This year, MoodleNet has gone from a Minimum Viable Product (MVP) to a more sophisticated beta, which is now ready to be tested as a federated platform. Along the way, hundreds of people have had the opportunity to test various versions of MoodleNet, to get involved in various ways (such as localisation), and to give their input into its future direction.
We’ve collaborated with a number of organisations and individuals outside Moodle HQ, most notably Eummena, a new Moodle Partner, who have added to the resource and energy available to the MoodleNet project. We’ve also had input from 3ipunt around plugin development, and many conversations with other Moodle Partners who are keen to get involved in 2020.
In last year’s retrospective, we explained that building a federated network like MoodleNet takes time. While ActivityPub-based alternatives to services such as Twitter (Mastodon and Pleroma), Instagram (Pixelfed), YouTube (PeerTube) exist, and more are being developed, MoodleNet is one of the first social networks to reimagine social networking from the ground-up for this particular audience. We want to empower educators by them creating and joining communities to engage in discussions with one another and curate resources.
Interestingly, Twitter seems to have realised the importance and power of federation, announcing a dedicated team to explore ways in which their platform can become decentralised. As we have often argued, the design of platforms and algorithms, as well as content moderation should happen for and with communities, instead of being controlled by a faceless corporation, potentially headquartered many thousands of miles away. So if Twitter is serious about this initiative we invite them to adopt the ActivityPub web standard and join the Fediverse!
This, then, is the story of MoodleNet in 2019. It’s been quite the adventure.
Q1: Testing the value proposition
We began the year in January with a core team of Doug Belshaw (Product Manager), Mayel de Borniol (Technical Architect), Ivan Minutillo (UX Designer & Front-end Developer), and Alex Castaño (Backend Developer). By the end of January we had begun testing the value proposition of MoodleNet with those who had signed up. Volunteer contributors helped to get MoodleNet translated into a number of languages, which we then tested with English and Spanish speakers.
Gaining feedback from testers, we evolved MoodleNet during the first testing period so that by early February, MoodleNet looked like this:
The question representing our value proposition was “Do educators want to join communities to curate collections of resources?”. After several surveys and lots of conversations, it appeared that, even from the initial version we presented to users during the first testing period, yes they did!
We iterated from version 0.3 in late February, to 0.5 in early March, and 0.7 by the end of that month. During that time, the second testing period demonstrated that users wanted a way of being able to distinguish the good from the bad in terms of resources, so we published an in-depth post exploring rating systems. In that post we shared the following diagram, commenting that ‘quality’ is subjective, and how we are interested in providing “the shortest path to the best resources”.
We learned a lot from testing MoodleNet’s value proposition, and most importantly confirmed we were on the right track. The tagline ‘Share. Curate. Discuss.’ was chosen based on what resonated the most among testers, and we ended up with a very clear list of priorities for future development.
Q2: Iterating and learning
In the second quarter of 2019, our focus was on ensuring that the version of MoodleNet we delivered met and, if possible, exceeded people’s expectations from the beginning. At this point, each MoodleNet installation was a standalone ‘silo’; in other words, it could not federate with other MoodleNet instances.
While we worked on federation on the backend, we made iterative improvements to the front-end in April and May with versions 0.9 alpha, v0.9.1 alpha, v0.9.2 alpha, and v0.9.3 alpha. At the same time, we thought carefully about how to make search in MoodleNet a useful and delightful experience, as well as the best approach to categorising resources and collections.
At the UK & Ireland MoodleMoot we ran a workshop where five emergent groups looked at MoodleNet from various perspectives. Our wiki page for the event captures the feedback we received from the ‘Online course design’, ‘Institutional use’, ‘Learning and teaching’, ‘Technology’, and ‘UX’ groups. Although this was very positive, we noticed some concerns around telling the difference between Communities and Collections, as well as the way discussion threads were organised.
— MoodleMoot UK & Ireland (@MoodleMootIEUK) April 17, 2019
In May we attended the Creative Commons Summit in Lisbon, Portugal. We did not run a session this time, instead networking with other projects, including OER World Map, Wikidata, and CC Search. In fact, our conversations with the OER World Map project directly benefited our approach to categorisation.
Towards the end of Q2 we said goodbye to Alex, and hello to Karen and James. We were delighted to have been able to attract and employ two such high-quality candidates, particularly to help with the challenges we faced around implementing federation.
We had a new page about MoodleNet arriving on moodle.com, but wanted to re-use the moodle.net domain for the main project website, so it was important that we set about sunsetting the existing service. In June, we worked with the Moodle Community team to formulate a plan which they then carried out perfectly over the following months.
We finished work on the first draft of MoodleNet’s Data Protection Impact Assessment (DPIA) and made it available for community consultation. This process, which we went through with Carlo Polizzi, Moodle’s Privacy Officer (as well as Simon McGarr, Moodle’s external Data Protection Officer and Legal Counsel), really helped us clarify the data workflows using a ‘privacy by design’ approach.
Q3: Rewriting all the things
While James and Karen got up-to-speed in June, Mayel prototyped a plugin for Moodle LMS, and Ivan took some time to redesign the front-end user experience. By the end of the month we had mock-ups ready of the new, much more conversational interface that makes the relationship between Communities and Collections much clearer. In July we did the additional work necessary to turn this into a clickable prototype, which we tested out with users.
It was around this time that we realised that we were going to have to change the database structure and rewrite much of the backend code in order for MoodleNet to be able federate while also providing a snappy UX on the front-end. This was not a small undertaking, but we were nevertheless optimistic about the amount of time this would take. Eummena, a new Moodle Partner, decided to support the MoodleNet team by hiring a full-time backend developer in the form of Antonis Kalou. He’s been working very closely with our team ever since.
Somewhat implausibly, looking back, we announced that the federation testing programme would start in August! Of course, making predictions around these kinds of things is hard.
Given the backend developers would be busy with the rewrite, we decided to turn the new conversational UI into code. The first version of this was ready by the end of August.
In preparation for the federation testing programme, and because of wider problems in the Fediverse around decentralised moderation, we created a draft user agreement and covenant for instance administrators. We also clarified our position around the licenses users would be able to select when uploading resources to MoodleNet.
In September, based on some feedback from the community, we reasserted the project’s stance on ‘free speech’ issues in the context of wider debates about and across social networks. We also revisited considerations around metadata and the way ‘likes’ and ‘boosts’ would work across federated instances.
Q4: Crunch time
By the time we got to October, whilst the backend refactor was in full swing, the MoodleNet servers were still running the new front-end with an old (unfederated) backend. That needed to change as soon as possible given the upcoming Global MoodleMoot, where we wanted to start federation testing and give people the ability to sign-up and start using an instance run by Moodle HQ.
At this point, Alessandro Giansanti joined us for a few weeks as a front-end developer to help us during crunch time. As a member of ZO (a cooperative he’s formed with existing team members Ivan and Mayel), he’s now also part of our team and we’ve really benefited from his energy and experience. Eummena also funded a front-end developer who has been helping Ivan and Alessandro.
MoodleNet sign-in page
As the Moot approached, we asked for some feedback from the community around resource uploading in MoodleNet, and we started to think beyond the beta for our 2020 roadmap.
At the start of November we were still planning to start federation testing in some form at the Global Moot. However, despite the team working late into the night for several days before demo day, we didn’t manage to have a fully working version ready by the time Doug and Mayel took to the stage, so we simply showed screenshots instead. The main challenge, as we explained in our reflections on what happened, was resolving the small but numerous breaking changes in GraphQL API to get the frontend working correctly with the shiny new backend.
After taking some time off to recover from an intense period of work, the team have regrouped in December and resolved the remaining broken functionality in MoodleNet. There’s many more features and improvements we want to work on, but the core MoodleNet functionality is now ready and hooked up with federation. We’ve even already had an initial external security review done, which showed no major problems. All in all – as we said in our recent ALT Online Winter Conference session – we’re actually in pretty good shape going into 2020.
Reflecting on mid 2019 in particular, it’s evident that we lost a bit of momentum for a while. Some of that couldn’t be helped because of the nature of the problems we had to solve, but we certainly could do a better job of estimating (or avoid trying to estimate altogether) how long it would take. We’ve fed these realisation into our thinking and planning for next year.
The team is particularly looking forward to users getting their hands on the new, federated, version of MoodleNet. We’re excited about the amount of power and flexibility we’re putting into the hands of educators, and can’t wait to see what individuals and organisations do with MoodleNet in 2020!