The GiveWell Blog

GiveWell Labs update

[Added August 27, 2014: GiveWell Labs is now known as the Open Philanthropy Project.]

GiveWell and Good Ventures have made substantial progress since our last update on GiveWell Labs, and we’re now ready to take a major new step: moving beyond “shallow” and “medium-depth” investigations of causes to “deep dives” that are likely to involve grantmaking. This post summarizes our progress so far and plans going forward; a future post will elaborate on our plans for “deep dives.”

Basic principles of our current work
For the foreseeable future, we expect our work to be informed by the following principles, some of which we’ve written about before:

  • As stated in our last update, we work closely with Good Ventures on this research. Cari Tuna is an active partner with us on these investigations, and we see Good Ventures as the initial target for our recommendations. Both GiveWell and Good Ventures anticipate other philanthropists (including some portion of GiveWell’s existing audience of individual donors) eventually participating in funding the opportunities we identify. Throughout this post, “we” should be taken to refer to both GiveWell and Good Ventures.
  • Our main goal is to find the most promising charitable causes; we think of the “cause,” rather than the “charity” or “project,” as the most relevant unit of analysis for us at this point. More at our previous update.
  • Much of our goal at this stage could be described as “learning how to learn”: trying out investigative processes, seeing what sort of information they yield, and reflecting on our processes. Because of this, in many cases we’re prioritizing causes based on the extent to which we have a viable plan for investigating them and personnel who are suited to carrying out the plan; we aren’t prioritizing causes based purely on our guesses as to how promising they’ll turn out to be.
  • We want to perform several different depths of investigation in parallel, because having a sense for how to investigate a cause deeply could influence how we perform lower-depth investigations. Because we think giving can accelerate our ability to learn (more on this below), we will be making grants in cases that are distinct from fully vetted GiveWell recommendations.*
  • Looking so broadly across charitable causes, we are running into a lot of deep judgment calls that we won’t necessarily reach even internal agreement on. Our goal is to find the giving opportunities we believe are best – taking intuitions into account – and we don’t expect to fully formalize our decision making processes (anytime soon). So while we will try to explain why we are making the choices we are, we don’t plan to let development of explicit frameworks hold us up from moving forward on causes that seem worth investigating more deeply.
  • Related to the above point: at this point, our shallow- and medium-depth investigations are focused on finding the information that seems to give the best return in terms of “propensity of information to change our views, per hour spent gathering that information.” It isn’t necessarily the case that these investigations will answer every question one would want to answer in order to choose between causes; the goal, rather, is that these investigations make our choices of causes more informed than they would be otherwise.

Updates on previously discussed activities
(See the “Cause investigations that are currently in progress” section of our May update for previous discussion of these activities)

Lower-depth (“shallow”) investigations. We have now published ten of these, available at In each case, we feel that we’ve gained a basic understanding of the problem, the activities that can be undertaken to address it, and the other funders in the space. We plan to continue doing these investigations, and over time the number of them could grow quite large, as there are many different approaches to philanthropy that could fit our definition of a “cause.”

Higher-depth (“medium”) investigations. We have published one of these – a landscape of the open science community – and have two more (on geoengineering and criminal justice reform) that are each very close to completion. In each case, we feel that we’ve gotten a reasonably representative view of who works on the issue, who funds it, and what the contours of the major debates and key questions are.

All such investigations (as with the “shallow” investigations) have been done by our full-time staff. We haven’t succeeded in identifying suitable consultants for them (as we previously hoped to), so our capacity for these investigations remains limited, but we intend to continue doing them (albeit at a slow rate).

History of philanthropy/philanthropy journalism. We have continued working with the consultant referred to in our previous post on this subject (Benjamin Soskis). He has compiled a bibliography of sources potentially relevant to understanding the history of philanthropy, and is now working on deeper case studies on particularly significant claimed philanthropic successes.

In addition, we have started a relatively low-key (for now) search for people who might write informative blogs about philanthropy. We have had preliminary conversations with Ashok Rao (recommended to us by Dylan Matthews), Shaun Raviv, and Mike Miesen about the possibility of our paying them – on a trial basis – to write pieces relevant to the current state of philanthropy, the activities of major players and promising giving opportunities.

Co-funding with major foundations. Good Ventures has partnered with the Gates Foundation on a project to help contain artemisinin resistance in Myanmar and has also had conversations with multiple other foundations about the possibility of co-funding. We will be reflecting on this work in a future update.

Understanding the basics of scientific research and political advocacy. A forthcoming series of blog posts will discuss our progress on this front. In brief,

  • We have identified some potentially promising approaches to scientific research, but we are currently focused on recruiting generalist scientific advisors, which we have come to view as a necessary component for finding outstanding giving opportunities in this area.
  • We have developed a picture of what to look for in political-advocacy-related giving opportunities, and have started doing shallow- and medium-depth investigations of causes that include (or consist of) political advocacy.

A crucial next step: deep investigations of causes, likely including grantmaking
We intend to continue work on shallow- and medium-depth investigations, but we also feel it’s important for us to get experience – sooner rather than later – with investigating causes deeply. Deep investigations are likely to include “learning grants” in the short term, and hopefully well-grounded giving recommendations in the longer term. At this point, with the amount of shallow- and medium-depth investigations we’ve done, there are several causes that appear to us to represent (a) unusually strong combinations of “important cause” and “lack of current philanthropic presence, at least in some respect”; (b) strong potential learning opportunities. We don’t feel confident that these causes will be the ones we find most promising a year from now, but we believe that investigating at least one of them deeply would be a good thing from the perspective of “learning about learning,” as discussed in the first section of this post.

These causes are:

  • Labor mobility. We were initially interested in this issue because of research suggesting extraordinarily high potential benefits to loosening global immigration restrictions, as well as the arguments of Center for Global Development scholars (Lant Pritchett and Michael Clemens) that it represents one of the more promising and underexplored paths to reducing global poverty. Our impression is that while there are many groups in the U.S. focused on immigrants’ rights, the set of people and organizations persistently advocating for looser immigration restrictions – particularly on humanitarian grounds – is quite small. The combination of potentially high importance and little in the way of existing infrastructure interests us, although there remain many questions about potential negative effects of loosening immigration restrictions and about whether the sort of impact estimated in the literature is anywhere within the range of what might be feasible.
  • Geoengineering research. We were initially interested in this issue because it appears to receive very little attention from funders relative to other climate change responses. (We have since heard about other approaches about which the same might be said, and will likely write about them in the future.) We have completed a medium-depth investigation of geoengineering research (forthcoming) and believe that the field is relatively small. However, the risks of growing the field could be substantial, so we would want to undertake more investigation before making grants in this area.
  • Criminal justice reform (medium-depth investigation forthcoming). As we will discuss in our posts on political advocacy, this cause has been highlighted to us as an unusually tractable political cause, in which results at a sub-federal level might be expected within a few years (more at our conversation with Steven Teles). While there are several funders with substantial budgets in this area, there appears to be very little in the way of funding and infrastructure around the sorts of ideas promoted by Mark Kleiman, which we find intriguing.
  • Factory farming. We aren’t sure of how one should weigh philanthropy that primarily aims to reduce animal suffering vs. philanthropy that primarily aims to empower (or reduce suffering for) humans. But if we placed relatively high weight on the former, we would be very interested in the cause of reducing animal abuse at factory farms, which seems to both (a) affect far more animals than other causes traditionally associated with animal welfare; (b) have far less funding and nonprofit attention behind it. A medium-depth investigation of this cause is in progress.
  • Open science. This cause appears to have more organizations and philanthropic attention than many of the causes above, but still a fairly small amount of philanthropic funding in the scheme of things. We are not as certain about its importance. We find this cause to be a potentially high-value one, but less obviously appealing from both a learning and impact perspective than the above causes.
  • Malaria control. We haven’t done a formal shallow- or medium-depth investigation of this cause, but our research on Against Malaria Foundation has raised many possibilities of underfunded aspects of malaria control, such as research on insecticide resistance. Despite the relatively large amount of funding in the area as a whole, we feel that there may be particularly underfunded aspects, and also that even well-funded aspects of malaria control could still have extremely strong marginal returns to more giving (an example of the latter dynamic is LLIN distribution).
  • History of philanthropy/philanthropy journalism. We initially approached this as a cross-cause learning opportunity but are now starting to think of it as a “cause”: a type of work that could accomplish a great deal of good (by helping future philanthropists to be more effective), but gets very little funding and attention currently.

In all of the above cases, we have done enough investigation – or are quite close to having done enough investigation – to feel that we’ve gotten a reasonably representative view of who works on the issue, who funds it, and what the contours of the major debates and key questions are. There are other causes for which our investigation is still in preliminary stages, but which may turn out to belong on this list after more investigation.

* Grants will likely be funded by Good Ventures, though there may be cases in which they are funded by GiveWell.


  • Uri Katz on October 6, 2013 at 11:55 pm said:

    When you claim that a cause, or sub cause, is well funded, does that take into consideration the possibility that the money is not being well spent? Perhaps in the most over funded causes, the best organizations are overlooked, and they present the best giving opportunities?

  • Alexander on October 7, 2013 at 3:28 pm said:

    Uri – it’s difficult to tell quickly and from the outside how well money is being spent (as opposed to how much is being spent), so the heuristic we’re starting with is looking for underfunded causes. However, we do recognize that we may have a different angle on a cause or sub-cause than other funders, and that can affect the definition of the cause and the assessment of how “crowded” it is. For instance, there are many foundations working on some aspects of criminal justice reform and international migration, but we see little focus on the kinds of ideas promoted by Mark Kleiman and Michael Clemens, respectively, which we believe may be quite important, and accordingly we believe that those causes may be under-funded. To put it another way, how we define a cause or sub-cause is tied up in our assessment of the gaps in what other funders are doing.

  • Uri Katz on October 8, 2013 at 1:44 am said:

    “the heuristic we’re starting with is looking for underfunded causes”
    So long as you eventually make sure that these causes will use new funding as effectively, or more effectively, in terms of achieving their goals (eliminating poverty, preventing climate change, etc.) than crowded causes, that seems like a good approach. This requires figuring out if a lot of funding in a crowded cause isn’t being wasted on unproductive organizations, and if there aren’t underfunded organization that look very promising.
    I understand this may be something you want to do in deep dives, I just think it should not be overlooked.

  • Alexander on October 10, 2013 at 7:42 pm said:

    Uri – we don’t expect to get to a point where we’ll be able to say we’ve looked at every cause; we expect for there to be a variety of causes about which we say, “that cause looked crowded from the outside, and if we learned from a credible source that there were still exceptionally good giving opportunities that are being overlooked, we would look more closely.” Of course, for causes that we do end up recommending, whether crowded or not, we expect to develop an internal view, where we can point to particular giving opportunities that appear to be overlooked and that we believe are competitive with the best opportunities that we have found elsewhere. One example of our work in a crowded cause might be global health: many large foundations and aid donors work on global health issues, but overall, we still believe there to be some particular gaps that we can identify that appear to contain outstanding giving opportunities (e.g. bednets). But I think it’s likely that we will miss some good opportunities in crowded causes simply because we are not well-equipped to figure out that they are good opportunities that are being overlooked for the wrong reasons.

  • Peter Hurford on December 5, 2013 at 6:22 pm said:

    Why is GiveWell not giving any attention to AI control? I know why Holden doesn’t feel confident in donating to MIRI, but AI control in the abstract still feels like an incredibly important issue, even if there isn’t a stellar org currently working on it, and I’d like to see it investigated more in depth.

    Of course, I understand that GiveWell shouldn’t have to explain why they’re not doing every cause just because some random donor thinks it’s important, but AI control is a fairly popular cause among the effective altruist crowd.

  • Holden on December 6, 2013 at 7:49 pm said:

    Peter, we are planning to look into this issue fairly soon. Note that at this early stage, our priorities are partly set by how easily we think we’ll be able to learn about a given cause.

Comments are closed.