Hacker News new | past | comments | ask | show | jobs | submit login

Isn't IPFS censorable? That's the impression I got from this FAQ entry: https://github.com/ipfs/faq/issues/47



Each node can choose to host (or not host) whatever it chooses. If that qualifies as censorship, you can't _not_ practice censorship.

There is no one who can say that something is not allowed on the network.

That's not to say you won't face legal or social consequences, IPFS isn't anonymous.


Compare and contrast with content-neutral networks, for example; we all like 'Net neutrality, right? But IPFS isn't content-neutral. Or content-oblivious networks; we all like Tor and I2P, right? But IPFS isn't content-oblivious.

IPFS has all of the tools required for censorious node owners to choose to block content, and the protocol doesn't have any underlying mitigations. It's not hard to imagine, especially in places where there's a monopoly of ISPs, like across large parts of the USA and the Middle East, that IPFS nodes might be easily isolated and vulnerable to relatively straightforward censorship.


Nothing obligates you to use a node which censors. At the same time, a node you own can't unintentionally (as with Freenet) end up with illegal or undesirable content.

A tradeoff has to be made. Yes, I censor when I don't pin someone's porn and serve it off my node. At the same time, I'm not preventing anyone else from accessing that information.

Another comparison would be to Bittorrent. Given x.torrent, my refusal to share its contents does, in a sense, make me a censor. But my refusal (unless I'm the sole possessor of its contents) doesn't prevent anyone else from sharing it. So the final contents remain uncensored (in the whole).

Hell, even freenet "censors" in that infrequently accessed content will eventually stop being replicated within the network.


You didn't say "IPFS", which made me realize that your post is largely repeating the same points in favor of censorship on Usenet.

This is only increasing my confidence that networks which are aware of their underlying contents are inherently unable to effectively counter censorship, because individual nodes can always be pressured to drop content, and the Pareto principle guarantees that this censorship will be effective as long as the pressure is put on the most popular nodes.

I hear and am sympathetic to your point of view. Google takes a similar stance with email, Gmail, and spammers; consider, "Yes, Google censors when we don't relay someone's spam and serve it off our servers. At the same time, we're not preventing any other mail relays from forwarding that mail." (I'm not speaking for Google, merely making a rhetorical argument.) This is widely considered a good thing.

I am merely disappointed that IPFS, which has a lot of backing and is growing in popularity, may become both the dominant content-addressable distributed object system, and also remain lacking in terms of anonymity and availability.

(Also, while I am no fan of Freenet, you are equivocating censorship with cache expiration. One is done by people and one is done by an content-oblivious algorithm. Tahoe-LAFS's garbage collection works in the same way and is also not censorship.)


In IPFS the user's computer connects to whatever node has the content, popular or not. This means that even if the the 100 most popular nodes in the world have enough pressure put on them that they refuse to host a file, I can still host the file on my laptop and everyone with an internet connection will be able to access it.


> censorship will be effective as long as the pressure is put on the most popular nodes

Which makes IPFS a good choice, because there are no "popular nodes", or am I missing something here?


Compare it with DNS with HTTP(s) where simple DNS blacklist and/or IP blacklist (DPI in some cases) is all you need.

In case of IPFS you can ship HDDs behind the firewall and distribute content internally via local internet, intranets, mesh networks or just by moving files around.


It's about as censorable as bittorrent with magnet links. Which is to say, if you are sharing something that someone with power finds really, really objectionable, they'll have no trouble bringing down the hammer. So yes, totally censorable.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: