Problems on the Blockchain | Hacker News Reader

Problems on the Blockchain

benbyford.com

33 pointsbbyford30 comments

Ethereum will move to proof-of-stake, as described in a blog post yesterday [1].

I agree with the author that blockchain-based tech is more appropriate for small networks at present, yet the tech (e.g. Lightning network) will only get better and enable bigger and bigger networks.

BTW, what the hell is up with capitalizing 'blockchain' as the article does? It's like capitalizing 'database'.

[1] https://blog.ethereum.org/2015/12/28/understanding-serenity-...

The size off this endeavour is starting to consume more effort than is sustainable, not only for most peoples wallets, but also for the planet generally.

This is a common misconception, and only really is a "problem" in the initial distribution phase of the currency. Economically, all other things being equal, the cost of electricity spent on mining will approximately equal the mining reward.

The mining reward is approximately equal to (Fees + 50/(2^(floor(Years/4)))), where Years is the number of years since the network's inception. As time passes and the block reward drops off, the security of the network (equivalently, the amount of energy spent on mining) will approach the transaction fees paid for the use of the network.

The problem with this line of thinking is if someone subsidizes mining by a small fraction of fees paid they can become the largest miner. (A classic monopolistic tactic.)

So, really in the long term the network is protected by ~5% of fees paid over an extended period.

Worse, miners must be on the internet, so infecting a few machines could let you control the network with minimal investment. At which point double spending is probably more profitable than the minimal coins mined before the infection is detected.

Unless the hashrate of the sum of miners declines once they stop creating the cryptocurrency and start only validating transactions, the amount of energy spent on mining will always be high and will always be ecologically-unfriendly.

Sure, any energy expenditure over and above what would be necessary to run a centralized database can be considered inefficient, and therefore "ecologically unfriendly". However, the hashrate isn't burning energy for no utility: the security of the network is directly proportional to the energy expended in hashing.

Considering that the energy is being expended for a purpose (security against malicious actors), and that it is economically infeasible to expend more energy than the marginal fee revenue in a given moment (once we transition from the current block subsidies to a pure fee market), I question whether calling mining "economically unfriendly" is fair. The energy usage will scale linearly (not exponentially) with usage, which seems perfectly alright to me.

I completely agree with everything you said; however to clarify what I meant originally, I say "ecologically unfriendly" because of all the ways to do distributed consensus, proof-of-work is the least efficient mechanism when compared to the various proof-of-stake algorithms and Ripple/Stellar's cascading consensus model.

They all have their own individual tradeoffs (some technical, some political), but when talking about energy consumption, there's no argument that PoW is the most costly.

I was wondering if instead of having all the miners churning away at a proof-of-work there would be a simpler way to validate a block.

For instance, could you have a defined chain of authoritative servers, where the authority passes in a way where it cannot be abused?

Example: A mines a block by signing it with a key, throws up random number corresponding to B. B then does the same and nominates C. C stopped working, so authority passes to D.

Is there a way to do this so that the obvious attacks are thwarted?

I'm pretty sure the system you're describing is similar to Proof-of-Stake. In a PoS system, an owner in control of a portion of the monetary base (their "stake") controls the private keys to that portion. Depending on the algorithm used, the keys they hold have a chance of belonging to a set of keys that have the ability to sign the next processed block. Some, I'm not sure if all, PoS algorithms use an additional PoW step that acts as a time-multiplier on whether your key is in the signing set, to ensure a block can always be found in the absence of many people staking, and to create a barrier to the speed of block emission.

Additionally, R3CEV (pronounced like receive, the name isn't an acronym as far as I'm aware) is discussing a system where all banks in the consortium sign blocks in a round-robin fashion with trusted keys, negating the need for PoW at all. I have a feeling this will improve the efficiency of many things in the financial sector, but I doubt it will have as big of an impact as a decentralized and trustless system, like PoW and (possibly, if the wrinkles are ironed out) PoS.

>For instance, could you have a defined chain of authoritative servers, where the authority passes in a way where it cannot be abused?

Ripple and Stellar use a federated network of validating transactions to reach cascading consensus (validators A, B and C said this txn is good, and I "trust" them, so I will say this txn is good too) that while not completely abuse-proof, is significantly cheaper than PoW and possibly PoS (not certain about PoS costs; since the network of validators can be modeled as a graph, the central-most validators in the graph could censor transactions, but if validators "trust" a wide-enough variety of validators, it won't happen).

It's much easier in a centralized system. The only advantage bitcoin has over previous systems is that it prevents double-spending in a decentralized system.

if you have defined authoritative servers, they can just hold ballance amount and you can dispense with the rest of the blockchain entirely

I don't mean defined as in static. Of course if you did that you could just use an old fashioned database.

Did anyone bother to calculate the inefficiencies caused by a centralisation or federalisation and whether they justify all the effort that is going into bitcoin and other decentralization technologies?

Who is to say that a more transparent federalised system wouldn't be more or less on par with the bitcoin blockchain in terms of power centralisation?

Bitcoin is now effectively a federated system anyway. Miners coordinated to fix a hard-fork situation, which means they have the power to favor or disenfranchise certain people. They don't seem to have that motivation at the moment, but it's very possible. E.g. https://twitter.com/Datavetaren/status/631080568487342080

That's was exactly my point. Bitcoin is not as decentralized as it was envisioned to be so maybe there are other sorts of federalized systems that can be just as good and easier to implement.

Further down the twitter convo: https://twitter.com/TheBlueMatt/status/631715218788360192

The centralized systems are more efficient. Is that a typo in your comment?

Assuming that you mean calculating the inefficiencies caused by decentralization - that is easy - it is the whole mining industry. A centralized server would not need this proof of work effort to timestamp a ledger.

Let me know when you think that might happen.

Whos to say what fictional future system might be better then the system that is literally working right now.

> The higher the consumption the harder it is for the general public to participate in the spoils

I don't think the purpose of bitcoin is for the public to get "spoils", such as from mining. The public only needs the exchange utility of bitcoin, not to make money from it.

Agreed with most comments in this article except for the ecological comment. A good perspective where the author is just pointing out need for improvement in certain areas.

> Presumptions: at this point I'm presuming that the reader has a basic knowledge of Bitcoin and the concepts in the Blockchain technology.

At this point I presumed that the author had a basic knowledge of Bitcoin, but we all know what happens when you presume.

Love to hear any additions or pointers you might have. This article was mainly to point out the obvious to much of the cryptocurrency industry (but not necessarily to the average Joe) and as discussed in the comments there are efforts and alternatives. My main worry is that people can sometimes put blind faith in a technology, when they should really knock it down and build on its sturdy foundations

There's a bit of irony in my statement as I don't assume you know or don't know. We, as a population, depend heavily on trust for day-to-day life. As an author of an article like this you might consider that speaking as trusted authority on something which provides computed trust will affect the trust the general population has in said technology and in you (and the rest of us Bitcoiners). The general assumption is the value of Bitcoin is linked to how much people trust it. When you post things that negatively impact it, perhaps intentionally or perhaps altruistically, there's going to be trust issues between people who disagree with you.

The best way I can describe this effect with cryptocurrencies is "it's a polarizing topic". :)

Blockchain technology, no matter the implementation, will always have some degree of risk associated with the trust it embodies. Nothing will be made to be perfect. The question I'd like to know the answer to is whether or not it's providing enough value for the applications it currently runs, and what other applications can be improved by adding the blockchain's natural immutability to them.

Thanks for writing something interesting!

Proof of Stake does not work.

https://download.wpsoftware.net/bitcoin/pos.pdf

That's not a proof that no Proof of Stake system can work. It's a list of possible attacks, at least some of which are mitigated by recent Proof of Stake designs. See Vitalik's post here: https://www.reddit.com/r/ethereum/comments/3y0ysi/understand...

A fairly detailed description of that system is here: https://blog.ethereum.org/2015/12/28/understanding-serenity-...

> That's not a proof that no Proof of Stake system can work. It's a list of possible attacks, at least some of which are mitigated by recent Proof of Stake designs.

I have yet to see a design which mitigates all of the attacks.

Indeed I do not believe that such a design is even possible.

Would love to be proven wrong though.

Some people don't believe that bitcoin works or is possible even though it literally is working right now.

All of the currently "working" "proof-of-stake" systems have central trusted third parties signing blocks.

Until bitcoin was created all currencies used centralized trusted parties too. These things are in their infancy. Saying something can't be done at this stage is begging to be on the wrong side of history.

As I said.

>Indeed I do not believe that such a design is even possible.

>Would love to be proven wrong though.

Reply on news.ycombinator.com