# Deep investigations of new causes

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

As discussed previously, GiveWell and Good Ventures have identified several philanthropic causes that seem like promising candidates for “deep dives” – investigations deep enough to be headed toward major giving recommendations. (In this post, as in the previous one, “we” should be taken to refer to both GiveWell and Good Ventures.)

We’ve had a lot of internal discussion about how we might investigate a new cause at this level of depth. Much of our discussion has centered around the idea of a “program officer” or “program director” – a staffer who is dedicated full-time to a particular cause, and therefore can invest very heavily in getting to know the relevant people, organizations and literature. However, at this time, (a) we don’t feel that we can spare any of our generalist staff for a full-time investigation into a particular cause; (b) there aren’t yet any people we’re ready to hire as cause-specific program officers. So we’re thinking about what we can do in the meantime to find potential program officers, as well as potentially make progress on “deep dives” in other ways.

Below are some of the possibilities we’ve considered. Many involve the possibility of expenditures in the range of $100,000 – whether grants or research consulting expenses – and we address this issue in a later section.* Possible ways to move forward on “deep dives” • Consulting arrangements with potential program officers. Along the lines of the “trial hire” approach we’ve mentioned before, we might work with people who seem like potential fits for the “program officer” role (we’re still not sure just what constitutes a good fit, but are generally interested in people who show both interest in the work and background that could be highly relevant to the cause) on a consulting basis – retaining them to source possible giving opportunities and/or outline strategies for sourcing such opportunities. An arrangement like this could lead to a full-time hire. • Consulting-only arrangements. We might work with a consulting group in order to develop a strategy for sourcing giving opportunities (and for narrowing our focus) within a cause of interest. This option could be more expensive and less directly connected to finding program officers than the above strategy, but we are interested in trying it, as it appears to be relatively common among major foundations. • Funding “working groups.” We might fund several people who have relevant background in a cause to collaborate on a proposed strategy; this is another suggestion that was made by a major foundation. • Exploratory grants (in the general range of$100,000, though possibly more or less) to small organizations, startup organizations, academics and others with spare capacity, and/or cause-relevant departments of existing organizations. These grants would not be vetted as thoroughly as our typical recommendations; rather, as with the above options, they would be treated as learning opportunities. If we can find people who have concrete ideas now about how to spend money productively (or who form concrete ideas after finding that we’re open to such grants), we believe that funding them and checking in on their progress could be as good a way to learn about the cause – as well as to refine our views of which people and organizations can most effectively use funding to accomplish good and communicate to us about their progress (the major qualities we’re looking for in a program officer) – as the above options.
• Exploring key questions ourselves. For some of our causes of interest – specifically malaria control and geoengineering – we’re less interested at this stage in “growing the field as a whole” than in answering particular key questions. In these cases, we might explore these questions ourselves, while keeping an eye out for anyone who is particularly helpful in our exploration and could be retained as a consultant or (down the line) as a program officer.

In general, we aim to take an attitude of

• Eagerness to experiment: when opportunities arise that seem reasonable and not overly expensive (more on expenses below), we plan to take them without a large amount of due diligence, in the hopes that we’ll learn more from trying than from investigating these possibilities.
• Selective followup: our guess is that many of the things we try will not lead to much in the way of outcomes or learning, and that if we aren’t selective about how we spend our time, we will risk sinking a lot of time trying to get good information from uncommunicative grantees and/or partners (this intuition is based partly on our past experiences trying to get information from potential top charities). We explicitly reserve the option to stop following up on any given grant, consulting arrangement or other project if we feel that we aren’t learning enough for it to be worth our time. Conversely, if a particular grant or arrangement leads to an important-seeming question, we may open an in-depth investigation.

The financial costs of learning
Most of the above approaches involve funding somewhere in the range of $100,000 per project over the next year (note that we could potentially fund several projects within a single cause). We are comfortable with this, and believe it is appropriate to engage in promising projects along these lines without much hesitation, whether they involve consulting engagements, grants, or both. If, over the next year, we engage in 50 projects (grants, consulting engagements, etc.) with an average budget of$100,000 each, that would constitute a total of $5 million in grants – a fraction of a percentage of Good Ventures’s projected lifetime giving, alone. If those funds can have any noticeable impact on (a) our learning, and thus ability to find better giving opportunities in the future; (b) helping people and organizations build capacity that can productively absorb later funds, they will be well spent. Note that we see success at (a) and (b) as increasing (and not just improving) our future money moved – we believe that we will have more influence if we are able to point to better giving opportunities. Currently, we’re very short on capacity – we have more to do than we have people to do it – while the total amount of money being spent is relatively small in the scheme of what is and (hopefully) will be available. Therefore, we think it’s appropriate and important to move further away from our historical insistence on deeply investigating all grants before they’re given, and instead move closer to an attitude of “money that might help us learn and build grantee capacity is money well spent.” We still don’t think it makes sense to spend money where there isn’t at least a reasonably large chance of our following up on it and deriving value from it, but in cases where there are real potential benefits of the kind detailed above, we think it’s appropriate to make grants without much hesitation. And we think that requiring ourselves to deeply investigate and/or write about each grant would sustain the current imbalance between funding (of which there’s more available than we’re using) and capacity (which we’re short on). A couple of other notes: • We believe that being willing and ready to spend money, in and of itself, could have substantial benefits for our ability to attract potential consultants, program officers, and giving opportunities. • Consultants and program officers are likely to be more interested in working with us if they perceive us as ready to put funding into a cause. • In the past, we’ve had the experience of having people bring funding opportunities to us based on their understanding of what we’re interested in. On the other side, we’ve looked for funding in the past by seeking out funders that support others in our space. By committing funding in a cause of interest, we hope that we’ll make ourselves easier for people interested in this cause to find. • Along the lines of being ready to commit funding when it could improve future giving opportunities, we are also looking for cases in which seed grants could help create better giving opportunities by our traditional criteria. If, today, we ran across a giving opportunity like GiveDirectly’s early$100,000 grant via the Unorthodox Prize – a chance to help a potential future top charity get off the ground, and in the process gain benefits similar to those detailed above – we would take that opportunity as well.

*Grants will likely be funded by Good Ventures. There may be cases in which GiveWell is an intermediary.

• BJ Terry on October 10, 2013 at 9:56 pm said:

It seems like you should be cautious of having program directors dedicated to a single or small set of causes. Part of the strength of GiveWell is that it compares the seemingly incomparable to provide the most efficient answers. But as soon as you give people territory, they will, possibly unwittingly (but also wittingly, as no one hires perfectly), be biased towards solutions in that area. This probably isn’t a huge concern as long as you are small enough, but at some point it can lead to misaligned incentives.

I don’t have any great ideas on avoiding this problem immediately, but perhaps you could have program directors with partially overlapping expertise and responsibilities, so each cause has two people who are completely up-to-date, but they also have other causes which they do not share that they can compare it to. This has the downside of every cause having “two bosses” a la Office Space. You already have co-Executive Directors so are probably familiar with any challenges there, though.

Another idea, just because I feel bad about only having one idea, would be to use internal betting markets to try to reveal as much hidden knowledge as possible about the different causes.

A third idea would be to have a group that randomly does deep-dive audits on the individual causes to determine that the cost effectiveness is being properly compared and that all qualitative factors are being effectively raised. This seems like it might be too much overhead and too bureaucratic.

• BJ Terry on October 10, 2013 at 10:45 pm said:

Someone told me that I should provide more concrete examples of idea A in my last comment, about having program directors with partially overlapping program responsibilities.

Suppose that you are at Google and the company is growing, so you need to determine where to allocate resources to grow the company most effectively. If every product director has responsibility for a single product, and you ask them how much investment they need, each will try to find the maximum amount of investment that they think they can possibly justify via the magic of PowerPoint and Excel, and then it will be up to their superiors to figure out how to discount their exaggerated requests for new budget. If instead each product director has responsibility for two areas, you can ask which of them needs MORE investment, or force them to put together their ideal joint budget, thus allowing you to rank the products’ investment possibilities straightforwardly.

Now suppose you are at Google and the company is losing tons of money, so they need to do some cutbacks. If you have a single product director for each product and you ask people, “how many people can you do without” they will try to protect as many of their people as humanly possible. If, on the other hand, each product director has responsibility for two products, you can ask them all which of their products is better or most in need of resources. If both of the product directors for a product point to their respective alternate product responsibilities, you know that the product in question is probably not one of your most important, so you can feel free to slash and burn.