#2536 Pol.social Server Problems - A Bitter Lesson for the Fediverse

Disclaimer: I appreciate and respect the work of volunteers and the FOSS idea. This article is not a personal attack, but an attempt to start a discussion about necessary professionalization. I believe that when hundreds of users and organizations depend on services, a conversation about quality guarantees (SLA) and raising standards becomes essential. And this text is about that need.

Update – September 10, 2025, 10:45 AM

POL.SOCIAL: BUS FACTOR = 1

The situation around the Foundation for Technologies for People (FTDL) servers is even more dramatic than one might have suspected. The Mastodon instance pol.social, after its alleged restoration, has been down again since yesterday – the outage has lasted over a day, and counting from Friday overall (and it’s Wednesday). Worse yet, key services such as Peertube, Matrix, or Nextcloud haven’t been available even for a moment since Friday, September 6th. So if any NGOs took advantage of the volunteer server offerings, they’re now patiently waiting for service restoration and access to their files.

Several hours ago, the administrator published a new statement acknowledging a cascade of problems and personal exhaustion. The statement was published through a channel that contradicts any principles of professionalism – on the very platform that had failed. Worst of all, the foundation has its own website ftdl.pl with information about planned work and disruptions, but apart from information about ORANGE’s fiber optic work and service access interruption, there’s currently no information. The donation page for infrastructure development is still running. This lack of communication speaks poorly of the type of incident. Contrary to hopes, the administrator didn’t announce professionalization and migration to a real Data Center, but continuation of the struggle with problems within the current, amateur infrastructure. It’s precisely this heroic but doomed struggle that has been going on since Friday.

A key question arises: if the data is allegedly 100% safe, why is service restoration so unimaginably difficult? The answer lies in the difference between data and system. Having intact files (data) is useless if the entire “factory” (system and configuration) that can process this data has been destroyed. “Manual restoration” in this case means manually rebuilding the entire server logic – all configuration files and network settings. If this is taking so long, it’s probably being done haphazardly, without automated deployment procedures. It’s like assembling a jet engine without instructions, having only a pile of parts.

A symbolic proof of the problem’s scale is that FTDL’s entire infrastructure is based on a single internet connection with 300 Mbps bandwidth – an unacceptable solution for stable hosting. The Signal Dadalo editorial team has a 3x stronger connection with a public IP and could theoretically create such a server farm, but this is such an amateur solution that infrastructure is kept in professional Data Centers.

This catastrophe brutally exposes the truth about the state of Polish Fediverse, which I summarized on the cursed-for-centralization BSKY:

[Reference to Bluesky post]

The current situation is not an accident, but a logical consequence of a culture where ideals and volunteer commitment have overshadowed the fundamental need for technical professionalism and stability.

End of update. The rest of the article was written on September 6th.

Another Outage, Old Problems

We are currently witnessing another painful outage of one of the most popular Polish servers in the decentralized Fediverse network. The pol.social server, running on the Mastodon engine, has stopped working again. Cutting off users from their accounts and network for over ten hours. This isn’t the first time. In July, users were cut off from their accounts for nearly three days, and the current interruption has already lasted almost a day with no end in sight.

The situation is all the more serious because many non-governmental organizations have placed their accounts on pol.social, including such important ones as the Watchdog Poland Citizens Network, which was strongly encouraged to create a decentralized account. These institutions, seeking an alternative to commercial platforms, trusted the project run by the Foundation for Technologies for People (FTDL) in the person of Piotr Sikora. Unfortunately, once again they are convinced that good intentions and noble ideas are not enough to guarantee the stability and reliability that are crucial for communication continuity.

Practical Illustration of Theoretical Risks

This outage is an excellent, practical illustration of the problems I described in detail in my critical analysis of the parliamentary interpellation regarding the federalization of social media, which despite my good intentions met with zero reflection and response from the community (read more about my criticism). I pointed out the naivety of proposals to implement the ActivityPub protocol in public administration without thorough risk analysis, costs, and maintenance problems.

Users as Hostages of Administrators

However, it’s worth pointing out other aspects of this outage. Mastodon users have repeatedly pointed to the problem of being slaves to admins. What does this mean? Users, including the mentioned organizations, become hostages to administrators’ competencies and capabilities (including time and financial resources). With practically zero possibility of easy data migration to another instance, their digital presence is completely dependent on one person or a small team. Where’s the talk of sovereignty here? Migration in Mastodon is theoretical, because it involves creating a redirect and importing contacts. All old posts remain on the old server. Let’s add that redirection works and post archives exist as long as the server and domain are still maintained with that declaration. Most often, shutting down the server and giving up running the instance causes complete deletion of data and redirects.

Ideology over Pragmatism: The Activism Trap

In the Fediverse itself, apart from frustrated users, no one seems to care about such incidents. The idea of “blind escape from corporations” is more important than building a really working, trustworthy alternative. The reputation of the entire ecosystem suffers, but critical voices are often silenced as an attack on the very idea. This is precisely the activism trap, where the ideal overshadows fundamental problems.

[Image caption: Statement by Piotr Sikora, pol.social administrator]

Lack of Professionalism and Stability Guarantees

The outage of a server run by the foundation shows that volunteer work cannot ensure service continuity at the level required by institutions. How can we seriously discuss transferring state crisis communication to this technology in this context, when even key social servers cannot operate stably? FTDL has one connection without a backup redundant fiber link, which violates all hosting principles. Additionally, in my opinion, they have excessively crammed too many services with too many features into limited resources. For example, nextcloud with some very system-intensive add-ons. Mastodon is also known for causing technical problems, which I will describe in a separate article.

Alarm Bell for Polish Fediverse

The pol.social case should be an alarm bell. This is proof that without professionalization, reliability standards, and real, working migration mechanisms, the Fediverse will remain a curiosity for enthusiasts, not a real alternative. Before we start dreaming about “digital sovereignty” in offices, let’s ensure that basic services work. Otherwise, every such initiative, like the one described in my analysis, will be doomed to failure from the start.

At the time of writing this article, i.e., at 8:30 PM, services are still not restored. We’re talking about mastodon, peertube, nextcloud, and others.