Comment by hwillis | Hacker News Reader

Comment by hwillis | Hacker News Reader

Parent Full thread

> I think once all the externalities are factored in, doing it all on a computer makes more sense even if POW is energy intensive.

No, it's not even close. Bitcoin currently uses as much electricity as 31 million US residents. One transaction uses more than two months of average residential usage.

> (including things like minting and handling physical cash and coin)

Why would you include that? Do you think people won't want to carry money? Do you think bitcoin makes cards, POS processors, and cash obsolete? Do physical wallets not take energy to make and run?

Regardless, the amount of energy used to make and use physical money is very small:

1. coins are irrelevant; the most expensive coins are cents, which cost about a cent to make. The total amount of coins made is much smaller than the amount of paper dollars made, so the value of energy used is small compared to the total cash made.

2. The US spends 1 billion annually to mint currency. Bitcoin spends 6.5 billion USD on electricity directly.

Transactions do not have an energy cost unto themselves, blocks do. If the entire world were to stop sending Bitcoin transactions for one hour, it would make almost no difference in the power consumption of the world's miners. Roughly 7 to 10 empty blocks would be produced. The mining process provides network security all on its own, transactions or no transactions.

>Why would you include that?

Because I said "all externalities". Bitcoin competes with the traditional financial system, including all the people and infrastructure involved in cash handling. That includes not just minting, but distributing it, collecting it, counting it, securing it, and so forth. And that is a cost borne not just by the United States federal government, but every government with physical currency and every bank on the planet. The idea that doing all this with physical objects is less power consuming than doing it with data is ridiculous.

If you want to make an honest apples to apples comparison, you include everything. Mining power usage is the single largest driver of bitcoin energy footprint granted, but it has a lot less other stuff attached.

You're telling us that since mining costs don't vary with the number of transactions processed, they don't count. Production costs that don't depend on the level of output are called fixed costs. [1] They're still costs. Pretending that a cost doesn't exist doesn't make it cease to exist.

[1] https://en.wikipedia.org/wiki/Fixed_cost

And I'm telling you that the production cost is tied up in blocks, not transactions. This is not up for debate, it is intrinsic to how bitcoin works. If you send a transaction, the only power consumption involved in that act is involved in the transmission of the data and its broadcast to the network.

The monstrous energy usage comes from trying to brute force a single hash. It is entirely decoupled from transaction volume.

> production cost is tied up in blocks, not transactions

Fixed costs are amortized. The block's footprint is spread across transactions. Arguing otherwise is like someone taking the power bill of a bank to infer the cost per teller-window transaction, and the manager arguing that teller-window transactions don't burn energy, branches do. Yes, sure. But also irrelevant.

And yes, if the branch tripled in size its energy use would reduce the per-teller energy footprint. (Assuming constant transaction volume.) But that's a hypothetical.

>You can still amortize the cost of the block across its transactions

Sure, if you're more interested in talking points then useful metrics. Increasing or decreasing the block size limit would lower or increase the "cost of a transaction" (since blocks are generated on a schedule commensurate with difficulty, which is roughly a function of how many miners there are) without actually changing the amount of power consumed.

What you're doing is basically correlating the world's average temperature versus the number of pirates and declaring that pirates are responsible for global warming. In reality, the two variables are unrelated and not even correlated.

Are you really arguing that the number of blocks and the number of transactions in BitCoin are two variables that are “unrelated and not even correlated“?

Of course they’re related. It’s trivial.

> the block size limit would lower or increase the "cost of a transaction"

Assuming constant transaction volume and other factors related to difficulty. You have a point. But it's far from a panacea for proof of work, and certainly not at the threshold to derail coming taxes and regulation. (Though one might find a way to structure the taxes such that they reward a productive increase in the block size.)

But what do you amortize the fixed cost over? Refugees can use bitcoin to hold money while fleeing dangerous situations. Not being forced to do a transaction is the value in such a case.

You can't send a transaction without mining a block. Therefore all the costs that are incurred mining a block are costs that are incurred sending transactions.

Changing the maximum block size would change the "energy used by a transaction" since block production is a factor of fixed time, not transaction volume, so the comparison is absurd on its face.

It's not a comparison. The fact is that the costs of processing bitcoin transactions are predominantly fixed. This tells us that bitcoin doesn't have economies of scale. The average cost of a bitcoin transaction doesn't decrease as the number of transactions increases. This makes bitcoin uncompetitive in relation to pretty much any other transaction processing technology.

And once that block size changes, the energy used per transaction will go down. It's really very simple, stop making it complicated. Blocks are there to hold transactions - that's their whole purpose. The fact that technically you can mine an empty block is an irrelevant detail of the protocol: bitcoin would not keep getting mined if there were no more transactions.

But what is the current block size, and what's the plan for changing it?

> The mining process provides network security all on its own, transactions or no transactions.

The mining process provides a fixed amount of network security. $125,000 per 15000 transactions. The fact that it would still cost that much money even with zero transactions is not a feature. It's not a good thing.

Each individual block is secured. You can't just add more and more transactions to each block without reducing security.

> Bitcoin competes with the traditional financial system, including all the people and infrastructure involved in cash handling.

Right now bitcoin competes with practically nothing. How much pizza is bought with bitcoin vs cash? You're effectively assuming that all those other functions are either irrelevant or that bitcoin can somehow do them for zero cost. Neither is true.

> And that is a cost borne not just by the United States federal government, but every government with physical currency and every bank on the planet.

60% of global federal reserves are in dollars. Dollars are the world's dominant currency. Compared to the massive disparity in energy use, it doesn't matter if even only 10% of global currency is in dollars. Dollars win.

Plus, if you're trying to compare with the places where cash really matters -where they can't use VISA or cell phones to transfer money- then bitcoin is certainly at a huge disadvantage after you have to buy computers for all those people.

> The idea that doing all this with physical objects is less power consuming than doing it with data is ridiculous.

Again, you're not replacing coins with data. You're replacing coins with USB sticks.

But you're right! It IS ridiculous, because it's completely insane how pathetically inefficient bitcoin is.

>The mining process provides a fixed amount of network security. $125,000 per 15000 transactions. The fact that it would still cost that much money even with zero transactions is not a feature. It's not a good thing.

Completely wrong, you cant know exactly how many transactions are processed per bitcoin block. The introduction of the lightning network and other 2nd layers mean any one of the ~1500txs per block could in reality be a batch of a 1,000,000 or more transactions being settled. There could easily be 15,000,000,000,000,000 txs per block, think about what that does to any $/tx calculation.

Bitcoin is cheaper than visa because it is designed to process infinite transactions for a fixed security cost.

Lightning network and other Layer 2 solutions are a joke. Those are only created because the Bitcoin network was intentionally designed to be slow and congested, to make it artificially expensive, and waste ridiculous amounts of electricity in the process. If those things actually worked then someone could make them work with anything else as a settlement layer, even traditional currencies. There is no actual reason for them to use Bitcoin.

So, the reason these systems use something like Bitcoin (or Ethereum, whatever) is because the goal is to build programmatic money, and you can build that on Ethereum and you simply can't directly build that on a "traditional currency" without first writing an adapter (such as some stable coin, as either a centralized regulated entity or as a decentralized protocol) to some decentralized programmatic system that you can then piggy-back on top of as a settlement layer.

The way that then, for example, (and this is really glossing) Lightning works is to provide a number of hypothetical signed Bitcoin transactions that rely on each other in a way where, if someone attempts to screw you over, you can put down one of the other transactions and move to a different state. You make it sound like Bitcoin is purposefully crippled to prevent this kind of performance enhancement, but one of the few and subtle big changes to Bitcoin was to support this!

https://en.wikipedia.org/wiki/SegWit

> The mining process provides network security all on its own, transactions or no transactions.

It seems to me that if nobody transacted, nobody would be able to sell Bitcoin, mining rewards would be rendered worthless, and mining would stop.

Sure, you can't point at any given transaction and blame any particular emissions on it, but surely it's reasonable to amortize it and assign a fraction of the emissions during each block to each transaction, considering that Bitcoin has no value without them.

No, because there is no relation between transaction volume and block production times, and this is a protocol level constraint. The network aims to produce a constant number of blocks per hour, which happens no matter how many transactions are available to be included in that block.

Here's an example: If, say, the maximum block size was doubled so they could hold twice as many transactions, this would cut your energy usage figure in half, but actual energy consumption by miners would barely change at all. Similarly, if the maximum block size was halved, your figure would double, but the energy consumed would not. If there were no such thing as a maximum block size, your value then scales linearly with the amount of transactions that can be crammed into 10-ish minutes of time.

I dunno, isn't that like saying "if planes were all half-empty, the carbon emissions of the plane would barely change, so you can't blame the carbon emissions on any of the passengers"? Yes, the "marginal carbon emissions" of a single transaction is ~0, much like the marginal carbon emission of taking an empty seat on an airplane is ~0, but by this logic nobody on the plane is responsible for any carbon emissions, which is obviously silly, so in lieu of any better accounting method, you just amortize the carbon emission of the plane across the passengers.

If the block size was reduced to, say, one transaction, then the economic value of the Bitcoin network would surely drop, mining rewards would be worth less, miners would quit, and the electricity use would fall. It's the economic value of the transactions that end up rewarding miners; I can't see how people transacting on the chain aren't partly responsible for the miners' emissions.

I can't think of any obviously better way to assign a number to that than just amortizing it across the number of transactions. It's a meaningful number that explains something about how much usefulness the network produces per unit of electricity. Doubling that by doubling the block size (or, halving it by halving the block size) would be a meaningful change!

>But by this logic nobody on the plane is responsible for any carbon emissions

That is strictly true. This is why the concept of a personal carbon footprint (something dreamed up by a British Petroleum marketing team) is inherently inane.

>Doubling that by doubling the block size (or, halving it by halving the block size) would be a meaningful change!

How and why? It would change your metric, but it wouldn't make the network consume one single watt less or more power. Its efficiency is unchanged.

A machine that spits out one widget per watt is less efficient than a machine that spits out 10 widgets per watt. Even if there is exactly one widget machine in the world, it is always turned on and always draws the same amount of energy, one type of machine would produce more widgets than the other. That machine is more efficient. It is more useful for the same energy input. It's a totally reasonable metric to compare widget-producing machines by!

If the bitcoin network is a transaction-producing machine, the more energy it takes to produce the same number of transactions, the less efficient it is. All else being equal, spending more money and energy for the same result is worse than spending less money and energy.

> If the entire world were to stop sending Bitcoin transactions for one hour, it would make almost no difference in the power consumption of the world's miners. Roughly 7 to 10 empty blocks would be produced

Yes, and then the average cost of a transaction would be even higher than the ridiculously high amount it already is.

Reply on news.ycombinator.com