Proof of Work Is Efficient | Hacker News Reader

Proof of Work Is Efficient

danheld.com

2 pointsundefined14 comments

“energy usage is subjective”. Hmm..

Yeah, came to comment on the same thing. I think that is supposed to refer to the part of the article that says "Claiming that one usage of energy is more or less wasteful than another is completely subjective since all users have paid market rate to utilize that electricity.", but phrasing that as "energy usage is subjective" is a terrible way to say that. The quantity "amount of energy used" is not subjective, and is also the plain interpretation of the phrase "energy usage is subjective". If one wants to say "the value of what the energy is used for is subjective", one should just say that.

In addition: 1) I don't think the comparison between smelting bauxite into aluminum and taking a hash of a value, is a useful one. The reason that taking a hash being a one-way function is an important property of it being a hash, is because of the relationship between the value and the hash of the value, and the fact that one can compare hashes of values to determine if it is (probably) a hash of the same value, or not, but can't determine the value from its hash. Nothing analogous to this is relevant in the case of smelting bauxite into aluminum. It's a stupid comparison.

2) The "type 1 civilization" section wont convince anyone who both thinks clearly and is initially concerned about energy use, because it is a general purpose argument for not worrying about energy use. The argument is just "being willing to pay for energy may result in better energy sources being developed, and therefore the limits of energy use being less of a concern." . But this can be applied to any situation where someone is concerned about energy use. It also doesn't address the concern that while some energy use may be worth paying for from someone's perspective, according to their subjective valuation of the results, it doesn't take into account the negative externalities of energy use, which is the thing people are concerned about. If people had to pay to revert all of these negative externalities, and still wanted to spend the money on the electricity, then that would address the complaint. But saying "don't worry about the large energy use, because it is a large energy use and therefore not a problem", isn't convincing.

3)Regarding measuring the energy use per transaction vs per block: Of course, it is true that the energy use doesn't depend substantially on the number of transactions in the block. Nevertheless, it isn't clear to me that energy cost per transaction isn't a relevant quantity? Actually, I don't see why to expect the "energy use per block" to be a particularly relevant quantity. Yes, one can describe and estimate the total energy used in attempts to mine a given block, using the difficulty of the block and the energy efficiency of the typical mining hardware in use when the block was mined, but if the target rate of blocks was e.g. half as many blocks per amount of time, but the average issuance rate was the same, and the maximum number of transactions per amount of time was the same, why wouldn't the hashrate be roughly the same, and therefore result in the energy used per amount of time being roughly the same, while the amount of energy used per block being around twice as much? I suppose if I want someone to secure a bunch of valuables for me, and also handle securely transferring those valuables to others at my instructions, I would expect some of the costs of this to be over the period of time that they need to be secured, and so in a rough analogy I can kind of see why the "total value secured" perspective may make some sense. But in that case, I still don't think the relative measure would be cost per block, but rather cost per amount of time. (this criticism I am less confident in, and is also substantially less strong of a criticism even assuming I'm right about it)

To be clear, I don't hate bitcoin. I just dislike bad arguments.

1) The bauxite argument is actually pretty good. Aluminum is one of the most energy intensive ores to refine. Comparing the energy use of Bitcoin to Bauxite refining clearly communicates the level of energy investment going on and negative externality going on. In Aluminum's case, we waste a ton of power, and create rather chemically nasty tailings for a small amount of a substance that we can use to build many things, and importantly, can be reused. In Bitcoin's case, we get the same nasty tailings (waste heat of billions upon billions of hashes) for one single block never to see anyother use ever again, plus the waste products of whatever generated that power. Go ahead and shove ten transactions in it. I could facilitate that with a single piece of ledger paper at far less energy investment.

2 Is spot on.

3 I don't think you give enough credit. One block of a block chain is essentially one page in a giant distributed ledger. Yes, you can make the argument that trying to do the distributed piece with non-blockchain methods is impractical, but guess what, that's how the rest of the world has been doing finance networks all along! It works just fine.

Blockchain/crypto is only a "big deal" because it allows one to trivial set up a payment network/transaction medium that is currently poorly integrated with most financial system's regulators or investigative agencies/controls.

Blockchain, and PoW in particular solves nothing except escaping national government's ability to enact monetary policy, and not even that is a permanant given.

re. 1 I mostly just meant that comparing it not being reversible to hashing being one-way, isn't informative about anything. It may be analogous in other ways, sure.

regarding 3, I still don't think the energy use rate would be substantially changed if the target block rate had been somewhat different, so I don't see a reason to connect the energy use to the individual blocks? It seems likely to me that given a fixed issuance schedule, fixed electricity prices, etc. , that probably the energy/block is a result of the energy/time combined with the block/time , rather than the energy/time being a result of energy/block and the block/time . In terms of explanations, I think the energy/time explains energy/block, not the other way around.

Reasoning: miners increase their power usage as long as the marginal expected income rate per watt is greater than the cost per watt, and the expected income rate is determined by the price that bitcoin sells at, times the expected coins mined / time. We could break the "coins mined / time" into (coins per block) * (expected number of blocks), but this is I think mainly a technicality, and doesn't particularly factor into any decision making. The power use would, theoretically, increase until no one would profit by increasing how much they spend on power for mining, and so until there is nowhere that the cost of the power of mining more is less than the average rate of income of mining. This ties the power use to the issuance rate (in coins/time,not coins/block), independent of the block rate.

Reply on news.ycombinator.com