October 2023 update
It’s almost October, which means two things: 1) it’s time for another update, 2) it’s almost my birthday 🥳 ! September was quite the busy month, full of extensive work on papers and proposals, the FediForum unconference, applying to grants, and a growth in support & funding!
Funding Update
First things first, how are we going with recurring monthly income to support my work via my Ko-fi Memberships? Well, fantastic news: we’ve reached €620 per month in funding! The next goal is €1000 in monthly funding.
Crossing the €500 milestone wouldn’t have happened without all my fantastic supporters, and those who promote my Ko-fi to their followers. You all help keep my work independent, allowing me to entirely focus on moderation tooling for the Fediverse, without being distracted by for-profit freelancing.
This month we also saw a few large one-off donations, which were extremely helpful in making ends meet, allowing me to pay my rent & bills with less financial stress. Also, if you’re wondering why all my membership tiers are now asking for address details, it’s due to taxation and legal requirements on book-keeping here in Germany, that’s all that information is used for (I’d rather not collect that information, but unfortunately it seems I have to).
As a reminder, I’ve been focused on the Fediverse since March 2023, where I initially used all my savings (€16k) to support myself whilst I worked on getting up to speed and building connections to be able to effectively build good moderation and trust & safety tooling, and my funding target of €5000 allows me to break-even.
Nivenly Foundation Sponsorship
In other funding news, I’ve recently been funded by Nivenly Foundation to work on some moderation tooling. The initial product of that funding is two papers, one being a 6000 word technical proposal for a tool called FIRES, and the other was a 3000 word essay about moderation tooling, our current shortcomings, and how I believe we can do better in the future. So far I’d estimate that at least 100 hours of work has gone into researching & writing the two papers, and both will be published soon.
The FIRES proposal has been undergoing a private round of peer-review and will ultimately be an open-source project, which aims to enable subscribing to moderation advisories and recommendations, and to label those with structured data, giving moderators more information as to why a certain moderation action is recommended.
Nivenly Foundation isn’t Just funding the writing of papers and proposals however, as they’re also funding development of the first reference implementation. There will be an official announcement soon, along with the publication of the proposal.
Whilst Nivenly are providing funding a specific scope of work, the income I receive through Ko-fi Memberships also allows me to have more flexibility and to make contributions to both Mastodon & Pixelfed codebases, and to participate in the discussions & chats hosted by IFTAS.
Additional Income
In something of an unexpected piece of income, I was able to invoice for some freelance work where I provided technical advice and code review to a soon to be launched service for moderators & instance admins. This was unexpected because initially I’d thought it was open-source which would almost certainly not see me paid for my work, but it turned out to be closed-source (for now) project and the organisation behind the it wanted to own the Intellectual Property (IP), which required compensation to allow the transfer of IP to that organisation.
Whilst this does violate my desire to only work on not-for-profit and open-source software, as I’ve kinda had it with the for-profit side of the tech industry for now, it did also give me additional income that helps extend my runway length. In the future I’ll be explicitly clarifying the open-source and non-profit status before contributing to projects.
Now that we’re all caught up with funding matters, let’s get on to the work I’ve been doing!
Mastodon
First up, Mastodon, this month I worked on 3 different pull requests, and started work on an additional 2–3 branches. The pull request for install time prompts to setup a denylist (“blocklist”) will likely be closed, as the implementation only covered instances that have access to the mastodon CLI, not all users. There’s also problems with the overall architecture of that piece of the Mastodon codebase that makes handling updates to those denylists tricky.
In Mastodon 4.2.0, I shipped a bunch of changes to the streaming server, which whilst not moderation tooling related has been a long term project of mine and greatly improved the reliability of that component, allowing for more detailed monitoring, improved uptime, and opening the door for future performance improvements.
The other pull requests & branches focus on making the information relating to domain blocks more easily accessible, including making the audit log a key part of the instance & federation management interface and preventing unsafe URLs from appearing in the audit log.
Additionally, I’m working on changes to the API to allow for narrower scopes required for querying the impact statements for domain blocks, and changes to the OAuth 2.0 endpoints to enable feature detection & introspection, which will assist in building next generation moderation tooling.
Pixelfed
Over the past month, I’ve also been getting more involved with the Pixelfed project, assisting Dan in the background with a number of things. I’ve also been having a proper look at their moderation tools and figuring out what can be adjusted, what needs to be changed, and what other projects can learn from Pixelfed.
FIRES Proposal
Whilst the proposal isn’t yet ready for public release, it has gone through a extensive round of peer-review privately, with lots of great feedback. If you were at FediForum or read the notes, you may have already seen me talking about some of the core ideas in the proposal, although I may have referred to it as FFS – it’s just been renamed as part of the peer-review to make the intentions & functionality a bit clearer.
FIRES is an acronym for Fediverse Intelligence Recommendations & Replication Endpoint Server. Two fundamental aspects of the proposal are that it shifts perspective away from prescriptive denylists (“blocklists”) towards moderation advisories and recommendations that can change over time, and moves away from unstructured textual information explaining moderation decisions towards structured data that labels decisions.
Additionally, it aims to provide the infrastructure we need to be able to subscribe to moderation advisories and recommendations. It is very much so thinking two steps ahead of where we are currently, so may not be fully usable right out of the box, but in the coming months we should start to see the benefits of this new approach.
Our present infrastructure of denylists (née “blocklists”) will only get us so far: we need more structured data so we can make informed decisions, we need more options for moderating than just defederating, and the ability to subscribe to changes in efficient & resumable ways.
I’m currently working on the final edit of the FIRES proposal, tightening up the language & structure, and addressing all the feedback from those who participated in the private peer-review phase. The proposal should be published in the next week or two, pending final review & the official announcement.
Next Steps & October Plan
The majority of October is going to be focused on the development of the reference implementation of the FIRES proposal, along with additional changes to Mastodon to improve workflows & auditability of federation management, and continuing to improve the OAuth 2.0 endpoints.
I’ve submitted a total of three proposals for funding applications with NLNet, to secure additional funding for moderation tooling. The largest is for the entire FediMod tooling ecosystem, which includes FIRES, and tools for managing and accepting recommendations and proposals, as well as being able to collect receipts and preserve data for law enforcement.
The second proposal was for Pixelfed to improve to federation management and adding the ability to federate moderation notes on reports, allowing moderation teams to communicate with one another.
The third proposal was on behalf of Oliphant, who wants to work on more moderation tooling, especially tools that integrate with the FediMod/FIRES ecosystem. For this proposal I’ll be managing the project, providing technical advice, and performing code reviews.
CSAM & Germany
The final item that should gain resolution in October is how Fediverse servers hosted in Germany need to handle CSAM and other illegal content. I’m now in contact with Freiwillige Selbstkontrolle Multimedia-Diensteanbieter (FSM) for the protection of youths in online media, who will be taking my questions and ideas to the Bundeskriminalamt (BKA, the German federal police) as well as to a forum of other German organisations focused on youth protection. I’d initially contacted the BKA directly but as of writing I’d received no reply.
So, that wraps up the October 2023 Update, the summary being: onwards and upwards, towards a better, safer Fediverse! If you’ve not read the September 2023 update, then you can find it also on my medium profile.
Remember, if you’d like to support the work I’m doing, the best way is through my Ko-fi Memberships!
