Hacker News Reader

Comment by DennisP | Hacker News Reader

I've been kicking around the Ethereum community since the early days, and what people were calling "web3" back then is not what we have now. The original idea was for everything to be hosted on IPFS or Swarm[1]. The latter was to be an Ethereum-native protocol similar to IPFS; it turned out to be harder than expected and hasn't gotten traction so far but it's still an active project. One of its goals was to enable payments for keeping files available.

Still, decentralized naming and digital signatures go a long way towards turning p2p filesharing into something more. I think it'll all come together eventually.

[1] https://www.ethswarm.org/

View on HN

When you look for services that offer commercial IPFS, it turns out they're mostly reselling AWS with a markup. A big markup, like US $500/year for 400GB.

Wow. And that's in the world where you can get storage for 1.50$/TB/Month if you rent baremetal servers.

Where can you get that?

Hetzner. SX64 is 64tb of raw storage and 74 euro per month. If we reserve some space for data redundancy it is closer to 2 dollars per terabyte per month though.

Just what we need. As if AWS itself wasn't already expensive enough.

Exactly. It’s been weird seeing so many people who have lived through several technology curves failing to imagine where these current trends are pointing.

So, the next wave of DRM

What proportion of people who had the perspective/foresight to predict "several technology curves" (and potentially profit off of them) will waste their time fighting battles in the comments of hacker news threads about the potential next one?

If they did it well enough, they might have plenty of time to waste. Just because you made a few good investments doesn't mean you have to dedicate every waking hour to making more of them.

What are some decentralized naming initiatives?

Handshake is the one I’m most interested/invested in.

- https://handshake.org

Wild! It's self-evident that you can't solve addressability with a blockchain, you can't get anywhere near the requisite scale. What am I missing?

Oh, I dunno. From a quick google there are about half a billion registered domain names. Assume an average length of 20 bytes and that's just 10 gigabytes. An ETH address is also 20 bytes, so 20 GB total. Blockchains are already bigger than that.

There are about 15 million new registrations per year, which is only one every two seconds. That's also easily within today's capabilities.

Throw in rollups and data sharding, and names barely make an impact at all.

It's not an issue of the amount of data, it's an issue of latency, both to query and to update. A reasonable latency budget for reads is single-digit milliseconds. A reasonable latency budget for writes is single- or double-digit seconds. You just can't do that in a global permissionless distributed system.

Ethereum today does writes in double-digit seconds and reads in milliseconds.

Not measured from my laptop, it certainly doesn't. And those writes cost money.

Average block time is 15 seconds and if you're doing a 1559 transaction then you can make sure it goes in. Maybe with bad timing it takes two blocks. Due to the random variance in PoW, it can take longer than 15 seconds, but the proof-of-stake chain issues blocks every 12 seconds consistently and the network will migrate to it in about six months.

If you have a full node then reads are from your local disk. If not, there's network latency like anything else.

My regular domain names cost money too.

Of course I will not have a full node. How do I discover the IP addresses of the nodes which maintain the data? How do I keep that set of IPs up to date?

ENS has gotten a lot of traction on Ethereum .

https://ens.domains/

How do I discover the IP address of an Ethereum node to which I can submit ENS queries without a centralized service?

Reply on news.ycombinator.com