EA Still Needs an Updated and Representative Introductory Guidebook

post by Evan_Gaensbauer · 2019-05-12T07:33:46.183Z · score: 28 (18 votes) · EA · GW · 17 comments

After the publication of the first edition of the Effective Altruism Handbook in April 2015, Will MacAskill's first book Doing Good Better (DGB) was published in July 2015. For a few years, DGB was the go-to introductory book for EA, though the EA Handbook had also originally been intended to fulfill that purpose. DGB was merely a comprehensive and broadly representative introduction to EA that took optimizing the message for a wider audience into account than the EA Handbook. Of course, as EA is a movement predicated on change to become more effective, and also as a relatively young and still growing movement, EA dramatically changed over the course of a few years.

So, in 2018, the Centre for Effective Altruism introduced the EA Handbook 2.0, meant to serve the role that both the first edition of the EA Handbook and DGB, but updated to better represent the EA movement. However, this provoked controversy about the proportionate representation [EA · GW] the EA Handbook gave to different causes; in particular, over how much more space was dedicated to AI alignment, existential risk, and long-termism in EA compared to the community's other priorities. So, many effective altruists since then have still preferred to use DGB as an introductory handbook to EA. However, there is one problem with continually recommending DGB in its largely unedited form from 2015 that the EA Handbook 2.0 remains correct in addressing: it is out of date.

As EA Forum user bdixon recently pointed out in his article assessing if climate change deserves more attention within EA [EA · GW], the prioritization of climate change in DGB appears to underrate the degree of warming climate change may bring about in the next 100 years, and some of the potential negative consequences of climate change. As the EA movement's priorities and methods change over time, DGB remains stuck presenting EA as it was in 2015. So, something like the EA Handbook 2.0 should exist to replace DGB, but last year conversation stultified on how such content should be presented, or how often such introductory modules or handbooks to EA should be updated. Consider this post an attempt to reboot that conversation.

17 comments

Comments sorted by top scores.

comment by michaelchen · 2019-05-13T11:39:23.315Z · score: 20 (8 votes) · EA · GW

I don't have issues with the EA Handbook's emphasis on the far future, but I do think Doing Good Better is much more beautifully written and emotionally compelling, so I'd probably still recommend it over the EA Handbook.

Here are some comments I have on individual articles in the EA Handbook:

  • Introduction to Effective Altruism:
    • Places too much emphasis on "tested solutions," which seems to advocate against high risk high reward interventions.
    • Overall, covers a lot of topics pretty decently.
  • Efficient Charity – Do Unto Others:
    • Written in 2010 (although the EA Handbook says 2013), so it has out-of-date cost-effectiveness figures. There is a note at the top which says that and recommends looking at GiveWell, but I think it'd be better to just directly edit the article, so that people who don't look up GiveWell's figures don't walk away with the impression that $5,000 to save a life is ineffective, or that we can save a life from fatal tuberculosis with $100 if we can't. In addition, while the article claims that it costs $5,000 to save a life from diarrheal disease, I haven't seen any figures from GiveWell which could provided an updated view.
      • SHIC uses the following lines in their excerpt of "Efficient Charity": "According to the World Bank’s analysis, immunising children for dengue fever saves one child’s life for $25,000, but we know that by donating to malaria prevention we could save about five lives for the same cost. If you want to save children, donating bed nets instead of immunising against dengue fever is the objectively right answer, the same way buying a nice car instead of a broken-down one for the same price is the right answer."
    • I really like this article though and I think it does well in terms of emotional impact. It might be good to put this before Introduction to Effective Altruism to get readers hooked.
  • Prospecting for Gold:
    • Feels kind of long-winded, and at some points the gold metaphor is a slog to read through rather than an actually helpful metaphor. I feel like it's a lot better for watching as a presentation than reading. We might be able to rewrite this to make it more succinct.
    • Cites data from DCP2, which has some pretty unreliable figures, and there's a DCP3 now which we can use instead. I don't think this point is too important though.
    • This does cover some important concepts like long-tailed distributions, marginal utility, and comparative advantage.

I'm not going to read/review the rest of the EA Handbook right now, but I think overall, lightly edited transcripts of talks don't make for great reading material, and we'd want to edit them a lot more to be more succinct and easier to read.

comment by Jon_Behar · 2019-05-13T22:14:16.977Z · score: 18 (7 votes) · EA · GW

For global poverty, there’ll be a great new option released later this year: an updated 10th anniversary edition of The Life You Can Save coming out in Q4. There will be updated numbers and examples, two new forewords, and increased emphasis on specific calls to action meant for a broad audience (e.g. initially asking people to make a recurring donation vs. a substantial pledge).

The price is also right, as we’ll be able to distribute free copies of the e-book (which will have links so people can take action more easily) and audiobook. The audiobook will have chapters read by celebrity narrators; this isn’t the time or place to list people involved in the project, but they’ll be a great credibility boost.

A lot of EA origin stories start with the first version of TLYCS. We’re about to have a chance to distribute a new and improved version to a much wider audience, and we hope the EA community will help spread it far and wide.

(I work for TLYCS the nonprofit, which is producing and promoting TLYCS the book.)

comment by Evan_Gaensbauer · 2019-05-14T05:50:28.422Z · score: 4 (2 votes) · EA · GW

Strongly upvoted. It's not clear to me introducing people to EA through books is the best way to go. I think if people first find EA by other means, and it appeals to them, giving books to people who initially find EA appealing could be a better way to get them to stick around. I think having free ebooks like this is something great to have to introduce to people who first become cognizant of EA. That it's a free ebook that doesn't have to try to strike some perfect balance between different causes makes it easier for people who just want to learn about one thing at a time. I wish more dedicated EA organizations did something like this.

The only other EA-aligned organization I'm aware who does something similar is MIRI. A lot of the books they release are way too long for most people, or aren't as directly related to AI alignment. Smarter Than Us by Stuart Armstrong is freely available online as an ebook through MIRI, and fits the bill though.

comment by aarongertler · 2019-05-23T12:38:01.395Z · score: 6 (3 votes) · EA · GW

I'm the current Content Specialist for CEA (the same position held by Max Dalton when he put together the second version of the EA Handbook).

We're aware that the Handbook isn't an ideal resource and have been thinking for a while about how we might want to update it. In the process, we've consulted a few stakeholders with expertise in different cause areas. So far, we haven't officially released any edits, but we might do so in the future; we're still uncertain about timelines and how we want to prioritize this project.

comment by Evan_Gaensbauer · 2019-05-24T06:47:42.653Z · score: 2 (1 votes) · EA · GW

Thanks for the feedback. That sound reasonable. I wrote the OP because this was a resolvable issue that lots of people disputed over EA that appeared to be left unfinished. There are other introductory guidebooks for effective altruism for different causes, etc. So that there isn't a general guidebook right now that satisfies different relevant parties in EA doesn't seem like a huge problem. Michael Chen pointed out multiple major problems [EA · GW] with one article in the current EA Handbook 2.0. They're significant mistakes that the article needs changed for it to hold up. I figure for the EA Handbook to deliver its message with integrity, it has to do that for all the articles in the EA Handbook. Since most of the articles were initially written as blog posts, I expect there are other holes in each which with hindsight we could point out. It's just that the articles in the EA Handbook 2.0 may not have been as professionally written as published books or scholarly articles by effective altruists, which is a quality we should ostensibly aspire to if an introductory book to EA is about EA putting its best foot forward to people new to EA.

Siebe Rozendal suggested [EA · GW] an updated version of Doing Good Better. I thought this would be too much work, but it seems like it might be less work to update DGB than it would be to update the EA Handbook, since that poses multiple difficulties. I thought that would require Will doing most to all of the work to update DGB himself, but The Life You Can Save (the organization) has worked with Singer to update the book of the same name. Jon Behar, who works for The Life You Can Save, explains it here [EA · GW]. It's a new edition 8 years later, so there must have been a lot to change. So, the CEA could do something similar where Will works with them to update DGB. CEA could consult with TLYCS (the organization) or work with them in some capacity to replicate the process they've used with Singer to update TLYCS (the book).

I honestly think it might be more tractable and more effective to update DGB than the EA Handbook 2.0. If that's the case, given that DGB is written more as an intro to EA as well, and it's more popular, I imagine some EAs would be willing to donate time and/or money to see an updated version of DGB happen.

Is that something you think Will and/or the CEA wold consider?

comment by SiebeRozendal · 2019-05-14T20:16:10.110Z · score: 5 (4 votes) · EA · GW

In the meantime, it would be great to have an article to send along with Doing Good Better - or to send afterwards - with a number of corrections, updates, caveats, and pointers to further reading. Ideally, this would be authored by Will himself, but I can see someone else doing a good enough job at it, or someone ghost writing for Will.

comment by cafelow · 2019-05-12T19:47:42.486Z · score: 4 (3 votes) · EA · GW

I agree Evan. I still give out many Doing Good Betters (hoping people read past the poverty sections) and only send people the Handbook if they've engaged with EA ideas for a while. The intro book would ideally be written with someone with a high status in the EA community and the world in general, as even an excellent intro book may not supplant MacAskill and Singer's intro books. I'm wondering if the best intro book would be a new edition of Doing Good Better with less on global poverty to allow for more on other cause areas. That would give the new book status and mean the outdated edition will automatically stop being the default intro book (Of course that would require a lot of Will's time, even if a coauthor was brought in).

comment by Evan_Gaensbauer · 2019-05-14T06:09:06.726Z · score: 4 (2 votes) · EA · GW

At first I thought just updating Doing Good Better would be too hard. After reading Michael Chen's comment [EA · GW] on what needs to be corrected in the current articles in the EA Handbook, on top of everything else that would need to be updated about it, it seems like it would be more work than updating DGB. It seems like DGB has probably been subjected to more stringent and professional editing since it had to go through external, commercial publication, and so updating an already solid work that covers the core facets of EA is probably easier than creating a new handbook from scratch. It also sounds like based on Jon Behar's comment [EA · GW] that The Life You Can Save organization helped quite a lot in updating Singer's original TLYCS book. I hadn't thought of an idea a whole organization could give some time to updating an EA book, as opposed to a single assistant or editor. So the CEA or another organization could help will with that, which would make the job of updating DGB a lot easier.

comment by Jon_Behar · 2019-05-14T17:45:33.381Z · score: 9 (4 votes) · EA · GW

To confirm: TLYCS the organization is playing a critical role in the book project; without the organization there absolutely wouldn’t be an updated version. The org has been essential every step of the way (including working to purchase all the necessary rights since ~2014). There’s a ton of work involved and Peter is doing a lot of it, but we’re trying to take as much off his plate as possible including pretty much everything on the promotion and distribution side. There are a lot of skills needed to pull this off, and this model plays to everyone’s comparative advantage: Peter is great at thinking and writing, while the org is better suited to set the distribution strategy (Charlie Bresler, TLYCS’s executive director, ran the marketing department for a large company with an iconic ad campaign).

Our hope is that a lot of people and organizations throughout the EA community will be able to use the book as a way to have more impact, such as GiveWell distributing the book to their donor base, groups/individuals sharing the book with people first learning about EA, making the book available for download at effectivealtruism.org, EA Global, etc. And of course our recommended charities and other effective nonprofits are mentioned throughout the book, with links embedded in the ebook version to make it easier to convert.

This seems like a good time to mention: TLYCS is fundraising for this project, and you can make an earmarked donation here. There’s more background on the book project in TLYCS’s recently released Annual Report.

The EA Meta Fund has made a $10k grant for this project which we’re extremely grateful for, but this barely makes a dent in the barebones budget, let alone what we think we ought to invest in this project. We’d love to see other donors from the EA community get involved as well.

comment by michaelchen · 2019-05-15T00:35:15.953Z · score: 7 (2 votes) · EA · GW

"it seems like [updating the EA Handbook] would be less work than updating DGB" → do you mean that updating the EA handbook would be more work? That would make more sense given the rest of your comment.

comment by Evan_Gaensbauer · 2019-05-15T05:01:34.467Z · score: 2 (1 votes) · EA · GW

Oh, yeah. Thanks. I fixed it.

comment by Evan_Gaensbauer · 2019-05-12T22:57:24.311Z · score: 4 (2 votes) · EA · GW

One question to ask is what the goal of such a book is. If it's to spread the basic ideas as far and wide as possible, a popular book like Will's or Singer's books seems like a good start. I'm not sure how those books have panned out for all the goals they were intended to serve, since I don't know who has measured them. I think there have been some attempts to measure the impact of DGB in various ways, but I haven't kept track of them.

I was thinking an updated edition of Doing Good Better might work. Unfortunately, with how busy Will is, it seems unrealistic to expect he'd be able to update date it soon. I was hoping someone else would come up with a suggestion with how a better EA Handbook could be created.

comment by Risto_Uuk · 2019-05-12T14:32:03.579Z · score: 3 (2 votes) · EA · GW

In addition, what do you think should be updated in Doing Good Better?

comment by Evan_Gaensbauer · 2019-05-14T06:10:09.636Z · score: 2 (1 votes) · EA · GW

I was more thinking a new guidebook to EA that is more satisfying to more people in the EA community would be better, since it seems like a lot of people were dissatisfied with the EA Handbook 2.0 release from last year.

comment by Evan_Gaensbauer · 2019-05-12T23:00:48.913Z · score: 1 (3 votes) · EA · GW

Well, after reading the article on climate change, I think that's something that should be updated in the book. It also didn't mention AI alignment much, so it could be more, but it doesn't have to be a lot more.

comment by Risto_Uuk · 2019-05-12T14:30:57.969Z · score: 3 (2 votes) · EA · GW

Your link referring to bdixon and climate change leads to Joey's post "Problems with EA representativeness and how to solve it". Can you share the post that discusses how Doing Good Better appears to underrate the degree of warming of climate change?