4 ms·

«A bitcoin transaction wastes as much electricity as it takes to power an American home for a week» This is false, or at best a deceptive statement.¹ A transac

by mrb 9y ago

«A bitcoin transaction wastes as much electricity as it takes to power an American home for a week» This is false, or at best a deceptive statement.¹ A transaction does not consume energy. It's like saying each bit on a CD-ROM "use fuel" when it's shipped on a mail truck. Miners work on a hash of a block, so the number of transactions in the block is independent of energy use. For example the transaction rate could double tomorrow, without increasing energy consumption. [¹] Also, this metric is derived from Digiconomist, which is flawed as it overestimate consumption by a factor of 2: http://blog.zorinaq.com/serious-faults-in-beci/ http://blog.zorinaq.com/serious-faults-in-beci/

SilasX 9y ago
You're right, in a sense. However, mining is using a certain amount of energy per block, and each block has a limited amount of transactions. So it's reasonable to amortize the cost of the bitcoin network over the transactions and so, in another sense, each transaction maps to its share of the power consumed in a block. It would be equally fair (and it would be fair) to say that the shipping cost of a CD is equal to its mass- or volume share of the cost of the truck delivery, and indeed, that's how you'd correctly quantify e.g. the environmental cost of Netflix's DVD-by-mail (it doesn't work to just say "one more disc doesn't cost any more so the environmental cost is zero"): http://www.slate.com/articles/health_and_science/the_green_lantern/2008/08/video_stores_vs_online_rentals.html http://www.slate.com/articles/health_and_science/the_green_l...
- mrb 9y ago
  But the limit in the number of transactions is not a technical limit, it's a political limit. We chose arbitrarily to set the limit to 4 million segwit weight units, however it could be changed to anything without impacting energy consumption. Therefore your analogy is incorrect: if the cargo capacity of a truck is increased, it will certainly impact its fuel consumption.
  - SilasX 9y ago
    >But the limit in the number of transactions is not a technical limit, it's a political limit. We chose arbitrarily to set the limit to 4 million segwit weight units, however it could be changed to anything without impacting energy consumption. It's at least partly a technical limit, because there are technical issues faced by scaling the Bitcoin network to the arbitrarily large number of transactions/sec you think it can handle. Each scaling solution has its tradeoffs[1][3], and therefore, to the extent that there's no "free [scaling] lunch", then there is inherent scarcity to what the miners can support, and therefore transactions should be attributed their proportional share of that limit when accounting for them. It only appears as a political limit because a) users generally don't want a fragmented network, and b) there's persistent disagreement on what tradeoffs are acceptable when scaling. >Therefore your analogy is incorrect: if the cargo capacity of a truck is increased, it will certainly impact its fuel consumption. What is that responding to? The issue was how to account for the cost of one unit that takes up a scarce slot in a known-cost thing. The fair (IMHO) solution, per previous comment, is "blame" each unit for its share of the total cost. How does an increase in the truck's capacity refute the analogy? The same logic would hold: if trucks could carry more Netflix DVDs at the same [environmental or whatever] cost, it would be reasonable to say that the cost per DVD shipped also went down. If trucks had infinite capacity, then it would be fair to say that the DVDs have no cost, since the cost inputs are independent of problem size (number of DVDs to ship). But -- per [1] and above -- we don't live in that world. [1] For example, if you simply said "each block can hold as many transactions as miners are willing to tolerate" [2], that risks making the ledger (blockchain) freakishly large. Since each node has to store the full ledger, that blowup, in turn, risks [excessive] centralization, where only a few miner-datacenters can run nodes, defeating the (or a major) point of Bitcoin. [2] which you think would make transaction capacity non-scarce [3] Wiki page for discussing the issues: https://en.bitcoin.it/wiki/Scalability https://en.bitcoin.it/wiki/Scalability
    - woah 9y ago
      The limits are primarily political, as there are other blockchains doing thousands of times more transactions.
mikeash 9y ago
Wouldn’t doubling the transaction rate require forking the network?
- mrb 9y ago
  No. If segwit use jumped to 100% (currently stands at ~10%) then the transaction rate capacity would roughly double.
  - mikeash 9y ago
    Didn’t segwit result in a fork?
    - mrb 9y ago
      Yes (soft fork). But it already took place, so there is currently network capacity to double the tx rate without another fork (although it will likely take 1-2 years for Segwit use to approach 100%).
      - mikeash 9y ago
        Given how imprecise "an American home" is, quibbling over a factor of 2 seems pointless. So if the network went all-in on segwit, a single transaction could instead power a smaller American home for a week? Goodie.
imtringued 9y ago
>For example the transaction rate could double tomorrow, without increasing energy consumption. A higher transactions rate might result in more transaction fees being paid which indirectly increases energy use.

by mrb 9y ago

SilasX 9y ago

You're right, in a sense. However, mining is using a certain amount of energy per block, and each block has a limited amount of transactions. So it's reasonable to amortize the cost of the bitcoin network over the transactions and so, in another sense, each transaction maps to its share of the power consumed in a block. It would be equally fair (and it would be fair) to say that the shipping cost of a CD is equal to its mass- or volume share of the cost of the truck delivery, and indeed, that's how you'd correctly quantify e.g. the environmental cost of Netflix's DVD-by-mail (it doesn't work to just say "one more disc doesn't cost any more so the environmental cost is zero"): http://www.slate.com/articles/health_and_science/the_green_lantern/2008/08/video_stores_vs_online_rentals.html http://www.slate.com/articles/health_and_science/the_green_l...

mrb 9y ago
But the limit in the number of transactions is not a technical limit, it's a political limit. We chose arbitrarily to set the limit to 4 million segwit weight units, however it could be changed to anything without impacting energy consumption. Therefore your analogy is incorrect: if the cargo capacity of a truck is increased, it will certainly impact its fuel consumption.
- SilasX 9y ago
  >But the limit in the number of transactions is not a technical limit, it's a political limit. We chose arbitrarily to set the limit to 4 million segwit weight units, however it could be changed to anything without impacting energy consumption. It's at least partly a technical limit, because there are technical issues faced by scaling the Bitcoin network to the arbitrarily large number of transactions/sec you think it can handle. Each scaling solution has its tradeoffs[1][3], and therefore, to the extent that there's no "free [scaling] lunch", then there is inherent scarcity to what the miners can support, and therefore transactions should be attributed their proportional share of that limit when accounting for them. It only appears as a political limit because a) users generally don't want a fragmented network, and b) there's persistent disagreement on what tradeoffs are acceptable when scaling. >Therefore your analogy is incorrect: if the cargo capacity of a truck is increased, it will certainly impact its fuel consumption. What is that responding to? The issue was how to account for the cost of one unit that takes up a scarce slot in a known-cost thing. The fair (IMHO) solution, per previous comment, is "blame" each unit for its share of the total cost. How does an increase in the truck's capacity refute the analogy? The same logic would hold: if trucks could carry more Netflix DVDs at the same [environmental or whatever] cost, it would be reasonable to say that the cost per DVD shipped also went down. If trucks had infinite capacity, then it would be fair to say that the DVDs have no cost, since the cost inputs are independent of problem size (number of DVDs to ship). But -- per [1] and above -- we don't live in that world. [1] For example, if you simply said "each block can hold as many transactions as miners are willing to tolerate" [2], that risks making the ledger (blockchain) freakishly large. Since each node has to store the full ledger, that blowup, in turn, risks [excessive] centralization, where only a few miner-datacenters can run nodes, defeating the (or a major) point of Bitcoin. [2] which you think would make transaction capacity non-scarce [3] Wiki page for discussing the issues: https://en.bitcoin.it/wiki/Scalability https://en.bitcoin.it/wiki/Scalability
  - woah 9y ago
    The limits are primarily political, as there are other blockchains doing thousands of times more transactions.

mikeash 9y ago

Wouldn’t doubling the transaction rate require forking the network?

mrb 9y ago
No. If segwit use jumped to 100% (currently stands at ~10%) then the transaction rate capacity would roughly double.
- mikeash 9y ago
  Didn’t segwit result in a fork?
  - mrb 9y ago
    Yes (soft fork). But it already took place, so there is currently network capacity to double the tx rate without another fork (although it will likely take 1-2 years for Segwit use to approach 100%).
    - mikeash 9y ago
      Given how imprecise "an American home" is, quibbling over a factor of 2 seems pointless. So if the network went all-in on segwit, a single transaction could instead power a smaller American home for a week? Goodie.

imtringued 9y ago

>For example the transaction rate could double tomorrow, without increasing energy consumption. A higher transactions rate might result in more transaction fees being paid which indirectly increases energy use.