Zhoreeq/ipfd: An imageboard, but images are stored in a peer-to-peer network
github.com
github.com
Tangentially, in a perfect world, the Internet Archive is the IPFS pin of last resort, a payment mechanism allows for the Archive to collect micropayments for perpetual hosting (so the web isn’t going through the perpetual boom bust cycle of img hosts), they can dark content that gets DMCA reported, but they’ll nuke or reject content both illegal and/or exploitative.
I guess that gets interesting real quickly, since with a descriptive enough 'key' you could assemble almost any digital file from entirely arbitrary data as short as two bits.
If you "ipfs add" a file, ie take a local file and add it to ipfs, the default is to also pin it. This is most likely what you want.
If you just fetch or cat a file, it will still put it in the local store but it will not pin it. However you still could serve that content until it is GCed from your local store.
Pinning is the action of getting the content into your local store and preventing it from being GCed in the future.
All that said, your point is still valid. Viewing a file is enough to also make you serve that content, and it's very non obvious when you are and when it will be purged.
P2P no longer has platform protections, you can't claim that some illigal image you're storing isn't of your own action. This is different from TOR where you're just passing through unknown, encrypted non stored network traffic.
Your IP is exposed in any actual P2P network, meaning that if you happen to be the hoster of an illigal image you're toast from the start. All it took for the FBI to take down silk road was one single IRC connection unproxied. That's a dangerous game to play.
Centralization is a lot more about legal, safety, money, and convenience. Especially legal and safety.
Probably the ideal end state for services like this is to aggressively filter your own feed, not to censor the service itself.
This is a weird American meme that I wish would die off.
Therefore censorship and freedom are not a dichotomy, they are linked dependencies.
This is a nonsense authoritarian meme that I wish would die off.
I don't think you'll find too many people who think fraud should be protected speech, for instance.
That's as simple of a truth at the heart of the American constitution as any.
But then you'd have a new problem: what if someone posts an image of the Tianenmen Square protests? They'll get added to the "illegal" lists despite their value.
So maybe you'd also need a few community-maintained "not illegal" lists with some sort of voting system to decide whether to host something...it'd be messy, but it's not an unworkable problem.
No censorship by design. Not even for <insert evil here>.
What there are too little of, are venues for discussion of the _perfectly legal but not necessarily tasteful, pleasant or genial things_. We have way too many walled gardens which are bent upon imposing their worldview on their patrons. We need a lot less of that.
Time and again we are reminded that taking away plentiful online venues for the discussion of things wont just make such discussions disappear or people holding unwelcome views go away for good.
In short, we need perfectly legal and lawful 'cesspools' and fewer curated and coercive walled gardens.
IPFS is just a fetching tool, when you "upload" to an IPFS node your files are just sitting on your local disk until someone else requests them. It's not cloud storage.
Regarding this project: Several of the existing centralized imageboards are already breeding grounds for terrorists. The world does not need this.
I'm extremely impressed with the progress that IPFS has made - I'm an active (daily) user, and have implemented archiving systems with their tools. Clusters and pinset management in particular have improved substantially, browser support for ipfs:// natively has started to roll out - to name two things I've worked with recently. The libp2p project itself is another example.
Apart from beaker which is a special case, who else supports it natively?
The last Chrome release related to this was in September: https://blog.chromium.org/2020/09/chrome-86-improved-focus-h... where they safelisted a number of protocols in registerProtocolHandler().