On Creating an Online Information Marketplace for Giving
From CharityScorecard
In 2007, the The William and Flora Hewlett Foundation published an Annual Report that opened with an interesting essay by the Foundation President, Paul Brest, about Creating an Online Information Marketplace for Giving.
I was delighted to learn about the Hewlett Foundation's efforts to create a nonprofit information marketplace. In the process of setting up this Charity Scorecard site I've done a lot of thinking about what sorts of features and characteristics would be useful in an nonprofit information marketplace. In September 2007 I wrote to the Hewlett Foundation to share some ideas for desirable features in nonprofit marketplace. Here are some excerpts from that letter...
Background
Open-source culture
I'm a software developer. I started my career in the private sector, but in 2003 I shifted to working on open-source software projects in the nonprofit sector. I believe that significant practical benefits arise out of the cultural norms that are common in open-source communities, and I'm pleased to see those norms becoming more mainstream. Projects like Wikipedia and OpenCourseWare are wonderful demonstrations of the power of open collaboration on web-based projects. Adopting some of the spirit and methodology of the open-source movement would help increase the chance of success of any new effort to create a nonprofit sector information marketplace.
Problems of an individual donor
In the early 1990s I had a modest amount of money to give away, and I was faced for the first time with the problem of deciding where to give it. This was before the web existed, so it wasn't easy to do detailed research about nonprofits or about social problems, especially for a small donor working alone. I ended up giving most of the money to a few organizations that were doing work on international family planning, and in 1993 I wrote about the experience in an open letter called The Gumption Memo. It was a 50-page memo geared toward individual donors, about trade-offs, prioritization, and decision making in the face of uncertainty. I posted The Gumption Memo on my web site in 1995, back in the early days of the web, and at the time it served as a tool that allowed me to be unusually open and transparent about individual giving, and as a way to invite collaboration with friends, family, and others.
In recent years I've again come into some money that I need to give away. I am faced with the same problem of trying to figure out where to give the money, but in the past 15 years the world has changed dramatically, and the problems and priorities now seem to have shifted from what I saw 15 years ago. The tools available to individual donors have also changed dramatically. New tools like GuideStar and Charity Navigator are enormously useful, and nonprofits themselves sometimes offer significant amounts of information on their own web sites.
For the individual donor, things are much better than they were 15 years ago, but there's still plenty of room for improvement. Organizations today have typically embraced the web only as a tool for one-way publishing of brochures and reports. Few organizations have started to take advantage of the web's potential for moving us toward open content and open public collaboration. Nonprofit organizations seem to be gradually moving toward modest levels of financial transparency, but few organizations are even striving for significant operational transparency.
Back in 1993 I used The Gumption Memo as a vehicle for openness and collaboration in my giving. In a similar spirit, today in 2007 I'm using my Charity Scorecard site to operate openly and to invite others to take part. I'm using Charity Scorecard to collect info about nonprofits of interest, to gauge how well the different organizations fit with my priorities, and to do so with an unusually high level of transparency.
The Charity Scorecard site is in part a small scale information marketplace, but it is also intended to serve as an experimental capital market, where money flows from donors to nonprofit organizations, and where that flow is based on publicly posted information rather than on personal connections or intuitive assessments.
Charity Scorecard is a small-scale, prototype project. At this point the total development effort invested in the site amounts to only a few hundred hours of volunteer labor and a thousand dollars of contracting costs. But, even in its current embryonic state, the site demonstrates a number of important characteristics that are missing from sites like GuideStar and DonorEdge. The Charity Scorecard site meets more than half of the criteria that I'll describe below in my list of suggested Ideas for a new nonprofit marketplace. It would be great to see ideas from Charity Scorecard percolate into mainstream sites like GuideStar. Conversely, it would be great if data from GuideStar were available for use in sites like Charity Scorecard.
Ideas for a new nonprofit marketplace
Sites like GuideStar and DonorEdge have shown the value of publishing, in a standardized format, basic information about nonprofit organizations. But the first-generation sites like GuideStar and DonorEdge are basically “closed” sites. They intentionally restrict access, which limits their usefulness.
In contributing to a second-generation nonprofit marketplace, the Hewlett Foundation has the opportunity to lay the groundwork for a far more open solution. I'd like to encourage you to consider pursuing the following 14 goals for a second-generation nonprofit marketplace.
1. Free access
Anyone should be able to view any page of the site without paying a fee.
2. Instant access
Anyone should be able to view any page of the site without registering. Imagine how much less widely used Wikipedia would be if you needed to login just to read an article. GuideStar's policy of limiting page views only to registered users seems like a bad idea. If I'm looking at a page on the site, I should be able to e-mail you a link to that page, knowing that when you click on the link you will see the same page that I'm seeing, not a blank login screen that requires you to register before you can see what I sent you.
3. Open-content licensing
All of the content should be licensed under mainstream open-content licenses, such as the Creative Commons license used by MIT's OpenCourseWare project. Existing sites like Charity Navigator, DonorEdge, and GuideStar limit the value of the data by imposing copyright restrictions that forbid the creation of derivative works (restrictions, restrictions, restrictions). Derivative works should be encouraged, not feared. The licenses should be chosen such that the site qualifies as an “Open Access Publication.”
4. Open-source
Any new software created in the course of building the site should be developed as open-source software. And, just as importantly, that new software should be built on top of existing open-source infrastructure rather than built on top of proprietary tools. For example, the site should run on something like the LAMP tools, rather than something like Microsoft's .NET framework.
5. Broad scope
Like GuideStar and Charity Navigator, the site should strive to serve as a central hub that provides information about thousands of nonprofit organizations, rather than just a limited selection of organizations that operate in some specific domain or specific metropolitan area. Local community foundations can provide separate portal sites that serve as windows onto a subset of the central listings, without maintaining their own duplicate databases.
6. Collaborative contribution
Adding content to the site should be a collaborative effort. No single organization should have an exclusive right to post content to the site. Individual nonprofits should be encouraged to post information about their own organizations. It should be possible for community foundations to add hundreds of entries about organizations in a particular area. Individual donors should be able to add information about organizations that interest them. Think Wikipedia.
7. “Post, Review, Revise” instead of “Submit, Review, Publish”
Following the Wikipedia model, new contributions should appear on the site immediately, prior to any peer review. Mistakes can be corrected later.
8. Structured schema
The content on the site should be structured based on a standard schema. The schema should include attributes for basic information about each nonprofit (name, address, phone, EIN, net assets, revenue, expenses, etc.). In this respect the site should be like GuideStar and Charity Navigator, rather than like Wikipedia (where most of the content on each page is unstructured text). The schema should include not just the structure for representing the nonprofit organizations themselves, but also structures for representing the projects and programs of the organizations, and for representing the financial information from previous years.
9. Collaborative schema definition
It should be easy for users to add to the schema. For example, the Greater Kansas City Community Foundation might, for the organizations in their area, track a wide variety of data, like number of volunteers or name of auditor. Any user should be able to easily add attributes like these to the basic schema, in just a few seconds, simply by editing a web page, without special knowledge about databases and without waiting through an official review process.
10. Neutral broker
The mission of the site should be to create a marketplace for information about nonprofit organizations. The organization that runs the site should refrain from any attempt to assign ratings, grades, or rankings to the nonprofits listed on the site. In this respect the site should follow the lead of GuideStar, not Charity Navigator.
11. User-defined ratings
Anyone using the site should be able to create simple rating schemes based on their own goals. For example, a donor seeking a small organization with a fairly recent incorporation date should be able to create a simple filter to find organizations that match those criteria, and should be able to see how a given selection of organizations rate based on those criteria. Users should be able to save those criteria to return to later, and they should be able to share those ratings with others.
12. Bulk export download
It should be possible to download the entire data set as a single file, for offline use. Similarly, it should be possible to download a file containing the data for any given selection of organizations, with the data in standard formats like XML, CSV, or ODS. Charity Navigator has a feature like this, where they offer the content “in an electronic format of your choice.” GuideStar also offers bulk data via their Custom Data Services, although only under restrictive licensing terms.
Having a bulk export feature with an open license makes it possible to “fork” the site, so that a new clone of the site can be set up overnight. Forking is important both for facilitating innovation and for establishing a data commons over which no single organization has monopoly control. The bulk export feature is also important because it facilitates large-scale data mining operations, such as integration with outside databases, or statistical analysis of the nonprofit data. The recently-published book Super Crunchers, by Ian Ayres, does a nice job of showing why statistical analysis is a powerful tool for both public sector decision making and nonprofit sector decision making.
13. Web service access
Live data from the site should be available in a form that other sites can easily consume, to facilitate “mashups.” (Ideally the data should be available in JSON format through a REST API.) A mashup is a web site that combines data or services from two (or more) other web sites to create a third site that offers features not found in the original sites. For example, the Oakland Crimespotting site is a mashup built by taking data from the Oakland Police Department's CrimeWatch site and mapping it using Microsoft Virtual Earth. HousingMaps is a mashup built by taking data from craigslist and mapping it using Google Maps.
14. Collaborative development
The design and development of the new site should be an open collaboration. Discussion should take place in open forums with publicly archived logs. At a minimum, the project should have a mailing list, wiki, IRC channel, and source code version control system, all of which should be open to public oversight.
Getting there from here
Hopefully this set of possible goals will look roughly similar to what you've already been thinking about, although the details may differ significantly. All of these goals are focused on (a) trying to make as much information as possible be conveniently available to as many people as possible, and on (b) creating a level playing field that will for allow for competition and rapid innovation in the tools and services used for accessing the data, as well as for the evolution of the data schema.
Half of these goals could be achieved quite quickly, perhaps in 2008, if there was leadership support for them. But, unfortunately, other goals will be harder to achieve. In particular, the goals of collaborative editing and collaborative schema definition seem to require an underlying software infrastructure for which there is not yet a good open-source solution. Fortunately, the new generation of Web 2.0 software is moving toward making these sorts of collaborative database tools. Proprietary applications like Freebase, JotSpot, and Dabble DB serve as proof-of-concept demonstrations.
