Wikipedians and Wikimedia Foundation partner to experiment with microgrants
In early 2014 the Wikimedia Foundation began an experiment to better support individual contributors to Wikimedia projects, by giving out smaller grants to more individuals (complementing our existing grants to organizations, which mainly fund offline activities). We started by selecting a global south community that did not already have a local chapter meeting its needs: Arabic Wikipedia. We wanted to make grants that the community would find useful, so we asked them in a consultation, what kinds of small resources do you need? “Books!” was the primary answer we got, so we focused the pilot in that direction.
At this point, WMF staffers connected in the organizers of The Wikipedia Library, a community project (also WMF-funded) that helps editors access reliable sources. The Wikipedia Library already had experience delivering journal access to lots of editors on English Wikipedia, but they had not yet set up similar programs for other language communities, nor experimented with offering resources besides journals before. Their community-coordinator model appeared to offer a scalable way for distributing small resources to many editors, and they were looking for new ways to expand beyond serving the needs of English Wikipedians. Partnering on an Arabic pilot was a natural fit.
The next step was to find local partners in the Arabic community to lead the Arabic Wikipedia Library. We ran signups for local community coordinators to vet requests and purchase and track books, and selected two: User:Mohamed Ouda and User:عباد ديرانية.
To buy and globally ship books requested on Arabic Wikipedia, we needed pages where editors could ask for a book, payment options that volunteers could securely use to purchase books, and a way to track everything as it happened.
We made our first test purchases using Amazon.com and Neelwafurat.com (a popular Arabic bookseller). It was surprisingly difficult to get money to the local Arabic Coordinators for purchasing books in ways that were both user-friendly and easy to track. Providing them with prepaid cards, our first strategy, seemed like a good direction, but we weren’t able to find a card that WMF could purchase in the US for use by coordinators internationally. We ultimately employed a very old strategy – bank wire transfers – and worked with WMF’s finance team to add standardized processes for two other payment transfer options – Paypal and Western Union – to meet our needs for controls and flexibility.
Leveraging the existing journal access program run by The Wikipedia Library, we looked to a page design that could expand globally through a more modular set of pages. If The Wikipedia Library was going to serve many different communities, all with different needs, then its portal needed to be clear and distinct but its options needed to be adaptable and flexible. We translated the new kit into Arabic Wikipedia Library Pages: a portal page, book purchases, journal requests, and one for sharing sources between editors.
The kit pages used a customizable request template which let volunteers make requests and then interact with the local coordinators to facilitate on-wiki tracking of which books they wanted, when they received them, and how they used them.
Over the four months that the program was running, we purchased 14 books out of 19 that were requested. We shipped books to Spain, Saudi Arabia, Egypt, and Tunisia. Books like: Turks and Moroccans and Englishmen in the Age of Discovery, and July Revolution: Pros and cons after half a century. On the average, books cost $20 and shipping cost $10.
Our biggest challenge by far in purchasing books was shipping. It has been difficult to get booksellers (even regional ones) to ship books from the country where the book is stocked to many of the countries where Arabic Wikipedians have requested them. In the case of Amazon, postal codes are required for shipping, and it turns out that some editors in the MENA region do not have postal codes. We failed to have books shipped to Palestine, Jordan, Morocco, and in one case Egypt. In the first month of the pilot, this prevented about half of the requests from being successfully processed. Trouble with shipping a significant portion of requests made us hesitant to broadcast signups more widely. As a result, we fell short of our target of having 40 books successfully purchased, shipped, and used to improve or create new encyclopedia content during this pilot.
Another significant challenge was in reporting itself. It was hard to know if books were received, because despite volunteer-coordinator-pinging, only 2 books were ever marked as having gotten into the hands of the editor who they were shipped to. At this point, we still don’t have enough data to understand if the books had any impact on Wikipedia, as no editors came back to update their request with a short list or link of the articles that they improved or created.
What we learned
We originally set out to learn more about supporting the needs of individuals in the global south, test WMF grantmaking systems for making many small grants to individuals around the world, raise awareness of WMF grantmaking in communities outside English Wikipedia and Meta, and expand The Wikipedia Library beyond its English home base. Here are some of our findings:
1. Moving money to individuals globally is even harder than could reasonably be expected, and multiple options are needed to fit different users and countries.
For processes to scale easily, they need to be consistent. But the global financial reality is not particularly consistent. At the start of this pilot, we knew that trying to process lots of small money transfers to individual contributors would increase the burden on our finance department. We also knew that WMF’s standard method for sending money via bank wire transfers can take up to 2 weeks for an individual to receive, involves a lot of back and forth with individuals and banks to confirm details like SWIFT codes, and that bank transfer fees can eat up large portions of small grants. So we were hoping to find some new methods for sending a few hundred dollars at a time to our coordinators.
Over the course of this pilot, our finance team added standardized processes for sending money to individuals via both Western Union and Paypal, which we’d had only limited use of in the past. These are great options to add to our toolkit because they tend to move money to individuals in many countries more quickly than bank transfers. And we’ve also confirmed we still need a variety of other options, because individuals and countries come in all shapes and sizes. Paypal, for example, is the best option for many contributors to receive money in many countries, but Paypal doesn’t work in Egypt.
2. Moving physical things to individuals globally isn’t easy either.
It turns out that tangible objects aren’t easily transferred between countries either – unsurprisingly, we ran into regional infrastructure problems. Over the course of the pilot, we tried several bookselling websites, and we even considered having a book shipped to point A and then forwarded on to point B so that requests could be filled. Ultimately, though, shipping tangible items globally is a barrier to scale. For future experiments, it may be better to focus on transactions that can be entirely completed online.
3. Community volunteers and WMF staff have complementary strengths that make us great partners and can lead the way to scale if done right.
Community members know their communities! They understand the local processes (and policies), they speak the language, and they have built relationships with other editors. But, coordinating planning and timing can be a challenge, and it wasn’t always easy to know when to involve which members of the team, balanced with a desire to keep things moving forward as quickly as possible. Engaging all team members early and often is an area we can still improve on, to help everyone maintain a sense of shared ownership of the project.
4. Community-building and impact measurement takes time.
9 months into the project and 4 months into the active pilot, we still don’t know much about the ultimate effect we’ve had on contributors or on Wikipedia. We will need to follow up with measurement again in future months, and we may also need to come up with better ways to collect data to determine impact (see next learning).
5. Microgrant reporting may not be a feasible means for collecting data on impact.
Coordinators were more successful at handling requests than they were at getting recipients to report on how they used books, or even to confirm that they got them! Reporting is always a challenge for grants (or even surveys). In this case, we aimed for very small and lightweight reports (linking to an article that had been improved), but still lack this data. A requirement that editors coming back for a second book need to report back on their first book may gradually bring in this data, but it remains to be seen if that will be enough motivation in the long run to get people to respond, or if the program will lose steam before this happens.
6. It’s important to design for scalability, but easy to get caught up in over-designing it before it is needed.
We put a lot of initial effort into setting up book-purchasing accounts with controls for reconciling purchases. Some of that infrastructure ultimately went unused when we found issues with purchasing and shipping that were different than expected. On the other hand, we also put effort into building the kit for local satellite Wikipedia Library branches, which will be used well beyond the initial Arabic test case. Our development was better harnessed in that case, perhaps because there we understood the community needs we were designing for, and left it open-ended in cases where we didn’t yet understand the needs.
7. Having a well-defined target community to partner with is a clear benefit to your experiment.
We were not designing an experiment in a vacuum. Rather, we piloted via a program that had already demonstrated a working community model, connected to a new target community expressing a need to expand this model in new directions. This helped us better target our efforts and waste less time figuring out how to approach the pilot.
8. When the costs of your experiment start to outweigh the benefits, it’s time to wrap up and turn your ‘failure’ into learning.
Ultimately, we learned a lot from this experiment, and it has pushed our thinking, processes, and relationships forward in useful ways. At this point, we’ve learned enough about what doesn’t work to recognize that it is time to change direction. The tendency for all participants involved in a struggling pilot is to blame themselves and then try harder. But knowing when to stop trying to ‘make it work’ helps us conserve the most important resources we have: the time, energy, and morale of volunteers and staff — which deserve to be spent on future projects with brighter chances to succeed.
The Wikipedia Library remains on Arabic Wikipedia, but we’re taking focus off making book requests work. Editors can still request books for the time being, and if they’re easy to send we’ll still ship them, but the Arabic coordinators are resetting expectations to clarify that not all requests can be met, and we’re not going to waste more volunteer time on complicated workarounds or invest further in solving these issues. If/when there is sufficient data on successfully received book requests at some point in the future, we’ll still aim to analyze the impact of book grants on the encyclopedia, to continue learning from this project.
This report will now be used as a starting point to go back to the Arabic community again for further consultation. We leave it to the Arabic community to decide whether to continue the Wikipedia Library and attempt to focus on providing other types of resources, and/or move in some other direction for supporting Arabic editors.
The pilot participants:
- Siko Bouterse (Head of Individual Grants), Haitham Shammaa (Learning Strategist), Asaf Bartov (Global South Advisor), Janice Tud (Grants Administrator), Ocaasi (heading The Wikipedia Library), Patrick Earley (WMF Community Advocate), Mohamed Ouda (Arabic Library Coordinator), Abbad Diraneyya (Arabic Library Coordinator)