Donations List Website: tutorial and request for feedback
Contents
Introduction
Donations List Website (DLW) is a website that tracks the public or explicitly shared donations of individuals (mostly in the effective altruism and rationality communities), private foundations, and other entities. The website was created and is maintained by Vipul Naik, with contributions from me (Issa Rice).
The aim of this post is to introduce DLW to the world and gather some feedback. DLW has been in active development since January 2017, and is planned for launch by the end of December 2019. At the moment DLW is in a pre-release state, where Vipul and I are adding both data and features to the site.
The first half of this post gives a tutorial of DLW, focusing especially on the features of the site that are not obvious to casual inspection. The second half of this post is a request for feedback; Vipul and I are interested in using the feedback to help us decide where to focus our efforts for future development.
Tutorial
This section is a tutorial of the main features of Donations List Website (DLW), with a focus on the parts that are probably non-obvious to casual users.
Main page
The main page of DLW shows information aggregated from all donations tracked by the site. The main page has the path starting with /
or /index.php
.
The main page of DLW has an optional parameter called display
that changes the kind and number of donations listed. Currently, the possible values are:
individual-scale
, which displays donations made by individuals (in other words, excluding donations made by private foundations and other group entities)full
, which displays all donations tracked by DLWempty value (the default), which restricts donations to the top 30 donors, donees, and cause areas
Donor page
Each donor tracked by DLW has a donor page. Donor pages have a path starting with /donor.php
. An example is the donor page for the Open Philanthropy Project.
For the Open Philanthropy Project in particular, there is a table of disclosures (from before it stopped announcing disclosures). Currently no other donor has disclosures data.
Donee page
Each donee tracked by DLW has a donee page. Donee pages have a path starting with /donee.php
. An example is the donee page for LessWrong 2.0.
Some entities are both a donor and a donee, in which case there will be separate pages for each role. An example is the Berkeley Existential Risk Initiative (BERI):
Donor–donee page
It is sometimes interesting to look at a specific donor–donee pair and see all the donations made by the donor to the donee, their relationship, a list of documents that mention both of them, and so on. For this, DLW has a donor–donee page for each such pair. Donor–donee pages have a path starting with /donorDonee.php
. An example of this is the page for Open Philanthropy Project donations made to the Machine Intelligence Research Institute.
Influencer page
The influencer page looks at the amount of money influenced (a.k.a. the “money moved”) by an entity. This is tracked separately from donations, so it is possible to compare the amount of money moved that is claimed (gathered from the entity claiming money moved, e.g. GiveWell) with the amount of money moved that is actually accounted for (gathered from the entities making or receiving the donations, e.g. Against Malaria Foundation or individual donors). Influencer pages have a path starting with /influencer.php
.
The only entity for which we currently track money moved information is GiveWell. You can see the amount of influence claimed vs amount of money accounted for in the money moved table.
Even for non-GiveWell entities, influencer pages exist. An example is Chloe Cockburn. In these cases, there is no comparison between the “money moved claimed” and “money moved accounted for”, because we have not entered any data for the “money moved claimed”. However, it is still possible to see a list of all donations that were influenced.
Cause area filter
Most pages on DLW accept an optional cause area parameter called cause_area_filter
. Using this parameter filters the donations to show only those with a matching cause area.
To give some examples:
Open Philanthropy Project’s AI safety donations (note the
cause_area_filter=AI+safety
in the URL)AI safety donations in general (note the
cause_area_filter=AI+safety
in the URL)Since the cause areas used on DLW track sub-cause areas as well, it is possible to use it a little like a search feature. For example, using
cause_area_filter=chicken
on Open Philanthropy Project donations shows all donations that have the string “chicken” somewhere in the cause area.Chloe Cockburn donations influenced with sub-cause area matching “bail reform” (for influencer pages, currently the filtering only works on some of the tables)
The cause area filter supports regular expressions, so for example you can search for all Open Philanthropy Project donations with cause area containing “AI safety” or “chicken”. (This example is a bit silly and I’m not aware of a better example.)
Donee pages don’t have this option, because most donees only work within a single cause area.
About page
The about page discusses how DLW is built and reliability concerns for the data.
Request for feedback
This section is an official request for feedback from readers of this forum. Vipul and I want to make Donations List Website (DLW) as useful as possible to people who think about effective giving, flow of money in the non-profit world, real-world decision making, and similar topics. To help us decide what improvements to prioritize, we would like to hear from you what excites you most about DLW, what you could see yourself using it for, what improvements would make you use it more, and so on.
Below, we provide some structure for feedback. We would appreciate feedback as (1) comments on this post, or (2) submissions to this Google Form. The differences between the two are that the Google Form has fields for each question (more structure), can be submitted anonymously, and individual responses will not be shared publicly.
Did you already know about DLW? Do you use it for anything?
To start off, we are curious to get a feel for whether people have already discovered DLW on their own and whether they use it for anything.
As an example response, I will talk a little about my own experience using DLW. I’ve found it useful for the following:
Since October 2017, I have been working on a website called AI Watch that tracks activity in the world of AI safety. More recently, I’ve expanded the scope of AI Watch to cause areas outside AI safety, resulting in Org Watch. Having access to data from DLW has helped me find names of people involved in various cause areas, which has made data collection faster for my work on these other websites.
I had been curious about the donations of certain individuals (examples: Vitalik Buterin and gwern) but as far as I could tell, there was no canonical location listing donations made by them. DLW has solved this problem for me for these individuals.
One day, I suddenly became curious about the indirect flow of money from the Open Philanthropy Project (Open Phil) to the Machine Intelligence Research Institute (MIRI). By “indirect flow of money” I mean not grants that Open Phil has made to MIRI, but where Open Phil grants money to some intermediate entity, who then donates to MIRI. Having access to both the DLW and Org Watch datasets meant I could do some exploration by looking at (1) organizations that had received money from Open Phil, that had also donated to MIRI, and (2) individuals who worked for organizations that received funding from Open Phil, who were also individual MIRI donors.
Potential improvements – what are your thoughts?
This subsection lists some potential improvements we’ve thought of.
Add more foundation giving data
Currently, DLW tracks the giving data of foundations including the Gates Foundation, Arnold Foundation, Hewlett Foundation, Good Ventures/Open Philanthropy Project, and Ford Foundation. However, there are many more foundations whose donations we could track.
Add more individual giving data
Similar to the above, there are many individuals whose donations we could track.
Add more data about things related to the effective altruism or rationality communities
DLW already has an emphasis on these communities, but there’s still more that can be added.
Add more money moved data
Currently we have only entered claimed money moved data for GiveWell. However there are some other entities that publish money moved data, like Animal Charity Evaluators.
Systematically track considerations that donors make
Donors concerned with effective giving have many considerations as they think through their decision. These include:
Giving now vs later
Donating to a single charity vs multiple charities
How to weigh local knowledge vs analyses published by others
How much weight to place on cost-effectiveness calculations vs less quantitative forms of modeling and reasoning
Whether to use a donor-advised fund
Size of average donation (relative to total wealth)
Which population ethics to use
Which worldviews to use, and what entities to see as “equal” (all animals? all humans? all people with a certain IQ or ability?)
Currently, the reasoning above is scattered across many different blog posts, comments, articles, and so forth. It is difficult to answer questions like “Where and how do people who give consideration to the giver’s dilemma donate?” and “Where do people who share my population ethics and worldview donate?”
It seems plausible that DLW could make progress on aggregating some of this discussion and linking it with the existing donations data.
Classify cause areas better
Although donations on DLW are usually tagged with a cause area, so far this has been done in a somewhat ad hoc manner, so there are some problems like the same cause area being phrased in different ways and the cause area hierarchy being a mess.
I did some work on this previously in a cause prioritization context (see entries for cause area classification and list of classifications of philanthropy on the Cause Prioritization Wiki), but this sort of work is also valuable for better classifying donations.
Include giving data (of existing donors/donees) in a more timely manner
Even after we add donations data for a donor or donee, new donations are made over time, so there is a question of how often DLW should update the data for existing donors and donees. Currently the updating schedule is somewhat arbitrary.
We have written up a series of steps for updating existing data for donors and donees, so updating donations data for most entities is pretty straightforward.
Include giving data (of existing donors/donees) more completely
Occasionally a subset of donations made by an entity is easy to access and process by a machine, while the rest are not. An example of this is the Gates Foundation, where donations tracked through the International Aid Transparency Initiative are easy to process, but where the rest of the donations are difficult to process at the grant level.
Add more documents
There has been a lot of discussion over the years about effective giving, personal donations, organization updates, and other documents that would belong on DLW. We have added only a fraction of the possible documents, so adding more is another option.
Add more graphs
Generally we try to add graphs for each table displayed, but there are some exceptions like the donee page.
Summarize existing data more
It might be the case that the data of interest is already part of DLW, but isn’t presented in the most useful way. Examples of this are donor intersection pages and donee intersection pages, where one wants to compare two specific donors or two specific donees.
Other features?
We are interested in hearing ideas other than the ones we listed.
Acknowledgments
Thanks to Vipul Naik for feedback on this post and for funding my work on it.
- Why “cause area” as the unit of analysis? by 25 Jan 2021 2:53 UTC; 85 points) (
- Funding chains in the x-risk/AI safety ecosystem by 9 Sep 2019 2:56 UTC; 22 points) (
- 18 Dec 2020 14:56 UTC; 1 point) 's comment on Propose and vote on potential EA Wiki entries by (
I just used DLW as it was a bit unclear to me where my funding ultimately came from. And I wanted this to be transparent to people I am interacting with outside of EA, in order not to come across as obfuscating who is funding my work. Perhaps this information on ultimate funding source was available elsewhere, but searching around, DLW was the first source that let me rather accurately track funding back to its original source.
Cool project! I’m impressed by the thoroughness of the lists.
This will take a lot of ongoing attention to keep up-to-date. Are you planning to do list maintenance as ongoing volunteer work? Are you intending to seek funding for ongoing maintenance?
Also, who do you have in mind as your intended customer?
So far almost all DLW work by me has been contract work paid by Vipul, and my guess is it will stay like this for the foreseeable future. We have not sought other sources of funding, but are interested in receiving funding (for my ongoing work).
As mentioned in the post, the intended audience is something like “people who think about effective giving, flow of money in the non-profit world, real-world decision making, and similar topics”. To give some examples:
Ryan Carey has a GitHub repository with data analysis of Open Phil’s grants data. (Last updated 2018-02-11.)
Here is another example, but part of a closed Facebook group so I won’t describe it. (Probably last updated around 2018-01-14.)
Michael Dickens wrote a post entitled “Where Some People Donated in 2017” recording where some EA and EA-peripheral people donated. (Last updated 2018-02-14.)
In June 2017, there was a Facebook event called “What’s Up With the Open Philanthropy Project?” The event looked at some of Open Phil’s work, compiling some documents about some of Open Phil’s grants in the process. Looking at the timestamps, I think the Google Docs were created when the meetup began, and were filled in during the course of the meetup. (Last updated 2017-06-09.)
I think in all the above cases, either the current version of DLW or an improved version in the future performs a superset of the data collection/analysis, is continually updated, and provides a single location for all the data and analysis.
Vipul has also made comments (1, 2) and at least one post using in part data collected by DLW, to make observations or answer people’s questions.
I’ve found the service to be pretty valuable so far, though mainly for keeping track of OpenPhil. I have funded $645 of work on this (via Vipul to Issa) so far.
This seems like a huge amount of work for what seems like a somewhat uncertain impact.