Upcoming proposal and discussion on content moderation

@Ciaran would a middle ground not be to make it extremely difficult to censor by making the thresholds extraordinarily high?

There appears to be a false dichotomy drawn between off-chain and on-chain governance systems. It’s being posited that off-chain governance systems are immutable and unstoppable. But, of course, upgrades in btc and eth are possible through hard forks. So actually, if we want to be precise, Bitcoin and Eth are not immutable and unstoppable - it’s just that it’s extraordinarily difficult to push through changes as widespread coordination and convincing is involved in hard forks. The DAO hack is an example of that.

Would it not be possible to similarly make certain things such as censorship similarly difficult on-chain? For example, the US Constitution is very difficult to change but it is possible, and as a result it hasn’t changed much. Perhaps for those things one would need a massive overwhelming majority, following would be disallowed, quadratic voting was introduced, etc. etc.

Seems like an agreed NNS constitution could solve the issue. That’s effectively what btc and eth has. It’s called what it takes to get a hard fork done.


If something like that happened today, maybe, maybe not. (I’m in no position to speak for the Foundation.) But it would be entirely for technical + legal reasons, not because of any grand scheme to position the IC one way or the other.

What I don’t understand though is your insistence on “what would the Foundation do about X”. On the one hand you’re arguing that the IC’s governance model should actively and explicitly prevent moderation (although, as said I don’t see how that’s possible without making the protocol immutable). On the other you appear to be assuming that the Foundation will maintain control of the IC’s governance indefinitely. (If not, what does it matter what the Foundation thinks about anything?)

It almost looks like you don’t trust the claimed 99% of the community to fall on your side of the argument and need a higher authority to step in and enforce it instead.

1 Like

The majority actually rarely decides. This is especially the case when the majority interest is a collective action problem around a general goal (to provide a globally trusted platform) rather than a very specific one (to remove a piece of content you don’t like).

Yes this is in fact what I’m actually arguing for. I’m not under the illusion that Ethereum and Bitcoin are shiny rocks we discovered in the ground :slight_smile: They can be changed. In practice however, since they are so hard to change, the only consensus you’ll ever get is for protocol changes. Ledger amendments simply aren’t feasible. Why? Because as Vitalik laid out in his debates with Vlad Zamfir over this, if somebody uploads something bad to Ethereum, and it hard forks to remove it, it’ll be uploaded again minutes later. So it’s simply not feasible to police the platform for content you don’t like at protocol level. Policing must happen at layers above.

Edit: I’m not arguing for an agreed constitution. If you have one, it’ll devolve into bureaucracy precisely because of the situation I outlined above. And governments won’t stand by and have some unaccountable non-state body making calls on these things. There’s a snowball’s chance in hell of that.


I don’t believe providing a means to obfuscate traffic to allow for hosting of DMCA content should be on the agenda for ICP

… Being hosted in globally distributed data centers, if the country they are hosted in has a problem with the lack of transparency the whole server just gets yoinked.

Unless I’m missing something, if there’s content a state actor doesn’t like they in would navigate as a user of the service then see what internet addresses are serving the response and then go knocking at the data centers doors.

The key difference with the plausible deniability aspect is that The Internet Computer nodes host the content so baking in a way to shield the IP address would imo remove the plausible deniability which would in turn lead to nodes being pulled from locales? I think we should be focused on building up more deterministic decentralization that is best-achieved through complying with local laws where the nodes are hosted. This will also help us to build an image that will be trusted for DeFi and to compete directly with AWS where central control and censorship of ideas is starting to take root.

Badlands (hosted on consumer hardware and networks) would be the way to go for underground ideas… AI image / content recognition should probably still be implemented though and tweaked for categories that are voted to be restricted. I would say “disable access first, appeal/tweak later” is right for most circumstances regarding AI-based content moderation of the worst types of content.

1 Like

Absolutely God damn right. This censorship is absolutely ridiculous. This is supposed to be an unstoppable web 3.0 blockchain.


My introduction to the internet computer was that the project is starting off centralized with a lot of power in Dfinity’s hands. As I understand it, this trade-off was out of necessity to implement the scale of infrastructure and code that is required to see the project get going. The goal and trajectory is decentralization, but it is a process, not an event. So I am not discouraged that we have some centralized exposures at this point. I think we all want to get more decentralized, Dfinity included. It’s in all of our interest to get to that point.


I am curious too if the community has any appetite for making some of the code governance more difficult to change than others. Right now, anything can be changed by a majority vote (50% +1), and with a minimum participation threshold of 3%. Should some aspects require a higher threshold?

The NNS is here to stay (unless we vote it away? I don’t think we could do that though). But perhaps we say that certain proposal types (like removing a canister) require a supermajority. What would be high enough? 2/3 (with a nod to the IC’s byzantine fault tolerance)? or 3/4? Or 95%?


Thank you for this reminder. It was a guiding concept for me for the longest time and I just realized that I’ve lost sight of it recently. I definitely still believe it is true. I think Dfinity is doing a fantastic job and they are facilitating movement toward decentralization faster than I wanted or expected.


Hey everyone, we have a possible compromise solution that we think a lot of people might be happy with. Please head over here to explore the idea with us: Boundary Nodes as Censors

1 Like

This doesn’t solve the problem I’m afraid. Let’s say you have it at 95%. Someone uploads something abhorrent that 99.9% agrees should be censored. There’s a vote to take it down. It passes and is taken down. Then it’s uploaded 1000 times in 1000 different places again. Do you have another 1000 votes? Dfinity’s competitors know this rule. They’ll exploit it and spam the network with stuff they know people will want to censor. Even without these inevitable attacks, the amount of stuff that’d be referred to the NNS would be enormous. You can’t run an internet scale platform with direct shareholder voting on every single issue. It’s completely not feasible. The responsibility for monitoring the network would have to be devolved to some entity if you want censorship at protocol level.


No one is proposing to do this long term. It is the only option now, but discussion on how to actuallyscale this – from technical (e.g. blocking (re-)installation of canisters with some hash) to process (e.g. implement DMCA counter notifications without direct NNS involvement) – would be more useful than restating the obvious.

No one wants “censorship at protocol level” but some of us do think that staying within the law has its benefits. Even later on, in a utopic world where the protocol would protect all participants (node operators, neuron holders) from liability, you would still need mechanisms to stay within the law (e.g. at the very least move content that is infinging / objectionable to a jurisdiction where it is not so). Else you’ll just end up with zero nodes in the US or EU (due to copyright infringement), zero nodes in China (due to whatever the CCP wants), zero nodes in islamic countries (due to whatever cartoon) and so on.

Chanting “decentralized protocols are alegal” is not going to change reality, no matter how loud you scream,


I’m responding to people suggesting continued censorship but a really high bar as a long term rule. See who I responded to here and in the other recent thread with a suggested technical fix. Many are suggesting it!

Although on a second thought, it may actually be feasible. (Not great, by any stretch of the imagination, but feasible.) E.g. I’m pretty sure you could get 3% of neuron holders by voting power to spend 24 hours a day reviewing DMCA takedown notices if that meant doubling (or whatever) their governance rewards as compared to neuron holders not involved in it. At zero cost to the network (you’d be simply distributing the same rewards unevenly).

1 Like

And decentralised protocols not being subject to the law is a fact of life no matter how you may scream otherwise :slight_smile:

Individual nodes within the protocol are of course free to choose their own strategies.

I believe you may be confusing copyright infringement with censorship.


How am I doing that exactly?

No one is proposing censoring the protocol. Only the content.

I continue to be lost for words!

DMCAs are many times used to censor content, will the NNS have to analyze every claim individually to decide whether its valid or not? Cause that doesn’t seem doable.

How is that any different?

1 Like