[vbuterin]

When I talk to people in the world about what about blockchains they find exciting, the aspect of permanence really is one of the things that people find attractive. I agree it's use-case dependent, but the problem is that once you go down the use-case-dependence rabbit hole, developers and users have to really think "am I creating something permanent or not"? And once you have to think even a little bit, you lose 50%+ of the magic. Having properties that you just get by default that you don't even have to think to obtain is quite important.

The good news though is that you don't need permanent storage of history to be supported by the base protocol itself. Consensus nodes don't need to know about history to verify the current chain. There are plenty of other mechanisms for storing historical data: bittorrent, centralized archives, Filecoin-style networks, etc etc. You just need the blockchain's throughput to be not _too_ high, so that it's actually possible for these protocols to store what comes out.

Definitely nothing is permanent, but you can get pretty close to "permanent unless civilization collapses", and that's much better than "data could drop at any time if a few people just forget"!

[contingencies]

In anthropology I believe it is accepted truth that when you ask people questions regarding motivations or rationalisations for their behavior, particularly group behavior, answers given do not necessarily equate to observed reality, ie. "Oh yeah I'm fully in to decentralisation and taking down the establishment." -> "I want a lambo."

[stickfigure]

> Definitely nothing is permanent, but you can get pretty close to "permanent unless civilization collapses"

Let's be honest here and acknowledge that we're talking about "permanent unless your particular fork of one particular blockchain collapses". There's a vast chasm between that and all of civilization.

[dannyw]

I disagree. If bitcoin or ethereum gets completely killed today, I'm confident you'll still be able to download the chain in four decades (assuming civilisation still exists) out of pure history and data hoarding.

[meowface]

Right. The "value" / perceived value won't be permanent (and may be very subjective and disputed from the start), but in terms of long-term data integrity and availability, blockchains can be useful. An NFT ownership transfer that occurred in 2020 on some blockchain may not be considered too valuable in a hypothetical 2040 world where almost no one is using that (or perhaps any) blockchain anymore, but you can likely at least retrieve that record and be pretty sure the data is accurate and wasn't tampered with.

So a (at-one-point reasonably popular) blockchain or protocol forking or falling out of use won't actually result in that data being lost to time. You'd probably need a major worldwide catastrophe for there to be a significant risk of that.

[littlestymaar]

> but you can likely at least retrieve that record and be pretty sure the data is accurate and wasn't tampered with.

You will have no way to tell if this data came from the legitimate Ethereum blockchain that was in use in 2021, a forked chain, or even a completely fake one which has zero blocks in common with the real one.

The authenticity guarantee in blockchains doesn't come from cryptographic schemes, it comes from the network agreeing on some shared truth. If the network is no more, you have no way to tell the truth.

[meowface]

If the network is totally gone, it's indeed more tricky. But if it still exists, and if there's no indication that the main network and chain was ever disrupted between then and now, you can be pretty confident the data is accurate if you join the network.

If no one uses it anymore or if it's so little-used that it's schismed into tons of other chains over the years, I think you can still probably obtain the original data you're looking for, but it'll take more effort to verify its authenticity.

Especially if the data you're looking for occurred at a time when the network was healthy and intact (like 2020), all you need to do in 2040 is find a block number and corresponding root hash that existed in 2020. I think these'll likely be possible to find even if the network's dead, and you can compare them against several different sources to increase confidence that they're not fake. Then as you scour the internet and download different published blockchain copies, you can truncate it back to that block number and compare the hash.

It's possible your search will be futile and you'll be unable to find a trustworthy record of block numbers and root hashes or that you'll be unable to find a verified 2020 blockchain, but I think your odds will be pretty good if the internet hasn't collapsed. Either way, as Vitalik pointed out, the odds are way higher you'll be able to find that compared to some data you entered into some SaaS in 2020.

[littlestymaar]

> you can be pretty confident the data is accurate if you join the network.

And how do you know the network you're joining is actually the original ethereum network and a completely different blockchain?

Like every peer to peer network in existence, an Ethereum node needs to connect to a reliable first node (a “tracker” in the bittorrent protocol, idk what's the name in the ethereum world). If they are down, you're on your own to join the network, and you have little guarantee that the network you're joining is working on the original ethereum blockchain.

In fact, after the ethereum foundation is gone, what guarantee do you have that you are running an actual ethereum node and not something running a modified protocol?

[bottled_poe]

When the pillars of some derivative blockchain have collapsed, how will we be certain about which download is the untampered version?

[ric2b]

For Bitcoin the correct one is the one with the most accumulated proof of work that doesn't contain any invalid blocks.

For Ethereum 2.0 I believe it's something about asking a friend?

[shawnz]

I think PoS chains typically use the chain with the most coin-days staked.