Thanks for being so kind Sean. I think you work harder than just about anybody I know: no excuses needed for you!
Bernadette_Young
Karma: 488
My own little victory dance:
When I set my threshold for my pledge in 2009, I pledge both an absolute percentage and a threshold of £25000 above which I would donate all money. I had done my research and tried to set a level that would cover both my own needs and that allow for having a family.
This is the first year we have had to pay any real costs for our daughter. For the first year of her life I was on maternity leave, so the cost was an opportunity cost of reduced salary. When I returned to work last May I had two new pretty huge expenses: full time child care (£10k a year, shared with my partner), and—because I was returning to do a PhD—full time university fees which cost £7000 a year.
I’m really excited to have been able to meet both my pledges this year, and within the threshold I’ve set, to have been able to pay for my study, childcare, as well as my own every day expenses. I’m feeling much more confident about being able to meet my pledge in the future while still providing for my daughter.
End of year celebration thread!
Your comment above indicated you had measured it at one time but did not plan to do so on an ongoing basis: “However, we can’t control that, and it would not be helpful to assess that on a systematic basis, beyond that base rate” That approach would not be sensitive to the changing effect size of different methods.
Not really I’m afraid. That reasoning seems analogous to the makers of glipizide saying: we know lowering blood sugar in diabetics decreases deaths (we do indeed have data showing that) and their drug lowers blood sugar, so they don’t need to monitor the effect of their drug on deaths. Your model can be faulty, your base statistics can be wrong, you can have unintended consequences. Glipizide does lower blood sugar, but if you take it as a diabetic, you are more likely to die than if you don’t.
It would also be like the Against Malaria Foundation neglecting to measure malaria rates in the areas they work. AMF only distribute nets, but they don’t actually care about (or restrict themselves to monitoring) how many people sleep under bed nets. The bed net distribution and use only matters if it translates to decreased morbidity and mortality from malaria.
If you are sharing information because you want to increase the flow of money to effective charities, and you don’t measure that, then I think you are hobbling yourself from ever demonstrating an impact.
We may have different perspectives on academic readers: I’m a relatively junior medical researcher. Three of my papers have over 100 citations. The view I expressed here is the one shared by my Principal Investigator (a professor at Oxford University who leads a multi-million pound international research consortium, and has an extensive history of publishing in Nature and Science). Humanities and medical research are likely to have some differences, but when fewer than 20% of humanities papers are thought to be cited at all, I’m not sure that supports humanities papers being read more extensively.
I don’t see any contradiction between saying:
I believe that, at the level a general reader will engage with it, this piece distorts the ideas of effective giving towards the damaging ‘good charities have low overhead’ meme, and will not in expectation increase donations to EA charities
In order to show the contrary, you need a more concrete endpoint that website clicks.
No matter how many steps there are between an action an an endpoint, the only robust way to show an association between them is to include measurements of the end point you care about: surrogate markers are likely to lead your astray. For instance, I don’t give much weight to a study showing drug Y lowers serum protein X, even though high levels of serum protein X are associated with disease Z. To prove itself worthwhile, the drug companies need to actually show that people on drug Y have lower rates of disease Z, or better yet, deaths from disease Z. Drug companies complain about and manipulate these principles all the time, because solid endpoints are take more time, effort and money to measure, and their manipulation around them has cost lives. (See the diabetic medication glipizide: short terms studies showed it decreasee blood sugar in diabetics—an outcome thought to improve their mortality—but longer term data showed that it makes people taking it more likely to die.)
Of course you’re free to measure your work however you choose: I would personally be unconvinced by website traffic, and if you are aiming to convince evidence minded people of your success I think you’d do well to consider firm endpoints or at least a methodology that can deal with confounding (though that is definitely inferior to not being confounded in the first place).
At any rate, that’s enough on this from me.
I agree that maximising the good done with every effort is the essence of EA; I disagree that the wording and structure of your piece communicated that, even with those words included.
There’s a tendency for people who do a lot of academic writing to assume that every sub-clause and every word will be carefully read and weighed by their readers. We agonise for months over a manuscript, carefully selecting modifiers to convey the correct levels of certainty in our conclusions or the strength of a hypothesis. In reality even the average academic reader will look at the title, scan the abstract and possibly look a figure and the concluding sentences.
Communicating complex ideas in a short piece is really hard to do, and if the less concrete the link between the message you want to convey and the topic you are trying to shoehorn that message into, the harder it is to avoid distorting your message. You could seek feedback from people who aren’t already aware of what you’re trying to communicate, but that’s likely to be very hard to do in the time frame needed for a current news story.
If you want a measure of success, I think you need a much better end point than website views, which is a) subject to a wide range of confounders and b) only a proxy for the thing you are trying to achieve.
I don’t think it’s about mismatched expectations so much as I have a different assessment than you do of how much this piece is likely to promote effective giving.
If your intention was to promote consideration of impact, or recipient focussed donation behaviour, then I think this article misses that mark. Sure, the information might be there 15 paragraphs deep in one of a dozen links, but it’s not conveyed to me—even as an interested reader versed in effective altruism ideas.
If indeed your article was intended by you to promote Charity Navigator style research with the hope it will nudge people towards the idea of impactful giving (which is what I take you to mean by saying that flattening out of the message is “a bug not a feature”), then I respectfully disagree that such an approach will in expectation increase effective giving.
Owen I think these are important caveats.
One further risk is that message you are trying to convey has to be stretched or even distorted to be made relevant to the original story. This is a result of the “hijacking” approach, and unfortunately I think it’s evident in this piece.
The problem with Wounded Warriors as I understand it, is not that their proposed projects were not likely to be helpful (I haven’t seen evidence that would help me answer that), but that people in the organisation mis-spent funds, and did not use them according to the charities own stated aims. So the problem here is not whether Wounded Warriors are engaged in effective interventions, but that people within the organisation diverted money from interventions and spend it on luxury flights and accommodation for its staff.
It seemed to me that the characterisation of effective altruism groups in the Time piece as organisations “pushing the nonprofit sector to become more transparent and accountable” is indistinguishable from Charity Navigator and others who are concerned with overhead as a metric of effectiveness. If we dilute the notion of an effective charity to one that has been vetted for financial transparency and accountability, we really lose the key message of how much different interventions vary in their impact.
For an example of how this can lead to opposite conclusions than EA reasoning: most EAs would agree it would be better for the world if programs like Scared Straight or Playpumps were bad at delivering their programs, since their programs have a negative impact. I expect it would be overall negative to deliver the message that finding an organisation with low overheads is both necessary and sufficient to ensuring your donation has a positive impact. I imagine that wasn’t your aim here Gleb, but it’s very much how it reads to me, probably as a result of the need to stay relevant to the news story you were tailing.
AGB reiterates a good suggestion he’d previously made on the facebook group: that no modification is needed for people who participate in EA without being a maximal sacrificer, and it’s entirely appropriate to call those people effective altruists. If we want a term for people who are hugely involved, or sacrifice a great deal of their own well-being (I’m not convinced self-sacrifice is a good metric here, but that’s another conversation), can’t we just find a modifier for those people? Dedicated/devoted may be less problematic when you don’t have to search for a counterpart that isn’t dismissive.
Thanks for pasting that comment here—I was sure there had been a really good discussion on this, with a general consensus that “softcore” needed to disappear. Perhaps I was just really persuaded by your comment and assumed others were likewise.
I agree there’s less an issue a designation for “very involved” being a bit negative, but I’m moderately opposed to “hardcore” because although it is used as you describe, I think its strongest association is with porn.
Thanks for this post Julia.
I know lots of people have been seeking alternatives to the ‘hardcore vs softcore’ terms that seem to have sprung up, and I agree that alternative terms are preferable to those two for many reasons. However I think you’ve addressed a much more important issue, that any binary categorisation is artificial and likely to be counterproductive.
Seeking donations from high net worth individuals/ financial ‘elites’ is a crowded market. The giving pledge is just one campaign targeting these people, which is already connected to networks of very wealthy person. Do we have good reasons to think that EA would have a comparative advantage in such a crowded market?
Another significant disadvantage I see to becoming another group that concentrates targeting high net worth individuals is that we would be perpetuating the myth that only very wealthy people can make a difference, which more moderately wealthy people often cite as their reason for not taking charity more seriously.
Also, any new pledger has some non-zero chance of breaking the pledge (see the GWWC fundraising prospectus for their current estimates, though some people have argued these are under-estimates). The chance of different people is probably largely independent. If this is true, then at the margin, two 10% pledgers have a lower chance of both defaulting and would probably lead to more money being moved (ie narrower 95% confidence interval on the amount moved).
If you use the search function in the main facebook group it’s quite straightforward to find plenty of discussion.
“If people make bad decisions then that’s unfortunate, but all other things being equal more information leads to better decisions and EA is the last movement which needs to have its strings pulled. ”
To be clear: I am not advocating censorship. I’m advocating putting information in a context that makes its scope and importance apparent. It would be naive to ignore that some ideas have mimetic pull, particularly if you’re being counter-intuitive by advancing an argument that aid is bad.
“I don’t have the time to write about everything.”
No of course not, but of all the problems in all the gin joints in all the world, you picked this one. That is a form of cause prioritisation, and I think it’s reasonable to draw some inference from that action.
I think the phrasing of the ‘problem’ is bad, but the title really isn’t the only issue.
“Considerations entirely outside the model: impact of development on wild animal suffering, climate change, technological progress, global economic development, etc.”
I’m afraid this really doesn’t read to me as being clear about how narrowly a focus this argument takes. I have literally seen people say “Now I’ve heard about the poor meat eater problem I’ve stopped donating to SCI”, so simply saying you don’t draw any conclusion is not, I think, sufficient justification for advancing such a one sided argument. Those things you wave away in the sentence above will in all likelihood completely dwarf the numbers below.
I do think that considering meat eating in the developed world to be “a separate problem” that is not a “relevant topic” is discriminatory. It’s an ‘us and them’ divide, which is purely conceptual.
I don’t think my objections are a reductio ad absurdum, I just think they are harmful actions that are not justified by the reduction in animal suffering they might indirectly lead to. I do find it odd that you see social stability as having positive x-risk only in developed countries though.
Hi kbog, I appreciate you’ve done a lot of work here, but I’ve downvoted because I have a strong ethical and practical objection to this issue being discussed as ‘the poor meat eater problem’. These objections have been hashed out every time this topic comes up. It makes me very sad that the meme persists, and I think it’s terrible for it to be associated with EA discussion.
I think the so-called ‘poor meat eater problem’ is based on 2 fallacies, at least one of which appears somewhat prejudiced: 1.: the decision to focus on only one long term consequence of advancing development 2: the failure to apply the same reasoning to developed countries
Clearly the vast bulk of meat is eaten by rich people. If you want to discuss the implications for animal suffering of economic development, then why limit the discussion to poor countries? Why limit the consideration to ‘what might be the effects of increasing aid to developing countries’. If one was to take this line of reasoning as am important one (ie we should limiting animal suffering by limiting human economic development, since the latter is associated with more animal product consumption), then I would question why you don’t also recommend the following:
Support political candidates most likely to trash the local economy (reducing local economic development)
Support political parties that oppose universal health care (thus keeping the local poor too poor for meat)
Support for parties that oppose gun control (murdered people don’t eat meat!)
Support for anti-vaccination policies (nor do children who die of whooping cough)
Support wasteful or ineffective use of research money (to make sure we don’t make any discoveries that would advance human development)
I hope you think those would all be terrible things to do, and I think the suggestion that we should limit our help of the global poor because they may as a result consume more animal products is likewise awful.
(edited for clarity, typos and niceness)
I don’t think ‘we can’t know it’s not a problem’ is a helpful guide to deciding if something needs action. Have you seen any evidence of voting being used by cartels or sock puppets? As you say, it’s just as possible for up-votes to be done for nefarious reasons (though I have serious doubts as to whether that’s the case) - but requiring comments for up-voting would also be onerous and reduce people’s interactions on the forum.
I think the suggested policy would make the forum worse by raising the bar to participation. Greg has explained the problems with it quite articulately above, so I won’t recapitulate his comment.
Thanks!