Mirlo Federation

The folks at Mirlo, a music selling website like BandCamp, put out this press release:

Good news everyone!

NLNet is awarding Mirlo €40,000 to help us fund our work on decentralization. We submitted the paperwork in April and heard back earlier this month. We're excited for what this work will allow us to do.

Before we dive into what we'll be working on, let's talk about what we mean by "decentralizing"! In the internet world, the norm is for a service to be centralized (for sake of brevity, this will be a little hand-wavy and simplified). This means that when you go to Facebook or Instagram, you're accessing servers controlled by Facebook, and everyone using Facebook is using those same servers. It also means that you put all your trust and faith in how Facebook and Instagram will use your data.

One alternative to all of this is to decentralize, and there are generally two ways to do this (and these two ways are generally more of a spectrum). The most radical way is that everyone controls their own data on their own device, and through the use of standards is able to communicate with others. The best example of this kind of system is e-mail. If you want to you could set up your own e-mail server and send your own e-mails. An other option is to use a more "federated" network, where several users join together, and with their joint server, "federate" with other servers based on agreed upon standards and principles. The most popular examples of this are probably Mastodon and Bsky (you could set up your own versions of those two as well, but it's not the norm). Often this kind of "joining forces" has usability benefits, mainly because it requires less technical knowledge (you don't have to be the person who knows how to install the software on a server).

At Mirlo, we've been aiming for federation for a while. Essentially what we mean here is that we aim to build software that other people can install on their own servers (or home computers!). While we'll continue hosting mirlo.space, this means that in the future there could be servers in, say, Montreal, Budapest, or Delhi, that hosts the same code, with musicians getting to choose where they upload their music. Because music purchasing is essentially decentralized (you download the album, you get the merch mailed to you, etc) a website like Mirlo is pretty easy to set up somewhere (it gets a little trickier if you want to prove to another website that you have paid for a piece of music, but we won't get into that here).

What does this mean for anyone currently using Mirlo? Your music will be safe and you'll be able to continue using Mirlo as a simple server provider.

It'll be a lot of work, but we have some ideas on how to proceed, which NLNet will help fund. Each of these could be a blog post on their own, but here's an overview of what we're going to be working with this money:

• Everything needed to self install Mirlo on your own server: theming, instructions, onboarding, etc
• Investigating integrating Taler and setting Mirlo up to work with several payment processors (we currently only support Stripe).
• Integrating a very simple version of ActivityPub (the protocol that underpins Mastodon) into Mirlo. We're being mindful that this will not be people's primary ActivityPub service, so we'll keep things simple for now.
• Exit from Mirlo for artists: being able to take all of your music / posts / followers and move to another service (maybe another Mirlo instance!). 
• Improving the accessibility of Mirlo across all sorts of metrics.
• Moderation tools: streamlining how we handle artists and users who violate our terms and our content policies.
• Punting decentralized streaming, with a mobile app that accesses several servers.

We think all of this is actually a fairly big lift, and we're excited that we'll be able to spread some of the labor for this out so that it doesn't just fall on our stewards. NLNet works by funding workers directly rather than companies, and preferences the work being done in Europe, which means we'll be tapping into our connections there, something we're very excited about to help us build our capacity and diversifying our community.

As for timeline? NLNet's funding works on a "you did the work, you get paid" model, but there's no pressure to do so on any given time line (except for within a year). We'll probably be approaching all these tasks in tandem, with some folks focusing on different parts of the project. We have estimates of how long they'll take (ranging from one week to 6 weeks per sub task), but as some of you may know, estimating software development is notoriously hard. We'll be announcing publicly when we accomplish a task, so you'll know when they're done.