Hacker News Reader

Comment by NathanKP | Hacker News Reader

I personally think that IPFS is the actual web3. All the noise about blockchain powered NFT's is obscuring the really interesting tech underneath: the distributed, versioned filesystem that can be used to actually host those corny images that people are buying and selling for obscene amounts.

The other thing that is interesting is ENS, which provides a distributed blockchain version of DNS so that you can give your IPFS hosted things a proper human readable name instead of just using long hashes.

IPFS and ENS together feels like it could be an actual web3. But that real tech often gets overlooked amid the buzz and speculation of people buying and selling NFT's and pumping and dumping shit coins to make huge profits. Some people are doing the actual tech work, but the rest are gambling in a flashy casino, making a lot of noise, and calling it web3.

View on HN

I agree, although I think that having a decentralized and modern payment layer is also a part of Web3.

People owning their own data, instead of companies like Facebook and Google, also seems promising, so I haven't written off the ownership layer yet.

NFTs I don't really care about and don't understand the hype.

Metaverse can also be a part of Web3 (Decentraland, The Sandbox, etc), but we'll see where all that goes along with Meta, Microsoft, and all the other companies talking about the metaverse.

Serious question: why would anyone use IPFS for storage compared to available options?

Everything I see points to it being ludicrously more expensive, not having good systems for redundancy and availability, etc.

It's not meaningful to talk about IPFS being expensive, it's like saying bittorrent is expensive. Both are just ways for people to seed a file, and other people to retrieve that file from its hash. There's only as much redundancy as there are people who want to seed it. IPFS is in wide usage because of libgen haha. Currently the resolver can take 30ish seconds, and there's a new, fancier resolver that supposedly takes <5 seconds, which is still pretty high compared to what people are used to but it's a work in progress. There's also IPNS which allows for the controller of a keypair to update the content associated with an identifier.

Filecoin is an incentive layer for IPFS that I'm much less bullish on. It's probably what you were thinking of, it's a way to pay money to have someone host or serve something for you on IPFS. It seems pretty underbaked to me but I think it's in the same order of magnitude, cost wise, as AWS.

What I really want is bittorrent, but with "live" file repositories, that can be updated by the maker.

Otherwise, I'm actually too well served by the regular cloud storage options to be honest.

That's IPNS

Peer to peer networking is still expensive because there aren't as many peers in it and it is not likely that data is cached at a peer close to you. This also reduces redundancy and availability.

There is nothing fundamentally stopping a peer to peer file system from having speed, redundancy, and availability, except that it doesn't have the economy of scale yet.

I think the fact that providers like Cloudflare are offering an IPFS gateway (in private beta) is an encouraging sign that we could soon reach the point where there are many more peers and providers caching IPFS at the edge, in distributed locations around the world: https://developers.cloudflare.com/distributed-web/ipfs-gatew...

ENS is only on Ethereum which that is not only expensive to purchase and renew the domain (due to gas fees) but with the way ENS is designed, you don't truly 'own' the domain as the keyholders are the ones that have access to the .eth TLD.

So basically it is more like a 'centralised' domain service with 'Ethereum blockchain' branding similar to GoDaddy, rather than 'a distributed blockchain version of DNS'.

The idea is interesting but the execution is not quite as advertised. Especially on a blockchain that is unable to scale.

Reply on news.ycombinator.com