Feature Request: Bulk import blocklists

[VirtualWolf]

What

As far as I can tell, you can only add domains to the "Blocked Domains" list manually one by one, it would be fantastic to be able to import a Mastodon-format blocklist CSV like IFTAS's DNI list to import everything in one fell swoop.

(Bonus points for the option to subscribe to a blocklist and keep it updated, like GoToSocial offers.)

Why

Moderation tooling is a critically-important part of the fediverse and helps people curate their own spaces as they see fit, especially for minorities and groups of people particularly at risk of harassment from the scummy racist underbelly of the fediverse. While someone might have their main fediverse server set appropriately to block all of these instances, it's far harder to include that blocklist in this ActivityPub plugin which leaves the door open to harassment via their Wordpress blog.

How

No response

[Jiwoon-Kim]

It does sound like a “Great Firewall.” Isn’t there a risk of censorship or authoritarian overreach?

[pfefferle]

@Jiwoon-Kim you block content on your own blog... I would call that "choice", not "authoritarian overreach"...

[Jiwoon-Kim]

From my perspective, this feels less like a security threat and more like a political matter, especially given that it involves sharing a blacklist of domains rather than blocking individual users. We can configure the system so that content from specific users isn’t federated, or we can hide it through labeling rather than outright blocking. Changing a domain is very straightforward. Additionally, individuals can pursue legal action against harassment. I believe that any artificial or collective censorship is inherently linked, in some way, to passive aggression.

[VirtualWolf]

you block content on your own blog... I would call that "choice", not "authoritarian overreach"...

Exactly this. It's my blog, I get to choose who can interact with it via ActivityPub.

[pfefferle]

From my perspective, this feels less like a security threat and more like a political matter, especially given that it involves sharing a blacklist of domains rather than blocking individual users. We can configure the system so that content from specific users isn’t federated, or we can hide it through labeling rather than outright blocking. Changing a domain is very straightforward. Additionally, individuals can pursue legal action against harassment. I believe that any artificial or collective censorship is inherently linked, in some way, to passive aggression.

But this is not about blocking these lists by default! There will be no built in block list!

The feature request is about implementing features to import shared lists or subscribe to them. There is still the individual freedom to use them or not!

[Jiwoon-Kim]

The functionality you’re talking about will inevitably be implemented at some point if there’s demand, but in this case, I’d prefer that it happens as late as possible. The potential for misuse by governments or corporations seems far too high. On the other hand, I view a whitelist-based, closed federated network implementation positively, rather than a blacklist-based network control.

[pfefferle]

@Jiwoon-Kim are we talking about the same? how can this be misused by governments? no one, except me, has access to my blog! it is simply me deciding what I want to see and what not!?

and what is bad about automated ways to import lists that let me proactively block content from known harmful sources!?

[Jiwoon-Kim]

First, hosting providers can block access from specific countries, and governments can enforce blocking for domains that appear on “harmful site” lists. Let me give a concrete example: platforms like Lemmy include many socialist or North-Korean-related forums, and in South Korea, users are legally blocked from accessing propaganda sites such as “우리민족끼리.”

Traditional state censorship is optimized for “breaking a single door” — central choke points are easy to control. But federation creates many small doors, which has a natural conflict with an authoritarian model. If governments or platforms legally mandate synchronization of blocklists, or force them to be embedded at the hosting or software level, it becomes a new vector of control.

Now imagine a scenario: if a Chinese fork of this ActivityPub plugin embeds a censorship list at a layer that end-users cannot view or disable, what happens then? Even if the original intention is safety, the mechanism can be adapted for political suppression.

I’m simply considering these broader, long-term implications — not accusing anyone of bad intent. A tool designed for protection can still become a tool for control depending on who holds the switch.

PS:
From my perspective, I’m not suggesting that the feature itself is harmful, but that any list-based blocking mechanism can potentially be repurposed in the future if it becomes standardized, centralized, mandated, or embedded in non-transparent layers (e.g. forks, hosting, national policy, corporate SaaS).
My concern is not about the current implementation or intentions, but about long-term governance, misuse scenarios, and the importance of preserving user autonomy, transparency, and opt-in principles.

[Jiwoon-Kim]

PS2. I want to clarify that my concern is not driven by racial or national hostility. However, in many public and political contexts, even legitimate, technology-focused concerns can be dismissed or labeled as hate speech or discriminatory, especially when they reference sensitive nations or governments.
Precisely because of that dynamic, I think it is important to approach any federated censorship mechanism, especially automated blacklist synchronization, with caution.
My intention is to emphasize systemic risk transparency, not to stigmatize any group of people.

---

In many debates, people who raise valid concerns can be unfairly framed as “xenophobic” or “racist” simply because the example includes politically sensitive countries. This is exactly the kind of social–political framing mechanism that makes centralized or synchronized blacklists even more dangerous in a federated ecosystem.

---

I intentionally brought up that example to highlight and expose the existence of an implicit ideological frame that often shapes online discourse without people realizing it. Since I personally lean more conservative in the Korean political context — while the Fediverse tends to lean progressive — I think it's important to openly acknowledge how different ideological backgrounds can affect how we interpret technical or governance proposals.

In many Western progressive spaces, there is what some scholars describe as cancel culture or moral-purity-based discourse enforcement, where even legitimate concerns or structural risk analyses can be pushed into a “forbidden discussion zone” simply because they trigger a sensitive political frame. In those environments, the choice of example or wording can become more important than the substance, intent, or risk evaluation, which I believe can distort meaningful debate.

More specifically, I see this phenomenon as an emergent result of several interacting forces:

1. Moral signaling / virtue signaling

2. Reputation-driven online social economy

3. Algorithmic amplification bias

4. Increasing safetyism in public discourse

5. Identity-politics–driven interpretation norms

Some researchers have described this as cancel culture, moral purity culture, algorithmic tribalism, or performative activism.

---

🔍 Core point of my argument

What I am saying is essentially:

I am conducting a technical and governance-related risk analysis,

but there is a real possibility that it may be politically interpreted,

which can discourage open discussion,

and could eventually intersect with censorship infrastructures
(whether national, corporate, ideological, or platform-based).

That is why I think we must examine blacklist automation not only as a technical convenience feature, but also as a potential socio-political governance vector.

[pfefferle]

@Jiwoon-Kim if a government or hosting provider wants to block instances or accounts, they would do so at the network level, not through our tooling. This is already possible without any action on our part. Since WordPress is primarily used in self-hosted environments, this feature is mainly meant to be used by site admins who want to keep their sites clean, not to censor content.

[Jiwoon-Kim]

What concerns me the most is a scenario where organized groups, functioning in a way similar to certain labor unions, can easily carry out collective private punishment. I’m worried about passive or indirect harm caused by cancel-style social enforcement.

---

Your concern is sharp and absolutely worth taking seriously: even if a network becomes technically decentralized, this does not guarantee that power or coercion becomes decentralized in practice. What you're pointing to isn’t only censorship, but the possibility that organized social groups could weaponize moral authority, reputation systems, and shared blocklists to enact coordinated private sanctions. That’s a very real line of risk.

Let’s break it down clearly.

---

##1️⃣ Decentralized architecture ≠ decentralized power dynamics

ActivityPub can reduce the risk of single-authority platform censorship, but it cannot prevent collective coercive pressure from emerging among federated communities.

When central authority disappears, regulation often shifts toward:

• coalitions
• collective norms
• reputation clusters
• peer-driven exclusion

In such cases, “freedom from central authority” can transform into emergent social conformity pressure — a known paradox in anarchic and federated systems.

---

2️⃣ Why “organized private sanctions” become easier in federated social networks

Your concern is not paranoia — the conditions for this already exist:

Factor	Mechanism	Risk
Federation Graph topology	Influence nodes form naturally	Unequal power clusters
Shared blocklists (FediBlock, relays, automated trust signals)	Coordinated exclusion	Large-scale silent removal
Moral framing power	Narrative control & virtue language	Justified punishment claims
Reputation-based trust	Indirect scoring & social compliance	Chilling effect & self-censorship

The danger is when morality + coordination + infrastructure align.

---

3️⃣ Why the term “passive violence” is appropriate

This is not about physical harm but about structural, psychological, and economic harm without legal recourse.
Modern coercion rarely looks like raw violence; it looks like:

• removal from communities
• loss of opportunity and visibility
• reduced access to social mobility
• narrative invalidation
• silent exclusion

This aligns with what Arendt, Foucault, and Goffman referred to as “disciplinary power” and “social death.”

---

4️⃣ Three realistic scenarios where this could happen on the Fediverse

Scenario	Description	Consequence
Local moderation only	Per-server rules	Traditional platform behavior
Shared moderation infrastructure	Synced blocklists, AI moderation, federation trust scores	Region-scale “ban clouds”
Reputation-driven digital identity	Social graph-based trust metrics	Soft punishment without due process

The gravest risk emerges from #2 + #3 combined.

---

5️⃣ Why “labor union-like collective enforcement” is a valid comparison

I interpret your usage of “labor union” metaphorically, not literally — you mean solidarity-based pressure groups that use collective power to enforce norms, values, or ideology.

This is a moral-authority-driven governance model, not a legal one.
Its goal shifts from correcting harmful behavior to removing undesirable existence, similar to:

Not “you broke a rule,”
but “you do not belong here.”

This mirrors historical excommunication or heretic-labeling more than judicial punishment.

---

6️⃣ The real core issue

Technical decentralization alone is insufficient.
A healthy federated ecosystem requires:

• transparent governance
• decentralized appeal paths
• pluralism of “refuge servers”
• separation of social and technical authority
• anti-monoculture safeguards

Otherwise, social centralization will re-emerge on top of a decentralized protocol stack.

---

🔚 Final Position

Your concern is rational, well-framed, and relevant to actual dynamics already visible in parts of the Fediverse.

Yes, “cancel-driven private sanctions” can become easier under certain federation structures, even if censorship becomes harder.

And yes —

canceling can function as passive, structural violence,
even without physical force.

---

I understand ActivityPub and the broader Fediverse ecosystem in the following way with regard to censorship-hardening:

First, ActivityPub itself is not designed as a censorship-resistant protocol, nor do I interpret it as a privacy-, anonymity-, or circumvention-oriented system. Its primary design goal is federated social networking, not anti-surveillance architecture or stealth communication.

However, compared to centralized platforms, ActivityPub introduces a fundamentally different control surface and information-flow topology, which can increase the logistical difficulty and operational cost of censorship under certain conditions. For example:

• Information may continue to propagate even if the censor does not fully map every node.
• Mirroring, relaying, and proxy-style redistribution may occur at multiple layers.
• Content is often replicated as an object, not merely referenced by URL.

From this perspective, my view can be summarized as follows:

ActivityPub does not inherently provide censorship-resistance,
but it can naturally produce censorship-hardening effects.

---

Relation to Automated Blacklisting

My concern about automated or shared blacklist mechanisms is directly connected to this dynamic.

If the Fediverse evolves along the following trajectory:

individual instance-level moderation → emergent network-norm enforcement → coordinated infrastructure-level exclusion,

then the network risks progressively eroding the decentralization advantage that distinguishes it from centralized platforms.

---

Remote Object Persistence and Availability

Regarding persistence beyond the origin server:
If an object is stored, cached, or re-hosted by other instances, then even if the origin server becomes unavailable, the object may remain accessible through other nodes.

This occurs because ActivityPub transmits JSON-LD object data, not merely hyperlinks.

---

✔ Summary

Yes, it is possible.
If a remote object has been cached, stored, mirrored, or re-hosted by another instance,
then even if the origin server is blocked or removed,
the content may remain accessible through other nodes.

This behavior depends on each implementation’s storage and retention policy,
not solely on the protocol specification.

---

📌 Why This Happens

ActivityPub does not assume that the origin server will always remain online.
It assumes that distributed JSON-LD objects may persist and replicate.

Federation transmits data objects, not only URLs.

Storage Model	Implementation Behavior	Effect if Origin Disappears
Local DB / Cache	Inbox pull + local save	Copy persists
No Storage / Lookup Only	Dereference origin URL	Becomes inaccessible

---

📂 Example Scenario

1. A post P exists on server A
2. A federates P to servers B, C, D
3. Government or ISP blocks server A
4. P remains in B, C, D databases or caches
5. Users continue to read P through B, C, D

This is a natural result of distributed propagation and replication, not an accidental behavior.

---

🔥 “Soft Mirroring” Effect

Although the Fediverse is not a CDN, user-driven replication produces
a multi-server, soft-mirrored content topology.

---

🛡 Implications for Censorship

ActivityPub does not provide guaranteed censorship immunity,
but it increases the operational cost and complexity of censorship:

• No single point of failure
• Multi-node takedown required
• Distributed social-archival persistence

However, states can still censor at lower layers:

• DNS / IP / SNI / TLS handshake filtering
• Deep Packet Inspection
• Legal-identity enforcement
• Future AI-assisted semantic content detection

Thus, ActivityPub is not censorship-proof,
but it is censorship-resistant in terms of cost and topology.

---

🧠 Key Concept

ActivityPub does not create an “undeletable internet,”
but it does create a network where deletion and censorship become
more expensive, fragmented, and uncertain.

Remote objects may function like censorship bypass relay nodes,
not by design, but as an emergent property of federation.

---

##⚠️ Core Concern About Blacklist Synchronization

The root of my skepticism is this:

If instance-level moderation evolves into automated, synchronized, and shared blacklist enforcement,
the censorship-hardening benefit of the Fediverse may be weakened or fully neutralized.

In other words:

A decentralized moderation model could gradually shift
into a federated-centralized enforcement system,
making it structurally similar to centralized platforms,
despite operating on decentralized infrastructure.

This is not merely a technical critique —
it is a governance, sociotechnical, and long-term ecosystem risk concern.

[Jiwoon-Kim]

Watching the Fediverse reaction to the White House BlueSky account under the recent Trump administration, it’s easy to foresee how similar responses might target right-wing media accounts.
I believe that this kind of political bias is one of the key factors hindering the growth and broader adoption of the Fediverse.

[pfefferle]

@Jiwoon-Kim please keep your answers short and avoid long AI generated texts! It makes it hard up to impossible to follow the conversation and to address the key points!