How would the rejection of The Daily Stormer play out on IPFS?

You’ve probably heard about the controversy surrounding the many companies that have opted to block The Daily Stormer - an American neo-Nazi and white supremacist news and commentary website - from their services. As privately held companies this is entirely within their right to do, but because of the way the web is effectively controlled by a few massive companies, these actions are not without their implications. This is described quite well in the Cloudflare blog post where they announce their termination of DS:

What I want to discuss here, is how you think this might have played out if the internet as we know it was running on top of the IPFS backbone.

Please correct me if I’m wrong, but I seem to recall reading about having some agency over which domains I host/serve on IPFS. I think that makes sense: I should be able to maintain a blacklist of domains, like that of Digital Stormer, that I don’t want to share any bits with. If they wanna spread hate, they can do it with the resources of their followers, but not mine.

So the way I think IPFS works is that, assuming the majority of the web would not support them, their site may be slower than average, but not unavailable. And that’s about it.

Am I missing a lot here? Is there more to it?

1 Like

People wouldn’t even have to subscribe to a blacklist in order to not host any of their content. Because your IPFS node only requests what you tell it to, all someone would have to do to not help redistribute their site is to not visit it (so it never gets downloaded and cached on your node).

I think you’re right that the site would never really be down on IPFS. Anybody who visited the site before using their own node would still be able to see it. For those users, the access would be very fast for site content that’s already cached or pinned on their node.

I’d imagine that if someone tried DDoSing every node that has any of the site’s content, those users might see increased bandwidth utilization depending on how targeted the attack is, how many people have the content someone is trying to DDoS, and how many resources the attacker has.

1 Like

As for official blacklisting:

In Germany and other countries, where there’s no freedom of speech, but only the freedom to speak your opinion—there is a legal difference—, sites/pages/content can be blocked on legal grounds more easily, either by law or by a court ruling. This is then done on the ISP level. A site like this could already be banned under existent laws in Germany. State or federal executives can order that, no courts or parliament vote for new laws needed.

On the IPFS such a block wouldn’t be possible on the ISP level, except for gateway subpages, so in these cases a blacklist for IPFS/IPNS hashes would come into play, but that particular blacklist would only be enabled for nodes under German jurisdiction. (Based on a node’s IP address?) If Brazil for example thinks it’s fine, there would be no need for the IPFS blocklist maintainers to put Daily Stormer on the Brazilian blacklist.

And after all, as far I recall, such blacklists will be opt-in depending on individual node settings, so even a German user could then still access the Daily Stormer, if he wanted to.

As for private per-node blacklists:

Is such a system planned? That a user can maintain his own blacklist only for his node? I assume that such a system would be a good idea, already with regard to malware site blocking, adblocking etc.

It does look like such a system is planned (see some of the discussion here). I don’t remember where I saw it, but I remember coming across a screenshot where somebody got an error message on the ipfs.io gateway about the content they were requesting being blocked, suggesting that at least on the official gateway some type of a blocklist is already implemented.

There’s been a similar case study presented for Mastodon

http://www.ethanzuckerman.com/blog/2017/08/18/mastodon-is-big-in-japan-the-reason-why-is-uncomfortable/

Full report here, which includes a case study on IPFS:

2 Likes