Translations in Brazil: much more than you needed to know

For many things, thanks to Fin Moorhouse, Konstantin Pilz, Laura Gonzalez and José Oliveira.

For the support, thanks to Juana, Leo, Fernando, Adriana, Luan, Luciano and Dani from Brazil – and to all the translators who worked with me.

And thanks to the authors who kindly allowed us to translate their texts – particularly to Tobias Baumann , Lars Doucet and Michel Justen.

Origin story

I started thinking about translating EA content in April, when I met Fin and Konstantin Pilz. I was then dreaming we could someday perhaps help translate The Precipice – just like my hero Anna did. But it wasn’t that simple; instead, I got in touch with Eli Rose (who was compiling info on translated EA content online) in May. In June, I applied for Open Philanthropy’s Proposals for growing the community of people motivated to improve the long-term future, with a project aiming at widespread dissemination of high-quality content. Here’s my proposal.

Having some previous experience with translations in Philosophy – e.g., I organized a volume of translated SEP entries on Philosophy of Economics – I thought that this project would be easy and perhaps synergetic with other plans I had for the rest of the year, such as my postdoc on Intergenerational Justice / Time Preference / Carbon Prices, and the creation of a research group on GCR in Brazil.

Deciding what to translate

In August 2022, I received the grant from Open Phil. One of my goals was to provide material for participants of a MOOC “Introduction to Effective Altruism” organized by a Research Group from UFU – Universidade Federal de Uberlândia; I am one of the facilitators in this course (there will be some live sessions), and, without this grant, it would have been hard to present content in Portuguese on EA organizations. Precisely because of a lack of translations, the course program was initially based on the first edition of the EA Handbook (2015), which is a bit outdated—as was our page on problems profiles on https://80000horas.com.br/perfis-de-problemas/ (finished just before 80000 hours website went through some major changes). Thus, because the students were not required to able to read in English, we started thinking about how to fill this gap, and I asked other facilitators if they wanted me to translate additional specific texts[1].

At first, I planned translating the 2nd Edition of the EA Handbook (2019), the Syllabus of CEA’s current introductory course on EA, and CEA’s resources page. But I ended up changing my mind; first, because some of that content had already been translated (and it was laborious to identify what had already been published online), or was not as good as more up-to-date texts, or depended on obtaining permission, etc. Second, I ended up consulting a SEO expert to increase the reach of our translations, posted on the <80000horas.com.br> webpage[2]; she rebuilt part of this website, and advised us on what sort of content would be more likely to optimize google search results and attract relevant traffic – i.e., people who were either interested in EA, or in subjects that are adjacent to EA[3]. This was one of the reasons we ended up tranlating the glossary and the table with EA-orgs (so that we would increase the chances of showing up in searches for the corresponding subjects). The relevance of this investment will persist, at least for one year – after which we will try to better assess its impact.

Minor obstacles

I thought I would have finished this by October – so as to support UFU’s course. I still fall for the planning fallacy… But in my defense, the MOOC was delayed (so decreasing my urgency), too.

Until December, we were very uncertain regarding copyright policy in EA – an issue we voiced in a meeting with other translators mediated by Laura Gonzalez (then working in EA Comms). Because of it, we had to pick the content we aimed to translate carefully, and ask for previous permission[4] – which consumed more time than I’d predicted[5] (as Jose had warned me). This was only solved when new EA Forum posts, posts made by CEA staff and by the “EA Handbook” user started being regarded as under a CC-BY license by default.

Also, spending money to translate useful content became harder and harder, because of a drop in the average price of translation. At first, I worked with many different translators (including two companies, Alphatrad and GTS), who charged distinct prices and delivered results with varied quality; but after bargaining, building more stable relationships, I ended up working only with a small sample of them[6]. However, I’m not sure how this market will be in the future, as NLP-based translations are getting better each day.

In hindsight, I don’t regret consulting an SEO expert, but some of the things she did or recommended were laborious and expensive (e.g., paying for a Semrush subscription, or editing links, etc.). It’d have been easier to invest more resources on Google ads (as they are quite cheap in developing countries). Some people had alerted me that they thought such a consultancy could be a waste, because, roughly, this type of expert sees webpages through “marketing lenses”—so she doesn’t understand our “product” and our “potential clients” very well (though, in this precise case, I can say she made an effort and was aligned); but these same people also predicted that buying ads would be useless, and now I think this not true (at least not if you time your campaigns well). However, hindsight is ²⁰⁄₂₀, and I won’t even be sure about this until October.

This delay has resulted in more stress than I had predicted: in October and November, many projects of mine had superposed the translation project—moving to Portugal to start my postdoc, working on the proposal of the research group (with the goal of organizing a conference), getting married, etc.

Outcomes and lessons

I list the content published on <80000horas.com.br> here; it sums up to aprox. 56,100 words—and about 20k only for the Glossary (and afterwards, Fernando got excited and included links to many subtopics, so adding another 16,235 words—but props, that’s on him). I think our major achievements were the glossary (based on a compilation of the EA topics – thanks Laura for sharing this), the spreadsheet with EA-aligned orgs (thanks Michel Justen) and the Longtermism FAQ (thanks Fin), as they compile and distill a good deal of info dispersed through the “EA-sphere”. But I was hoping that they would attract more attention than they have done so far.

Did it work? Well, sort of. I think it is useful to have high-quality material in Portuguese to point out when someone asks about EA. UFU’s course has just started, and the students will likely use the content – and others afterwards, if UFU keeps the material available online. So, my first goal was achieved; but back then, there weren’t many others translating EA content into Portuguese, and now I get to know a new project each week. On the other hand, when it comes to measurable traffic, I am convinced that, except if your translation is supporting a larger community-building project, buying ads is probably more effective than posting new content; from now on, I plan to have a specific budget only for this.

Part of the problem is (as I have argued with other people interested in discussing the impact of translations) that, except for high-schoolers (and even then, only for those who were not encouraged to learn reading in English – which is still pretty common in developing countries, but getting more rare), most people who are “potential EA material” can actually read in English well enough – it is just that they are more likely to read content in their own language, because of contingencies such as personal interests, social circles, online recommendation systems, etc. So translating web content is one way to improve on that—to fight for their attention; I suspect (epistemic status: about .47) that an even better way is to get translations of good books mentioning EA reviewed by major outlets (yeah, I hope to see a translation of WWOTF soon).

Also, as I mentioned in the beginning, I thought this project would be synergetic with my other plans; I was wrong – they are actually competing for my time and attention, and I’m not sure that translating web content (or funding it) is now the best thing I can do.

Next steps

The project is still ongoing—until I have spent the funding assigned to translations, which I predict will happen by the end of the month. More recently, I have been supporting the translation of additional EA Handbook content to be used by the Condor Initiative’s EA Intro Program; this hasn’t been published online yet.

One thing I have considered to do, in the future, is rewarding the best translations (and / or original content) in Portuguese ex post, like a certificate of impact—instead of choosing what should be translated in advance on a case-by-case basis. But that would require designing a mechanism, settig rules, etc.

Also, we expect (Juana and I) will start translating content from <riesgoscatastrophicosglobales.com> (if they agree to that) - so as to supply the needs (or attract the attention) of Portuguese-speakers interested in GCR. In the future, I consider translating and publishing content that might be relevant to promote philosophical debates concerning Longtermism, such as SEP entries Intergenerational Justice, Ramsey and Intergenerational Welfare Economics, The Repugnant Conclusion, and The Nonidentity Problem (it’s quite likely I will fund it myself).

[1] Adriana’s suggestion, if I remember it adequately. Only Luciano Cunha (Animal Ethics) replied, recommending some texts on S-risks (the first to be translated) and wild animal welfare.

[2] Why not <altruismoeficaz.com.br>? Well, in part because of convenience: this page is managed by my friend José Oliveira, who lives in Porto, while <80000horas.com.br> is managed by Fernando, who lives in São Paulo – where I used to live until September. Also, because we were then focused on building the Brazilian community (as I said, one of the goals of the project was to supply material for UFU’s course). Finally, because Jose is already doing a really good job and has already made <altruismoeficaz.com.br> kind of authoritative on EA content in Portuguese – while <80000horas> barely showed up in Google searches.

[3] Here are (in Portuguese) Fernando’s notes on how the traffic changed after we started, and the expert’s informal notes on what they delivered.

[4] When permission is not granted, we post instead a text commenting on the original material, displaying only excerpts. I believe this falls under the fair use exception, but would like to know if anyone disagrees.

If I seem too sensitive regarding copyright, that’s because I have seen stories when this went wrong, and people received a payment order out of nowhere because of pictures or texts published eons ago.

[5] An anecdote. Someone (I don’t feel comfortable dropping names) got in touch with an author to request authorization to translate a text included in the last EA Handbook; the author mentioned that such text was first published on a blog of the organization they used to work for, so our friend had to request permission to that entity. But then, the org replied that the post was a bit too old, they had changed their views and so didn’t recommend translating it (“but it’s on the EA Handbook!” my friend thought). Because of this, I consider the current EA Forum CC-BY policy to be one of the major achievements in the movement in the last year – and other translators know I’m not joking.

[6] I hired translators and revisors to do the work. With few exceptions, I would only proofread the material after it had already been revised – and then pass it to the webpage, who would still make minor editions.