Olá Duarte! Desculpa a resposta tardia. Sim, todo o trabalho realizado ao longo dos 7 anos foi fruto do trabalho voluntário de todos os envolvidos. A determinada altura tivemos conversações com a Open Philanthropy (por iniciativa da própria), mas como havia a intensão de todos os envolvidos continuarem a desenvolver o seu trabalho de forma voluntária, consideramos unanimemente que não haveria necessidade de uma dotação de fundos.
The main objective of this post is to provide information about the EA translation work we’ve been doing for the past 7 years for those we might ask permission from, to translate their work and then publish it on our Brazilian and Portuguese EA Website. We close the post with some suggestions that might help the translation work all around the EA community, and hopefully extend its reach.
Brief history of the Brazilian and Portuguese EA Website
Since 2015, just with volunteer work, we have translated, revised and published a selection of some of the best EA content we could find, amounting to almost 400 posts (roughly the equivalent to six 200 page books, or ~$64,000 USD in translations [1]), and we've also translated the EA Handbook (2015, Ryan Carey Org.) — with more than 340 downloads.[2] We post regularly, on a weekly basis, and are currently translating the EA Introductory Program.
We provide a large variety of (mostly introductory) content about EA cause areas (rotating publications on cause areas such as extreme poverty, animal cause, global catastrophic risks, generic EA content), and, while keeping up with the evolution of the EA movement, we also provide news about special moments (such as the EA Globals, pledge drives, special millstones, new developments of EA orgs, new cause areas, etc.).
At this point, we are the most viewed EA online resource for the potential ~274 million Portuguese readers around the world, with ~25,000 views per year.[3]
Why is this work important?
Assuring a good first impression: Imagine the first thing you saw in your internet search about EA was the Charity Navigator CEOs's rant about what they called “defective altruism”, or a video where it's literally stated that "Effective Altruism is an NGO that, following Peter Singer’s ideas, uses artificial intelligence to establish which is the best donation"(!). Well, although we are long gone from the first situation (that showed up in the top Google searches in Portuguese for some time), the second one is still a reality.[4] So it seems there’s still a gap to be filled (and to maintain it filled).
Trying to create valuable EA content in Portuguese: At this point, all the EA classics we've translated throughout time seem to have solved some of the problems mentioned above, and some of this content shows up regularly in our website statistics as the most viewed. But that's also the case for some outdated content and more amateur translations, so we should make sure we are not contributing to the problem.
Trying to maximize the reach of that valuable EA content: If the more outdated and amateurish content is a quick way into the website, maybe it serves a purpose: being the door that opens to what otherwise might be unseen good EA content. On the other hand, the weak reach of newly published content can be off putting (by comparison with earlier content, that frequently shows a statistical difference of over 10x). But we can also think about this mainly as a way to keep the website alive (as the EA movement evolves, if we stopped posting, it would soon be outdated), and thriving (as new content seems to generate more interaction[5]).
New gaps to fill: as we said, maybe we’re long past the time where the lack of good EA content in Portuguese might have led to a non-negligible risk of someone having his first contact with EA through an unreliable source and moving away without ever really engaging with the EA movement. Even if that's so, in such a dynamic movement, we will always have the same gap regarding new kinds of content (eg. wild animal welfare, AI, longtermism, etc).
Individual and community impact: If at an individual level we might see this volunteer work as an opportunity to enhance our writing and interpreting abilities in two languages (literally learning from the best EA writers), the updating with new knowledge, memory exercising, and critical thinking stimulation (with multiple revisions of different approaches to basic EA concepts) might be a great way to be knowledgeable about the EA movement and to be prepared to promote it.
Why do we mainly focus on introductory content?
Simple, diverse and compelling: By choosing a time reading scale of 6-12 minutes, a clear approach to different points of view (even within the same subject or cause area), and also compelling arguments to express those views, we’re trying to reach the greatest number of people, with different academic backgrounds, also having in mind we live in countries with a very different social, economic and cultural context from those where EA originated from (and Portuguese-speaking countries include not only Portugal and Brazil, but also African and Asian countries). So, although we post about all sorts of EA fringe subjects, we also post regularly about what a collaborator of ours called “uncontroversial altruism”, that is, we always return to the foundations of “the effective altruism tower”.
Can I be an AI expert if I only speak Portuguese?: This is just a short question as an example for a longer one: How much content can a small team translate to inform someone so they can be sufficiently knowledgeable about a scientifically specific matter? (And we might add: …without such a small team neglecting translations in other matters?) We are not trying to raise a false dilemma here, minding there's plenty in between, but we think that presenting a (simple, diverse and compelling) introduction to EA is a hard enough task, and some of those core arguments and values are also what might motivate those who can, on their own, go further away from our website's content to other EA more comprehensive sources. If your target audience is someone like the typical EA, you'll probably realize you'd be translating for someone who already reads fluently in English.[6]
How do we do our work?
Although this has all been the work of volunteers, we have always aimed at a professional quality, so here are some rules we established early on, and a brief account of how we do our work:
Publication frequency: By establishing a weekly rate of publication we’re not only taking into account the team's constraints, but also questioning if we could reach enough audience to justify a higher rate (that could compromise the content’s quality);
Personal fit: We try, as much as possible, to assign translations according to the volunteers’ areas of preference and expertise. So we think that having the help from a variety of cultural and academic backgrounds and different personalities might benefit this approach. For instance, someone personally connected to the animal cause, might have a more dedicated and enjoyable work translating EA content related to that specific cause, and that, in turn, might produce better quality work.
True to the original: Maybe a simple rule of thumb is, if we feel we have to change something the author said, and we're wrong, then that's a really big mistake, so we shouldn't do it. On the other hand, if the author is wrong, then maybe it’s a mistake to translate them in the first place (and that applies to style, as much as to content). So, as the saying goes, “traduttore, traditore” sounds pretty much the same, but one means translator and the other means traitor.
Uniform translations: To guarantee cohesion and consistency between translations (while preserving the differences between Portuguese from Brazil or Portugal[7]), and to ensure that everyone translates (especially new terms and even neologisms) the same way, we have been building aglossary. Also, one of the editors revises every translation.
Revision, revision, revision…: Ideally translators should revise their work as much as possible before passing it on as finished; then their work should get revised by someone more experienced (who does it more than once) before it’s published. And that means not letting the need to accomplish a publishing goal undermine the accuracy of the work, trying to assure that there's always time for that extra revision that catches that final typo.
Form and function: Each post is carefully formated (with the original links, titles, quotations, notes, translated graphics, etc.) and accompanied by appropriate and suggestive images, which are selected from free sources (see thesethreeexamples).
Then spread it: Every new post is shared in social media (especially Facebook groups and pages), and via our trimestral newsletter.
What problems are we facing?
Reduced team: throughout the last few years we went from 7 translators (and one person producing original content) to only 2 translators. Until now this has been enough to maintain the weekly posts, but a more diverse group would naturally bring forward the quality we aim for (see above: Personal fit). That being said, we hope those who moved on to other EA projects have found this experience enriching, both professionally and personally (at least as much as we have by working with them), and we also hope future volunteers may benefit from it too.
Copyright issues: Although we’ve always abided by the essential citation norms (by mentioning the author’s name and by linking to the source of the original content), it has not been our practice to ask permission to translate.[8] Be it for the sense of belonging to the same big international team promoting EA content, be it by the time constraints of doing this volunteer work in our free time (using a nonprofit platform), or be it, in some cases, for the sense of urgency (eg. the pandemic and the war in Ukraine) where, at some point, information in Portuguese was scarce or unreliable. Even so, we’ve had some good experiences of appreciation in the past, when we’ve informed authors or organizations about our translations. And sometimes they have even posted a link to our translation (which in turn gave our work more visibility). [9]
Are there other accomplishments?
This is some of our work outside the EA Website that might reach even more people than posting in the (BR/PT) EA Website, although we don't have the specific numbers to back this up.
Trying to learn from our mistakes: From the point of view of the Portuguese experience we must underline our updating about differences between European and Brazilian Portuguese, allowing for the preservation of those differences when revising translations and thus promoting a wider accessibility of the website's content.
Can permissions be easier?: Throughout the time it took us to write this post we’ve realized how frustrating it can be weighing the ethical consequences of the illegal behavior of translating without permission, or not translating at all (because much too often people or organizations either don't give permission, or don't answer requests for permission — at least in a useful time frame). And this doesn't even take into consideration the nuisance that an author might go through, for instance, after willingly giving permission to have an article published in the EA Handbook, being contacted by a french EA translator, then a Spanish, then a German, then a Chinese, then a Portuguese (from Brazil and then from Portugal)… Well, you get the picture! So, here’s a few ideas:
A Translations Coordinator: As Open Philanthropy has expressedinterest in investing in translations, would it be useful to have someone working full time on this? For instance, starting with the content thatOpen Phil is interested in translating, and posting an online spreadsheet updated with the status of the content according to the authors/organizations authorization:
Permission for Any Translation (PAT): like a Creative Commons Attribution license, anyone can translate;
Individual Translation Permission (ITP): permission granted on a case-by-case basis,
In this case (ITP) would it be useful to send the permission request via the Translations Coordinator? (Eg. for monitoring, accountability, and expediency, etc.)
Translation Permission Unknown (TPU): in the case the author/organization has not been asked or has not answered yet,
Translation Unauthorized (TU): the author/organization forbids any kind of translation.
The website states its translation policy: so every participant or collaborator would agree and follow the same policy. Here's three examples:
(ITP) Giving What We Can states: “Pleasecontact us if you wish to redistribute, translate, or adapt this work.” (And, from our experience, GWWC answers quite promptly to a translation request).
(TU) Vox (including Future Perfect) states: “...we do not grant individuals permission to translate and reprint, repost, or upload our content on personal YouTube channels, social media, blogs, websites, etc.”
Identify the post’s translation status: regarding older content, and if it’s hard to reach a consensus, for instance in websites with many collaborators, like the EA Forum (and other EA Blogs), they could categorize (at least) themost viewed posts with the same translation status (PAT, ITP or TU). And translators could move on with their work without bothering authors/organizations time and time again unnecessarily.
Submission of translation’s permission: regarding future content, EA Forum (and other EA Blogs) could require authors to submit their translation permission (PAT, ITP or TU), so that new posts would go live with that categorization.
Link to the translation: Every EA Forum post that had been translated (or on other EA Blogs) could show the language with the link to the translation (Eg. DE – German, ES – Spanish, FR – French, PT – Portuguese, RU – Russian, ZH – Chinese, etc.). This would allow people to find local EA communities/projects, and also avoid redundant work.
What other ideas would be useful to promote translations?
It's easier now!: Although we’ve mentioned the reduced team constrainments, if the permission issue is solved, and we don't fall into the paradox of choice, we can now benefit from plenty to choose from (by comparison with as early as 2011, when some of us started promoting EA). We can also benefit from the gathered experience, and the new translation technological tools, and that makes us feel motivated to keep on doing it. Always keeping in mind that this work can amplify our positive impact in a way that probably our other personal resources couldn't.
Acknowledgements: First and foremost I would like to thank all of those who have contributed to the Br and Pt EA Website, specially Rosa Costa, who is currently working with me, including in this post. I also would like to thank Laura González for encouraging us to write about our work, and for the discussions that led to some of the ideas presented here. Finally I would like to thank Justis Mills for his writing suggestions – any mistakes remaining are my own.
Although not all posts in the website are translations, this amount only takes into account the number of words in the website (with 0,13€ per word) and not the EA handbook, nor the revisions of almost all posts.
We are not sharing the link to this video so as not to contribute to the problem, but just for context we will add that it’s the second video on Google search about “Effective Altruism” (it follows Peter Singer's TED talk) and, oddly enough, it’s a didactic philosophy video comparing the moral theories of Peter Singer vs. Hannah Arendt.
We use WordPress and it states that blogging regularly is one of the ways to get more views and increase website traffic. This also seems to be a general understanding.
We follow some Wikipedia rules like not changing form the original translation in pt-BR to pt-PT (and vice versa) even if the last review is made by a pt-PT speaker (and vice versa), and try to find a grammatical formulation that doesn't seem strange to one part, and, for instance, whenever we can’t find a common word, we add (in brackets) the corresponding one (eg. 7 billion = pt-BR: 7 bilhões [pt-PT: 7 mil milhões]).
We’d like to thank Laura González for bringing these copyright issues to our attention, which has also prompted us to write up this post as a way of introducing our work and maybe have a positive impact on future translation requests.
This post is co-authored with Ben Garfinkel. It is cross-posted from the CEA blog. A PDF version can be found here.
Summary: Some strategic decisions available to the effective altruism m...
Disclaimer: Although I work on the Groups Team at CEA, I’m writing this in a personal capacity, and this post does not constitute an endorsement by CEA.
Agency - the realisation that you really can just do things.
TL;DR
Biosecurity needs people (of any background) who are agentic and have a high execution velocity and track record....
Summary
Making AI safe and good requires many people to be ready to address crucial problems. However, our current “pipelines” for finding, training, and accelerating talented individuals who could work on these problems are “leaky”, meaning there are many ways they can be improved. This post shows my full overview of the AI safety talent pipeline: where people come from, where they go to learn more, and why this is impactful. I also describe some “leaks” in the pipeline: stages that...
Muito obrigado pelo trabalho que têm feito :)