[maiavictor]

That's a blatant lie, running a full node is not required to "be decentralized", for the same reason you don't need to run a full node in Bitcoin. Light nodes have the same guarantees about the integrity and irreversibility that full nodes do. Perhaps you should try learning about them - see https://github.com/ethereum/wiki/wiki/Light-client-protocol - before stating things you don't understand? The fact that the worst thing a "fudster" like you has to say against Ethereum is that it had a hard fork to revert a millionaire hack caused by a bug in early stages of the project; whereas something not too different also happened to Bitcoin - see https://en.bitcoin.it/wiki/Value_overflow_incident - when it had the same age; says it all.

I don't often comment here, but I simply can't understand why an human being would come to the internet and hatefully attack things it doesn't like with straight made-up facts. I mean, I know this is so common, I just don't get why. This kind of thing makes me sad about humanity as a whole. Why can't we be nice to each other?

[dang]

People think they're being nice when they're not. Your comment is also an example, which presumably is why it was flagged. Can you please be civil and respectful when commenting here, no matter how wrong another comment may be?

https://news.ycombinator.com/newsguidelines.html

[nadaviv]

> Light nodes have the same guarantees about the integrity and irreversibility that full nodes do.

This is not true. SPV nodes blindly follow the longest chain and are at the mercy of miners. Running a full node guarantees you that all the protocol rules are being followed to the letter, while an SPV node cannot verify chain validity rules (like the 21M coin limit) and could be fooled to accept payments with money made out of thin air.

> Ethereum is that it had a hard fork to revert a millionaire hack caused by a bug in early stages of the project; whereas something not too different also happened to Bitcoin

The Bitcoin developers fixed a bug in the Bitcoin protocol. The Ethereum developers bailed-out a buggy smart contract written by a third-party, where the bug had nothing to do with the Ethereum protocol itself. I don't think the two are comparable.

Something that would've been comparable is the Bitcoin developers doing a chain-rollback to save the funds lost by MtGox. Which of course would be a horrible idea.

Also, when that happened in 2010, Bitcoin was a pet project valued at $0.08, with a total market cap of ~$250k. Ethereum was nearly a two-billion dollars project when they bailed out the DAO!

[maiavictor]

> where the bug had nothing to do with the Ethereum protocol itself. I don't think the two are comparable.

You're defining "bug" to fit your purposes. It wasn't "just a faulty contract". The entire protocol had a reentrancy situation that wasn't intended by any of its developers nor expected by any of its users, and that went against the expected semantics of its official programming language; it was a protocol bug, for any sensible definition. I don't see anyone neutral arguing it wasn't.

If you argue The DAO hacker had the right for the Ether he got because "that's what the code said", you could also claim the Bitcoin address had the right to claim his billions BTC, because that's what the code said back then. He only followed protocol rules and got all his money taken away by the hard fork.

[nadaviv]

> see anyone neutral arguing it wasn't.

Well, FWIW, I've never heard _anyone_, neutral or not, claiming the DAO hack was a bug in the Ethereum protocol before now...

And that's because, well, it very clearly isn't. hackingdistributed has a good overview[0] of the coding bug in the DAO that enabled the hack which I recommended you to read.

Looking at this another way, this could've been avoided by the DAO developers if they developed the smart contract more carefully. And the exact kind of bug that lead to the hack was still possible on Ethereum following the bail-out hardfork. So how can one claim the hardfork fixed a protocol bug?

> If you argue The DAO hacker had the right for the Ether

I didn't say that. I only argued about the differences between fixing a protocol bug and bailing-out companies that build on top of the protocol.

[0] http://hackingdistributed.com/2016/06/18/analysis-of-the-dao...

[DennisP]

I agree it wasn't a bug in the protocol. However, it was an issue that people in general weren't aware of, and even the official tutorial code on ethereum.org had similar vulnerabilities. For that reason I think the fork was reasonable.

By comparison, Parity's wallet bugs were just a result of carelessness, and independent audits probably would have discovered them. Consequently, the community strongly pushed back when Parity tried to get a fork for the second one.

[maiavictor]

> So how can one claim the hardfork fixed a protocol bug?

If I understand it, it isn't fixable without affecting every existing contract, so, the best fix was to patch the problems it caused, patch Solidity, introduce safer opcodes that avoid it, and invest heavier on security, which is exactly what was done (just see how much the EF is spending on security-related grants...)

[RustyRussell]

> for the same reason you don't need to run a full node in Bitcoin.

Just to be clear: the Bitcoin developers do recommend you run a full node. It is the only way to independently validate the entire chain.

[CryptoPunk]

Yes that's true. However it's silly to choose every user being able to run a full node over every user being able to write to the blockchain. There's no point in being able to read the blockchain if you can't write to it.

Practically, you don't need to independently validate the blockchain. You can rely on dozens of blockchain explorers to do that. You do however need to be able to write your own transactions to the blockchain to have any degree of control over your own funds. The 1 MB limit effectively prioritizes the former over the latter, which is something that was never part of Bitcoin's original roadmap, or what Satoshi would have recommended:

https://bitcointalk.org/index.php?topic=532.msg6306#msg6306

[DINKDINK]

>it's silly to choose every user being able to run a full node over every user being able to write to the blockchain

There's no purpose in asking a user to run a full node to write to a blockchain. A user/client runs a node to verify they've been paid. If you're trusting someone else to tell you you've been paid, you have centralization. The entire purpose of bitcoin is to receive payments without a trusted, third-party intermediary. Only the people who can run nodes can tell they've been paid.

[CryptoPunk]

>>There's no purpose in asking a user to run a full node to write to a blockchain. A user/client runs a node to verify they've been paid.

This is a totally illogical statement.

You can't get paid if you can't write to the blockchain. How are you going to spend what you're paid if you can't afford to write to the blockchain?

>>If you're trusting someone else to tell you you've been paid, you have centralization.

I've already addressed this: the risks associated with relying on a dozen blockchain explorers to validate the blockchain are nothing compared to the risks of relying on someone else to hold the private key that controls your funds.

What they're all going to lie about what transactions are happening on the blockchain? It's open source, public domain data. There's no limit to the number of sources you can poll until you get a confident estimate of the true state of your transaction. Consequently, the feasibility of such a deception being pulled off is astronomically low.

In other words, you're advocating throttling write-access to the point that the Bitcoin-Core blockchain is not able to provide practical utility to the vast majority of the human population, to ward off a total non-risk.

[esotericn]

You can currently add a transaction to the Bitcoin blockchain for (converted) approximately one US cent.

Reference: https://bitcoin.electronrelocation.com

In the future we expect fees to increase (as they must for security purposes), but the idea that someone that can afford to run a full node will not be able to afford to insert a transaction does not hold up under basic scrutiny.

[CryptoPunk]

>>You can currently add a transaction to the Bitcoin blockchain for (converted) approximately one US cent.

That's irrelevant. The only reason you can get away with a low fee on the Bitcoin Core blockchain at the moment is that it's not operating at capacity. As soon as adoption picks up again, fees will start to climb - they have to because capacity is constrained. At any globally significant level of usage, fees will be astronomical.

Remember, as recently as December 2017, BTC's average transaction fee reached $30.

>> but the idea that someone that can afford to run a full node will not be able to afford to insert a transaction does not hold up under basic scrutiny.

That's absurd. 1 MB of space in each block allows for a maximum of 604,800 transactions being written to the blockchain a day. There is no way that fees for writing to the blockchain wouldn't skyrocket, far out of reach of ordinary users, at any significant level of adoption.