Note that each of our funds raised estimates for 2021-2025 subtracts projected annual operating costs.
 Our approach to cost-effectiveness analyses is described on this page. The factors we take into account when deciding how to morally value different outcomes, like increased income versus death, are described here.
 We did recommend a $500,000 grant to GiveDirectly in 2020. Each year since 2016, we have recommended that Open Philanthropy make a minimum grant to each of our top charities. These "incentive grants" are designed to compensate charities for the time they spend engaging with our process and for the risk they take by allowing us to write publicly about their strengths and weaknesses.
 We have occasionally funded top charity programs that were modeled as less than 8x in our cost-effectiveness analyses. For example, see this $200,000 grant we recommended in 2020 to Sightsavers' deworming program in the Democratic Republic of the Congo, which we estimated at 2x cash. These cases have occurred when changes to our cost-effectiveness model caused an opportunity (that we had previously believed to be 8x or better) to drop below our 8x threshold. We still decided to fund these opportunities because either:
- The grants were small and to organizations with which we had a history of grantmaking. Because these organizations had planned around GiveWell's funding, we chose to make some of these small grants to be a good partner. This also gave us more time to plan a withdrawal of funding that would be minimally disruptive.
- There was specific additional work we wanted to do on the model that we thought might push it above 8x, and we wanted to maintain the organization's operations, so that we could have the option to direct additional funding to it in the future.
 In practice, the gap between spending at GiveDirectly compared to our other recommended charities is closer to three or four years. We typically fund organizations a few years into the future, so it often takes two or three years for a donation to be put to use. While we haven’t directed grant funding to GiveDirectly recently, it can put funds to use very quickly. You can read more in: Why we’re excited to fund charities’ work a few years in the future.
 This comparison isn’t exactly right. We think there is lost impact by holding money to spend in the future. But we would be surprised if the lost impact was greater than 10% each year. These differences wouldn't outweigh large differences in cost-effectiveness (e.g. GiveDirectly versus opportunities that are 5x or 8x better).
 This is all of the money GiveWell directed to charity in 2020, excluding about $40 million donated directly to GiveDirectly by GiveWell donors (see our 2020 metrics report). Because these funds are below the threshold for cost-effectiveness we have for this pool of funding, we’ve excluded them from this total.
 It takes time to allocate donations. We typically grant donations to our Maximum Impact Fund on a quarterly basis. While most donations are granted in the same calendar year we receive them, donations from October-December of any given year are typically granted by the end of the first quarter of the following year.
 This includes money we grant from our Maximum Impact Fund, as well as grants we recommend to Open Philanthropy.
 This figure excludes the funding GiveWell donors gave to GiveDirectly in 2020, since it is below our cost-effectiveness threshold. See our 2020 metrics report for a complete breakdown of the money we directed in 2020.
 We plan to publish reports on these investigations in the near future.
 We expect the majority of our funding will be directed to opportunities that are 8x or better, however we don't have specific estimates for the breakdown of funding opportunities in 2022 by cost-effectiveness.