If you've visited charity evaluators like GiveWell, you might wonder why the charities they recommend are unfamiliar — where is Oxfam, or Amnesty International? The Happier Lives Institute explains why these large and diverse charities, which they call “Multi-Armed NGOs” (MANGOs), are rarely recommended:

1. Complexity: MANGOs run hundreds of programmes — Oxfam International lists over 1,000 — making comprehensive assessment impractical for small evaluation teams.
2. Dilution: Even if some of the hundreds of programmes at a MANGO are highly effective, your donation’s impact is diluted across their portfolio, which probably includes less effective initiatives. This is a strong consideration, because, as Toby Ord points out with reference to health interventions in particular, the best interventions are sometimes tens of times better than the median. 
3. Fungibility: But can’t you avoid dilution by donating to a specific project a MANGO runs? Many of these organisations also receive unrestricted funding (funding they can allocate to any programme). If you donate to one programme they run, they may move unrestricted funding previously allocated to that programme to another (perhaps less effective) one. In this case, the overall effect of your donation is that a less effective programme gets more funding, and your favoured programme gets no extra funding.
4. Prioritisation mindset: Simply being a MANGO might be evidence that a charity isn’t sufficiently focused on the cost-effectiveness of its programmes. If it were, it would assess them and prune its least cost-effective programmes to do more good in the areas where it excels. Since it hasn’t — perhaps it isn’t sufficiently focused on impact.

It’s worth noting that these aren’t knock-down arguments against donating to popular, multi-armed charities. However, they provide a good reason to be sceptical of their effectiveness compared with FoNGOs (Focused NGOs), which focus on one or one type of programme and have been vetted by evaluators like GiveWell, Happier Lives Institute, or Animal Charity Evaluators. 

Back in 2021, before ChatGPT’s release, Daniel Kokotajlo wrote a scenario depicting the development of AI up to 2026. Much of the picture is strikingly similar to our actual world (chatbots that understand and create images and video as well as text; models quickly scaling up), and some is off-base (AI isn’t used for propaganda in the way this vignette predicts). Now, Kokotajlo (and others) have published a new scenario — AI 2027. 

This scenario is especially interesting now because OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei are publicly predicting that AGI could arrive within the next few years. 

I won’t give away much of the scenario (it contains a race between China and the US, industrial espionage, and highly advanced AI models kept from the public). I highly recommend reading it. If you enjoy it, and want to know more, check out this podcast episode, with Daniel Kokotajlo and Scott Alexander as guests. 

And if you’re looking for more information on when to expect AGI, this article from 80,000 Hours is a good start. 

• Blocking a local factory farm can mean even worse farming elsewhere. It might seem like an obvious win to block the construction of a polluting, cruel factory farm. But if you live somewhere with better than average animal welfare laws and increasing demand for meat (such as the UK), the net effect is more animals being farmed in worse conditions and imported. 
• Reducing the number of wild-caught fish may not decrease suffering. Catching fish from the ocean isn’t humane — fish often asphyxiate on deck, are crushed in nets, or processed live. But when we fish less, and demand remains the same or increases, aquaculture (factory farmed fish) starts to fill the gap. Since the 1990s, “wild caught fishing hauls have stagnated while aquaculture production has more than tripled”.

These examples all show that seemingly positive interventions can cause negative outcomes that we are only aware of if we pay attention to the economics. But this isn’t all bad news. Farm animal economics also helps us understand why the best animal welfare interventions, such as cage-free corporate campaigns, are so effective. 

80,000 Hours

• Podcast Host (Expression of Interest) (Remote or in London, UK/San Francisco, CA /  Washington DC, GBP £80K–£100K+, apply by May 6th)
• Podcast Chief of Staff (Expression of Interest) (Remote or in London, UK/San Francisco, CA /  Washington DC, GBP £80K–£100K+, apply by May 6th)
• Writer-Researcher (Expression of Interest) (Remote or in Berkeley, CA)
• Senior Product Manager (Expression of Interest) (Remote or in London, GBP £76K–£85K)

Fish Welfare Initiative

• Director of Programs (Hybrid in Eluru, Andhra Pradesh, INR ₹2M–₹3M, apply by April 30th)
• Director of Operations (Hybrid in Eluru, Andhra Pradesh, INR ₹2M–₹3M, apply by April 30th) 

Lead Exposure Elimination Project (LEEP)

• Head of Finance (Remote, London preferred)
• Recruitment Manager (Remote, GBP £50K–£55K, apply by April 20th)

Longview

• Nuclear Programme Associate / Senior Programme Associate (Washington DC, USD $90K–$125K)
• Executive Assistant to the Office of the CEO (London, GBP £70K–£85K)

ML Alignment and Theory Scholars (MATS)

• Operations Director (Berkeley, CA, USD $115K–$210K, apply by April 25th)
• Program Lead/Director (Berkeley, CA, USD $130K–$210K, apply by April 25th)
• Research Manager, Governance (Berkeley, CA, USD $48–$82 per hour, apply by May 2nd)
• Research Manager (Berkeley, CA, USD $48–$82 per hour, apply by May 2nd)
• Community Manager (Berkeley, CA, USD $44–$58 per hour, apply by May 2nd)
• Operations Generalist (Berkeley, CA, USD $40–$53 per hour, apply by May 2nd)

Obsolete

• Research Assistant (Remote, USD $37K–$65K for 6 months, apply ASAP)

Open Philanthropy

• Counsel (Remote (UTC-10 to UTC-1), USD $227.6K, apply by April 23rd)
• Strategy Fellow – Global Health and Wellbeing, Partnerships (Remote (must overlap 5–10 hours/week with PT), USD $172.3K, apply by April 27th)
• Partnerships Associate – Global Health and Wellbeing, Partnerships (Remote (must overlap 5–10 hours/week with PT), USD$126.2K, apply by April 25th)

Suvita

• Technology Specialist / Senior Software Engineer (Remote in India, apply by April 21st)

Transformer

• Managing editor (Remote, with a preference for London, Washington DC or San Francisco, GBP £77K–£112K / USD $100K–$145K, apply by April 30th) 
• Reporter (Remote, with a preference for London, GBP £50K–£70K / USD $65K–$90K, apply by April 30th)

We hope you found this edition useful!

If you’ve taken action because of the Newsletter and haven’t taken our impact survey, please do — it helps us improve future editions.

Finally, if you have any feedback for us, positive or negative, let me know!

– The Effective Altruism Newsletter Team