely to be developed and lead to a world-wrecking war by 2015, and extremely unlikely by 2003; and this based on good empirical record of past technologies and the lead time in expert opinion and prototypes/precursor technologies
It looks to me like “people are crazy, the world is mad” has lead you astray repeatedly, but I haven’t seen as many successes. What are some of the major predictive successes of “the world is mad” that held up under careful investigation of dispositive facts?
...none of that sounds like an explanation for why Gates hasn’t funded AMF. Shouldn’t that make it even easier for them to wave their hands? How does GF’s inaction produce good consequences here?
Gates kickstarted GiveWell’s previous top charity, VillageReach, with several million dollars before GiveWell got to it. VillageReach was relatively innovative and could use startup funding. AMF is not innovative in that way, but one out of a number of organizations that pay for bednets to be distributed by other organizations in developing countries.
GiveWell says that they wound up with AMF because other seemingly more promising interventions were low-hanging fruit that had been plucked (frequently by Gates) and were low on RFMF.
The Gates Foundation has funded much larger malaria net distribution funding programs, and encouraged donors to give through the consolidated Global Fund and Nothing But Nets. Like AMF, these provide funding to local net distribution partners. It has also engaged in plausibly more highly leveraged enabling projects, like research into effective distribution of nets and better treatments.
Now that bednet distribution is widespread and rapidly scaling up independently of Gates efforts (although in meaningful part thanks to its earlier interventions) they are working on vaccines, diagnostics, drugs, evaluations, advocacy, and matching schemes with other large funders.
The Gates Foundation actors:
Made clear their support for bed nets
Encouraged individual donations to bednets through larger vehicles than AMF
Focus on better/more leveraged malaria-control expenditures to spend on than bednets (and GiveWell seems to agree, see their blog posts on GiveWell Labs and large donors) at this time, with strong growth in non-Gates bednet spending
May notice that AMF is small, and its advantages over other net funders are soft, and perhaps nonexistent (GiveWell has repeatedly downgraded its effectiveness estimates for AMF, and has changed its recommended charities based on past mistakes previously)
Zooming out for the bigger picture: the Gates Foundation seems to be plucking the large low-hanging fruit. GiveWell has been searching through the cracks to find small missed opportunities, and finding it quite challenging.
Compare with asteroid risk: one can make a lot of complaints about insufficient attention to the x-risk impact of asteroids relative to mundane harm, etc, but governments have still solved 90%+ of the problem. It’s good to look for the further opportunities for improvement, but one shouldn’t lose the forest for the trees: the problem was largely solved.
I agree with the content and spirit of this comment — thanks for writing it.
There remains the puzzle of why the Gates Foundation has devoted so many resources toward education efforts, which look to be ineffective from the outside. I have high confidence that they could have found a more effective use of the money.
I have wondered too, and I am much less impressed with the education than health work. It’s possible they are just buying their “help my society” and “help people anywhere.” They also have programs to help the Washington area around Microsoft, so it’s pretty plausible that they feel they want to discharge several kinds of moral obligations to concentric circles of connectedness each of which gets some weight.
My steelman (which I don’t necessarily buy):
Education is a $1.1 trillion sector in the United States alone. Improvements in its productivity will therefore be a big input into economic growth, which affects our ability to do everything else. Moreover, the lifetime impacts of education are significant on worker productivity, and developing and harnessing human capital, which affects the economy, but also public policy and science.
There is very little competition in the sector. Effects of schooling on learning and productivity pay off decades later, and the science of efficacious teaching is ill-developed, so parents are rather imperfectly able to and motivated to improve student outcomes. Moreover, a substantial portion of the establishment in education research has been resistant to the use of randomized trials and the scientific method in education, for a variety of reasons. Powerful interest groups resist experimentation and the adjustment of policies to the current available evidence.
However, the history of philanthropy suggests that wealthy philanthropists can and have had large effects on educational policy and practices. So the opportunity to institute more systematic data collection and conduct a number of major experiments in educational outcomes, and shape policy around the results is one of the more promising ways to increase rich country GDP and virtues, with all the relevant flow-through effects.
The small schools fiasco involved putting too much money into that experiment prematurely, and a real perhaps-statistical blunder, but this was acknowledged and the program dropped in response to poor results.
This is offset by projects like videotaping vast numbers of hours of teacher teaching (correlated with outcomes), the creation of large national databases, causing several jurisdictions to experiment with pay-for-performance (by bankrolling the difference), etc. The chance of a huge win from this extension of science and experimentation is enough to justify things.
The GF also gives a bunch of scholarships to high-ability students (the Gates scholarships at Cambridge modeled after the Rhodes Scholarships, the Gates Millennium Scholars Program for top under-represented minority students in the United States). Again this might not be a utilitarian thing, but such programs provide a way to target talent that might otherwise be lost from key fields, and a huge opportunity for influence by handing out the money to people doing research and work that the GF wants to encourage.
It could be largely symbolic, e.g. the Gates Foundation paying attention to education reaffirms the status of education as an important thing to pay attention to.
I, for one, would love to read your response to Carl’s question about your “world is mad” thesis:
It looks to me like “people are crazy, the world is mad” has lead you astray repeatedly, but I haven’t seen as many successes. What are some of the major predictive successes of “the world is mad” that held up under careful investigation of dispositive facts?
Off the instant top of my head, central line infections and the European Central Bank. I’m busy working on HPMOR and can’t really take the time to consult my Freemind map for the top dozen items. Carl’s list does seem kinda lopsided to me (i.e. not representative), but again, got to make the update deadline on the 29th and all my energy’s going there.
Yes but I might need to be reprodded come July 16th which is when my schedule quiets down again. Having a general debate on whether the world is mad doesn’t seem like a particularly good thing to recurse on deep in a comment thread.
I currently have two ways to do this. One is RTM and the other is Boomerang. I use RTM for reminders to do something in the future where it’s not particularly important exactly when I do them and Boomerang for reminders to do something in the future at a particular time, e.g. if I know I’m going to be in a particular location around 3:00pm next Tuesday and I’d like to be reminded to do something at that location I’ll Boomerang an email to myself for that time. (Or, if someone emailed me about this thing, I can Boomerang that email instead.)
There is a good justification: they have good empirical evidence that the Gates Foundation stamp of approval attracts third-party funding (governments, other foundations, and sometimes small donors), which causes diminishing returns (in addition to Gates funding being large enough to produce diminishing returns in general, and being most useful for thick concentrated start-up funding, which AMF does not need as a provider of funding to other organizations doing bednet distribution). GiveWell also seems to agree that the GF gives to many opportunities better than AMF that are not available or easily parsed by small donors, and that other GF public health projects are not radically worse on average in expectation than AMF.
Internally, they do use DALY-like measures at the high levels, and use them in thinking about different kinds of projects for ballpark estimates and setting thresholds for action (junior-level employees work to implement). Also like GiveWell they take uncertainty about cause sign and magnitude into account, which boosts the relative virtue of picking out projects that are clearly promising within their field.
Some possibilities:
The ‘seal of approval’ does almost as much to attract funding to target charities as more detailed explanations;
The press releases explain the core case
The Gates Foundation funds and participates in roundtables, conferences, and other academic and nonprofit venues to convey its thinking, and commissions a lot of external work making information available (e.g. they fund the DCPP cost-effectiveness estimates)
The Gates Foundation has many employees working on small projects, and doesn’t want to constantly produce public-facing substantive claims piecemeal which might draw it into controversy
The full case for even good interventions may depend in part on sensitive issues like judgments about particular people
You haven’t been reading what the GF does put out itself (or the public info of the initiatives it supports, or the research they sponsor for public consumption); have you been reading the detailed GiveWell reports, so as to be able to make a comparison?
Yet I see you surprised by the higher-than-expected competence of elites more often and severely than I see the reverse, e.g.:
That there are good reasons of legal predictability for the role of precedent in common law
The ability of good math and computer science students to grasp something you identified as an FAI issue at first glance
The judgments of physicists about nuclear chain reactions
The intelligence of business elites
The capabilities of math, AI, and hard science elites
The ability of venture capitalists and scientists to tell in 1999 that Drexlerian nanotechnology was not likely to be developed and lead to a world-wrecking war by 2015, and extremely unlikely by 2003; and this based on good empirical record of past technologies and the lead time in expert opinion and prototypes/precursor technologies
It looks to me like “people are crazy, the world is mad” has lead you astray repeatedly, but I haven’t seen as many successes. What are some of the major predictive successes of “the world is mad” that held up under careful investigation of dispositive facts?
...none of that sounds like an explanation for why Gates hasn’t funded AMF. Shouldn’t that make it even easier for them to wave their hands? How does GF’s inaction produce good consequences here?
Gates kickstarted GiveWell’s previous top charity, VillageReach, with several million dollars before GiveWell got to it. VillageReach was relatively innovative and could use startup funding. AMF is not innovative in that way, but one out of a number of organizations that pay for bednets to be distributed by other organizations in developing countries.
GiveWell says that they wound up with AMF because other seemingly more promising interventions were low-hanging fruit that had been plucked (frequently by Gates) and were low on RFMF.
The Gates Foundation has funded much larger malaria net distribution funding programs, and encouraged donors to give through the consolidated Global Fund and Nothing But Nets. Like AMF, these provide funding to local net distribution partners. It has also engaged in plausibly more highly leveraged enabling projects, like research into effective distribution of nets and better treatments.
Now that bednet distribution is widespread and rapidly scaling up independently of Gates efforts (although in meaningful part thanks to its earlier interventions) they are working on vaccines, diagnostics, drugs, evaluations, advocacy, and matching schemes with other large funders.
The Gates Foundation actors:
Made clear their support for bed nets
Encouraged individual donations to bednets through larger vehicles than AMF
Focus on better/more leveraged malaria-control expenditures to spend on than bednets (and GiveWell seems to agree, see their blog posts on GiveWell Labs and large donors) at this time, with strong growth in non-Gates bednet spending
May notice that AMF is small, and its advantages over other net funders are soft, and perhaps nonexistent (GiveWell has repeatedly downgraded its effectiveness estimates for AMF, and has changed its recommended charities based on past mistakes previously)
Zooming out for the bigger picture: the Gates Foundation seems to be plucking the large low-hanging fruit. GiveWell has been searching through the cracks to find small missed opportunities, and finding it quite challenging.
Compare with asteroid risk: one can make a lot of complaints about insufficient attention to the x-risk impact of asteroids relative to mundane harm, etc, but governments have still solved 90%+ of the problem. It’s good to look for the further opportunities for improvement, but one shouldn’t lose the forest for the trees: the problem was largely solved.
I agree with the content and spirit of this comment — thanks for writing it.
There remains the puzzle of why the Gates Foundation has devoted so many resources toward education efforts, which look to be ineffective from the outside. I have high confidence that they could have found a more effective use of the money.
I have wondered too, and I am much less impressed with the education than health work. It’s possible they are just buying their “help my society” and “help people anywhere.” They also have programs to help the Washington area around Microsoft, so it’s pretty plausible that they feel they want to discharge several kinds of moral obligations to concentric circles of connectedness each of which gets some weight.
My steelman (which I don’t necessarily buy):
Education is a $1.1 trillion sector in the United States alone. Improvements in its productivity will therefore be a big input into economic growth, which affects our ability to do everything else. Moreover, the lifetime impacts of education are significant on worker productivity, and developing and harnessing human capital, which affects the economy, but also public policy and science.
There is very little competition in the sector. Effects of schooling on learning and productivity pay off decades later, and the science of efficacious teaching is ill-developed, so parents are rather imperfectly able to and motivated to improve student outcomes. Moreover, a substantial portion of the establishment in education research has been resistant to the use of randomized trials and the scientific method in education, for a variety of reasons. Powerful interest groups resist experimentation and the adjustment of policies to the current available evidence.
However, the history of philanthropy suggests that wealthy philanthropists can and have had large effects on educational policy and practices. So the opportunity to institute more systematic data collection and conduct a number of major experiments in educational outcomes, and shape policy around the results is one of the more promising ways to increase rich country GDP and virtues, with all the relevant flow-through effects.
The small schools fiasco involved putting too much money into that experiment prematurely, and a real perhaps-statistical blunder, but this was acknowledged and the program dropped in response to poor results.
This is offset by projects like videotaping vast numbers of hours of teacher teaching (correlated with outcomes), the creation of large national databases, causing several jurisdictions to experiment with pay-for-performance (by bankrolling the difference), etc. The chance of a huge win from this extension of science and experimentation is enough to justify things.
The GF also gives a bunch of scholarships to high-ability students (the Gates scholarships at Cambridge modeled after the Rhodes Scholarships, the Gates Millennium Scholars Program for top under-represented minority students in the United States). Again this might not be a utilitarian thing, but such programs provide a way to target talent that might otherwise be lost from key fields, and a huge opportunity for influence by handing out the money to people doing research and work that the GF wants to encourage.
It could be largely symbolic, e.g. the Gates Foundation paying attention to education reaffirms the status of education as an important thing to pay attention to.
I, for one, would love to read your response to Carl’s question about your “world is mad” thesis:
Off the instant top of my head, central line infections and the European Central Bank. I’m busy working on HPMOR and can’t really take the time to consult my Freemind map for the top dozen items. Carl’s list does seem kinda lopsided to me (i.e. not representative), but again, got to make the update deadline on the 29th and all my energy’s going there.
Will you respond when you have more time? :-)
Yes but I might need to be reprodded come July 16th which is when my schedule quiets down again. Having a general debate on whether the world is mad doesn’t seem like a particularly good thing to recurse on deep in a comment thread.
Re-prodding, as suggested.
For future readers: The discussion has continued here.
You seem to have a reminder system that you enter even trivial reminders into, and then get reminded later and act on them. What do you use?
I currently have two ways to do this. One is RTM and the other is Boomerang. I use RTM for reminders to do something in the future where it’s not particularly important exactly when I do them and Boomerang for reminders to do something in the future at a particular time, e.g. if I know I’m going to be in a particular location around 3:00pm next Tuesday and I’d like to be reminded to do something at that location I’ll Boomerang an email to myself for that time. (Or, if someone emailed me about this thing, I can Boomerang that email instead.)
Either Gmail+ActiveInbox or Things. These days, moreso the former.
I think most cellphones sold over the past decade or so (incl. the Nokia 3310, IIRC) have such a functionality.
Can you expand on this? I’m interested in what you use to organize your time and thoughts.