Purchasing research effectively open thread

by John_Maxwell1 min read21st Jan 201515 comments


Effective AltruismOpen Threads
Personal Blog

Many of the biggest historical success stories in philanthropy have come in the form of funding for academic research.  This suggests that the topic of how to purchase such research well should be of interest to effective altruists.  Less Wrong survey results indicate that a nontrivial fraction of LW has firsthand experience with the academic research environment.  Inspired by the recent Elon Musk donation announcement, this is a thread for discussion of effectively using money to enable important, useful research.  Feel free to brainstorm your own questions and ideas before reading what's written in the thread.

15 comments, sorted by Highlighting new comments since Today at 9:04 AM
New Comment

In terms of funding academic research, I think there's good reason to think that funding think tanks is probably best. Specifically, let's talk about George Mason University, the Kochs, and the Mercatus Center. Researchers are cheap to support, and by having one particular home that attracts many ideologically similar people, you get additional useful social networking effects that you might not get if you just gave some grants with the money. (Go down the Mercatus Center fellow list, and you'll see some familiar names like Robin Hanson and Tyler Cowen and (if you're libertarian / into open borders) Bryan Caplan.)

Researchers are cheap to support

Humanities researchers. Supporting, say, a high-energy experimental physicist can get quite expensive X-)

Researchers outside the physical sciences tend to be inexpensive in general - e.g. data scientists / statisticians mostly need access to computing power, which is fairly cheap these days. (Though social science experiments can also be costly.)

data scientists / statisticians mostly need access to computing power, which is fairly cheap these days.

This is true for each marginal data scientist. But there's a catch, which is that those folks need data. Collecting and promulgating that data, in the application domains we care about, can sometimes be very costly. You might want to consider some of those as part of the cost for the data science.

For example, many countries are spending a huge amount of money on electronic health records, in part to allow better data mining. The health records aren't primarily for scientific purposes, but making them researcher-friendly is a big indirect cost. Similarly, the census is a very expensive data-collection process that enables a lot of "cheap" analytics downstream.

While each data scientist might be cheap, there was a big up-front investment, at the national level, to enable them.

Actually, my uninformed guess (from casual familiarity and friendship with people from various fields) is that physics and chemistry are cheaper than biology to conduct. There's expensive equipment in Phys/Chem but it can be re-used over and over again by multiple labs. Biology on the other hand has major recurring costs in the form of maintaining animal populations and greater degree to which replication is important. And then things get cheaper again in the psych/social sciences, where experiments are often either computerized or conducted by undergraduates for credit and conducted on volunteers.

Basically, if you graph xkcd!purity by price, I think it is bell shaped with Biology at the peak. In an absolute "per researcher, per experiment" sense. That's not to say that biology might not be "cheaper" in terms of return on investment from an EA standpoint.

Well, statisticians can be thought of as kinda-mathematicians and George Pólya has remarked that

Mathematics is the cheapest science. Unlike physics or chemistry, it does not require any expensive equipment. All one needs for mathematics is a pencil and paper

Otherwise, e.g. pharma research and biochem in general can get pretty expensive.

In maths, you also need a waste-paper basket. Pólya must have been thinking of philosophy.

Agreed that equipment is costly, and thus the research as a whole is more expensive, but I think that has a small impact on the cost of the researchers. I also am under the impression that 'think tank' refers mostly to policy research--even if you've got a plasma physics think tank, say, they'll mostly try to answer the question of what plasma physics projects should be pursued and other meta-level improvements to the field rather than building tokamaks themselves.

As a person who's relatively unfamiliar with academia, here are some questions I have, along with guesses about their answers.

  • Is a large chunk of capital like Musk's better dispersed slowly or quickly? Ceteris paribus, slow dispersal gives an opportunity to learn what seems to be working, allows later grantees to build on the work of earlier grantees, allows serial grants to highly qualified/successful researchers, and might incentivize researchers to establish a track record or "resume" in the area where grants are being awarded to look (and be) more qualified for later grants. On the other hand, fast dispersal brings the positive results of the funding more quickly.

  • Is a large chunk of capital like Musk's better dispersed in smaller quantities to many individuals or in larger quantities to a few? Fewer larger grants might attract more qualified people; more smaller grants get more people involved and allow more leads to be followed.

  • What causes some research areas to get "hot", and how might philanthropists play a role influencing this? (For example, if Elon's money succeeds in funding research that's impressive by conventional standards, will that lend credence to the broader topic of AI risk reduction by association?) What's the best way to motivate researchers in general?

  • What's the best way to promote/advertise the availability of grant money in order to attract the best research proposals?

  • What are some tips for evaluating research proposals and the people who propose them (especially given that the proposals and the resumes of those who propose them have all been written to seem as attractive as possible?) Could research proposals be made available for public commentary before being accepted/rejected? (In the extreme case, this public commentary might take the form of a prediction market.) Or would some researchers be reluctant to share their ideas before they'd been fully developed, lest someone else take credit, so the effect would be to cut down on the number of proposals?

  • What people or organizations currently fund scientific research most effectively, and what do they have in common?

  • Givewell wrote a series of blog posts on scientific research as a charitable cause, focusing on life sciences research: 1, 2, 3. The second blog post discusses some ways established life science research funding mechanisms may fall short, providing opportunities for philanthropists; it also describes a "generalist scientific advisor" role Givewell is (was?) recruiting for that readers may be interested in applying for. Some takeaways I got from the series & comments on it: It's important to consider the current scientific funding environment in order to contribute in a non-replaceable way. To fund scientific research effectively, either the giver or their advisors should have good knowledge of the field they're funding. Scientists may consider basic research more prestigious/motivating than applications research.

  • I imagine if grant money is presented properly, receiving it could carry more weight than just the capital itself, in the same way that receiving a Rhodes Scholarship is more meaningful than just getting money to study at Oxford with. This could provide both an extra incentive to qualify for a grant and increase the prestige of the grant winners and the research they produced. Is this feasible and desirable, or a game best not played? (For a point favoring the latter position, see the "Nobel-Prize effect" discussed here.)

BTW, here's some previous related discussion on LW: Funding Good Research, The efficiency of prizes.

Is a large chunk of capital like Musk's better dispersed in smaller quantities to many individuals or in larger quantities to a few? Fewer larger grants might attract more qualified people; more smaller grants get more people involved and allow more leads to be followed.

This distinction may be less sharp than it seems. The recipient of a very large grant tends to develop a process of deciding how to use it that looks very much like another grant-awarding process. In effect, a really large grant subcontracts the work of disbursement. That's not necessarily a bad thing or a good thing, but it is a thing that happens.

One of the problems with slow dispersal is that due to Iron Law of Bureaucracy-type reasons, the money will likely not be dispersed in useful ways. See, for example, the decline and drift of the Ford Foundation and March of Dimes. This is why, for instance, the Gates Foundation will not exist in perpetuity but has an expiry date.

What is a good source that will help me evaluate the merits of donating to scientific research vs., say, the merits of donating to global poverty and disease alleviation?

Before thinking about how to best donate to research, I'd generally want to know about the merits of research donation in general.

My priors say research is important and underfunded to a degree comparable with how global poverty alleviation is important and underfunded, but how to quantify which is more important?

That's the kind of question that the Open Philanthropy Project is trying to answer. For example, here Givewell co-director Holden Karnofsky a back-of-the-envelope calculation for DALYs created by life sciences research. Here is a similar back-of-the-envelope question for AI risks research, from former SIAI (now CFAR) staff member Anna Salamon. You might try putting numbers from these calculations in to a spreadsheet or Instacalc to get a rough idea of how the outputs change given different assumed inputs. Previous LW posts on Fermi estimates: 1, 2.

Incredible idea. However I would like to add that while this contains many fruitful directions, a substantial amount of important research is not 'directed'. See 'Happy Accidents' by Meyers and such. Caution is warranted because 'research funding' is like crack to rationalists.

Endowed chairs are pretty expensive.