Donors want to find the most cost-effective ways to save lives. If spending a ton of money on fraud prevention doesn't improve cost-effectiveness, most donors would argue that it doesn't need to be done.

Thanks for the thoughtful write-up! A few things came to mind while reading:

• Part of the post felt like a false equivalency - to my knowledge $100M spent on animal welfare would actually net virtually no funds to conservation, as opposed to other approaches. Indeed as CB pointed out, many of the ideas people are pursuing are anti-conservation (I admit I am biased against funding wild animal suffering programs) - they actually openly advocate for further manipulating ecosystems.

• One particular section also caught my eye: "Zoonotic diseases, such as Ebola, SARS, and COVID-19, often emerge when ecosystems are disrupted, forcing animals and humans into closer contact. The loss of biodiversity weakens natural barriers that prevent the spread of these diseases. A diverse ecosystem can act as a buffer, reducing the chances of pathogens jumping from animals to humans." I am not sure how these two points flow from one another. I think it's perfectly possible to have biodiversity and still zoonotic diseases - as you correctly note zoonotic diseases primarily emerge from our interactions with wildlife, which feels independent of the biodiversity - the 2009 Swine flu epidemic is a good example of regular factory farming causing outbreaks of zoonotic disease. Many additional epidemics like SARS and Covid-19 appear to point to wet markets as their source - in other words its our eating of animals that is causing zoonotic disease - unrelated to the biodiversity of ecosystems.

• One thing that I appreciate about this post is the difficulty noted in weighing diffuse benefits from specific ones. I have no doubt that conservation of land and biodiversity has positive impacts for animal and human lives (e.g. preserving floodplains for water/flood management). Diffuse benefits in better temperature management, improved likelihood to identify antibiotics, etc., are difficult to quantify but 'feel' right. However 'feel' right is also what EA would counter in avoiding ineffective charities. I think there is more to be done in trying to quantify potential benefits. I wonder if there are opportunities to more quantifiably learn from projects like the Gorongosa Restoration Project that you cite.

For inspiration on implementation, see: Planet Money #915 How to Meddle In An Election

Question about the chart which is summing Q5 + Q6 - where is the 150%+ number coming from for depression (as an example)? Reading the chart before and after, I'd assume it should be at ~55%.

I don't know if this is what you are envisioning, but check out Blue Ridge Labs - they do a fellowship where they get techies to apply to an 'incubation period' focused on solving a social issue and teams form during the initial discovery phase. Many startups get major funding and are profitable.

Could be a good format to follow.