26 comments

[ 3.2 ms ] story [ 52.2 ms ] thread
(comment deleted)
From 2015. Seems like enthusiasm about IPFS has waned quite a bit, at the same time some places just support it without much fanfare?
Like who? For all that it's a great idea, I don't think I've seen it used for real applications (i.e. not just an IPFS demo).
Hundreds of thousands of artworks (for a given definition of the term) on ipfs via nft platforms like hicetnunc I believe, so that's one big use-case.
> Seems like enthusiasm about IPFS has waned quite a bit

The IPFS team seems to have shifted their focus to cryptocurrencies. It probably isn't a bad business decision, but it feels like the rest of the ecosystem isn't improving as fast as it could.

I don't know if IPFS enthusiasm has waned, I see a lot of excitement, but the real problem on the web is that browser support (outside Brave) is non-existent.

If users can't use it, there's not much point in trying to build for it. There are IPFS gateways that try to provide IPFS via HTTP, but they don't seem to be popular, I think there's some fundamental challenges there that make them the worst of both worlds.

An additional problem is the lack of language bindings/implementations. There is no fully-fledged implementation apart from the Go one, which makes getting many developers onboard a challenge.
The JS implementation is pretty feature-complete, though still classed as experimental/alpha.

https://js.ipfs.io/

Yeah, which is a good step forward, but it's definitely not usable yet in my experience. There is definitely a vibe of "why would anyone want to use anything other than Go" in the community. (I actually really like IPFS, I just wish that it fitted into the standard Unix Dec environment a bit better).
I've tried... so often... to use it for anything. It's fantastic as a concept, but for some reason would always translate into a pain when using it.

With things like docker-images it could have solved the hosting issue beautifully given they both use hash-addressing but there was always something like an unexplained incredibly slow bootstrap time, or problems with IPNS, or no real way to bind domains to hashes. It always felt like there wasn't a pratical "problem" they were trying to solve.

What would widespread adoption of IPFS mean for the business model of CDNs?
I think they would be happy to switch to a pinning service. Cloudfare runs a public ipfs gateway, after all!
It's great for CDN hosting, but honestly too good, which might indeed be tricky for their business models. If IPFS were huge, and you had a large popular static site, in theory you don't need a CDN at all, because IPFS is a CDN all by itself.

In practice, I think they'd end up moving towards:

* seed hosting - distributing content quickly & globally, so that the rest of the IPFS network can immediately access new content, and so that content (old unpopular pages) remains quickly available even if nobody else is looking at it

* caching/DDoS protection/etc for non-static content, like most APIs, which IPFS isn't going to be affecting any time soon.

This is all a way away of course, none of that matters until IPFS is super popular & widely used, if ever.

Slightly related note - what does everyone think of Urbit?
I have no idea what Urbit is, but there is a huge crowd of people on HN that will point out its run by a literal facist/nazi/devil/whatever whenever you mention it. I'd rather not have this topic mentioned here to avoid having these discussions again.
That does not answer my question, nor do I see how that's relevant.
Necessarily, projects which protect free speech will attract those with objectionable views because they're the audience. In the west, that may be white supremacists. In China, it may be people the west would be more sympathetic towards.

IMO the way to proceed there is not to throw the baby out with the bathwater but to fork (the community, the tech, whatever) and contribute to the shared vision of freedom while excluding any objectionable parts from the fork.

I dont care about that - i just dont want HN manchilds throw a tantrum.

This is irrespective of the actual deeds of Urbit and their users. This is a workaround for HN.

For anybody else with no idea what Urbit is, here’s the Wikipedia article. https://en.m.wikipedia.org/wiki/Urbit

IMO it sounds interesting, but the internet is littered with interesting decentralized protocols that never went anywhere. There’s a real question of “can they build a killer app”?

I watched their video but came out pretty confused as it seemed a bit too beyond my tech skills:

https://www.youtube.com/watch?v=M04AKTCDavc

That’s similar to my feeling too. I’m sure the technology is neat, but it seems very esoteric and hard to get started with.
It is just intentionally esoteric for the sake of being esoteric, not for any technical reason. It exists to make its creator feel smugly superior, not to solve any real problems for actual people.
IPFS attempts to solve one set of problems and is pretty good at it. There has certainly been more adoption as NFTs have used them as a storage layer. Also it's not entirely strange to see people start to use Github Actions to publish straight to IPFS as part of their webdev flow.

However, there are other good projects in a similar space. Hypercore protocol being one of them.

While IPFS has pushed ahead in field, I suspect they'll end up being the Netscape of decentralized tech. It's still early days.