The history of the source code of EPPI Reviewer as it appears today started in 2008 when the first lines of code for what will become EPPI Reviewer 4 were written. EPPI Reviewer 4 became available to the general public in 2010 and very much ever since, we wanted to release its source code. 14 years later, we finally did.

Why release the source code?

The main reason is simple: we want to maximise the usefulness of what we do. Allowing colleagues and researchers from anywhere in the world to look at how EPPI Reviewer works can be useful in a number of ways, many of which we (at the EPPI Centre) can't even imagine, so it has been in our list of wishes pretty much forever.

Relatively recently, another important reason has emerged too, summarised by a single word: "Sustainability". What would happen if the EPPI Centre were to close abruptly for whichever reason? (We don’t expect or plan for this, but we need to think from the users’ perspective!) EPPI Reviewer users everywhere would be stranded, without their data and without a way to recover it or continue working on it. Releasing the source code isn't the full solution to this scenario, but it's a crucial part thereof. It ensures people around the world can pick up from where we "left", in the unlikely scenario whereby we did "leave" (no, we have no reason to believe this scenario is likely!). This additional reason has emerged relatively recently and gradually, as our user-base keeps increasing steadily.
More in general, and perhaps more importantly, we do think that sustainability doesn't pair well with secrecy, one can choose to privilege one or the other, but not both. For a more general (and gentle) introduction to the relation between Open Source practices and sustainability, a good starting point is this article by Amanda Brock.

What took us so long?

The short answer is: we're a small team and want our work to be immediately useful. Doing what was needed to release the source code took weeks of work, during which we didn't make EPPI Reviewer better, so, for us to "find" those weeks, we needed our motivations to grow accordingly. Moreover, besides purely technical issues that needed to be overcome, we also needed to solve a number of other problems.

Competition/sustainability

When EPPI Reviewer 4 was launched, we had no idea whether it would be self-sustaining financially. Would it generate enough income to pay for our salaries and the hardware it ran on? We didn't know. As time passed, we found out it did, although UCL and research grants also contribute. Similarly, right now, we do not know if releasing our source code will mean the crucial source of income represented by user fees will be affected. In principle, once they have the source code, any third party could relatively easily setup an "EPPI Reviewer Plus" enterprise and become a direct competitor. Thus, we know a risk exists, and we need to account for it.

Small team

As mentioned above, merely finding the time to overcome the technical obstacles (which we won't discuss here) is a significant problem for a small team. Besides this, making a project open-source can mean opening it up to external contributors, and this generates more work as aspiring contributors will seek help in understanding the code-base and then their contributions in the form of changes to the source code would need to be tested and quality-assessed thoroughly.

The source code is in different places

This is perhaps the most obvious and easier to understand "technical obstacle": our source code lives in three different "places". Two separate "source code repositories" (one for the core of EPPI Reviewer, one for EPPI Mapper) and then also the Azure Machine Learning environment (and Azure DevOps), where the code running (our increasing range of) Machine Learning tasks lives and runs.
Overall, we needed to figure out how to solve or manage all these problems, and thus it took us 14 years to move from "wish" to "fact".

What did we do?

From 7 September 2024, the source code for the core of EPPI Reviewer is now available as a Public Repository on GitHub. This includes EPPI Reviewer 4, EPPI Reviewer 6, the Account Manager, EPPI Visualiser and more.

With reference to the problems listed above:
We are releasing the source code under a relatively new license called "Functional Source License, Version 1.1, MIT Future License" (FSL-1.1-MIT). This allows the license to cover the source code in one of two ways, depending on the purpose(s) for which it is used. If such purposes are commercial and include competing directly against EPPI Reviewer, then the source code is available (anyone can read it and perhaps obtain insights) but not open (using the code for such purposes is prohibited). Otherwise, the software is covered by the MIT license, which is extremely permissive and solidly fits in all mainstream definitions of "Open/Free Software". Moreover, once the code is 2 years old or older it becomes automatically covered by the same MIT license.

What this means in practice:

• If you want to use the open source for research, educational, not-for-profit purposes, you can/should consider the source code we're releasing as covered by the MIT license, and thus consider and treat it as Open Source.
• If you want to use the same code to create a commercial clone (or partial clone) of EPPI Reviewer, then no, the license that covers the code does not allow you to do so until the 2 year time period after release has elapsed.
• We may revise this licensing decision later on, as we learn more about the risks involved, and/or if we're lucky enough to find different long-term sources of funding.

When it comes to managing possible contributions from 3rd parties, we must admit that our core team is too small and our to-do list is too long, so as a general rule, we will not accept "pull requests" (the technical way by which any 3rd party can send us proposed code changes). Should anyone want to collaborate by contributing to the core codebase, they should get in touch with us instead, and we'll discuss each case individually.

For the time being, the released source code covers the vast majority of EPPI Reviewer – the core source code - meaning that EPPI Mapper and the code that runs in the Azure Machine Learning environment is not yet included. We will also release these at a later date once security issues are ironed out. (The machine learning code contains all kinds of ‘secrets’ to allow it access to storage and other Azure services.) We did not want to let the perfect get in the way of the useful, so we decided to go ahead and release what we can, and will release more as soon as we can. (Certainly in less than another 14 years!)

Concluding remarks

Releasing the source code of EPPI Reviewer feels like a major milestone to all of us and we are extremely excited that we finally managed to. Sure, this is does come with some important caveats and isn't all of the code that runs EPPI Reviewer, but on the other hand, it's the result of 16 years  (and more, as the origins of EPPI Reviewer actually trace back to the early nighties) of dedicated work of us here in the EPPI Reviewer team, so it does feel like a very important (if not final) step.

Should anyone wish to start trying to make sense of all this code, please do not hesitate to get in touch via EPPISupport - you will then be put in touch with the developers who will try their best to point you in the right direction(s).