4 ms·

Nope - the author is incorrectly using total energy / total transactions to get this number. Which, is understandable. However, if there were 2x the transaction

Nope - the author is incorrectly using total energy / total transactions to get this number. Which, is understandable. However, if there were 2x the transactions, the total energy would stay very close to the same. So no. There's a lot of energy to secure the network, but the actual energy usage per transaction itself is a small fraction of that misquoted number.

grey-area 5y ago
Unfortunately the network can’t handle more transactions. Decentralised, fast, cheap: pick two.
- hwillis 5y ago
  Bitcoin picked centralized, slow, and expensive. Most mining is done in giant warehouses, and the whole point is that more computing power results in zero faster processing. It remains to be seen if there are good ways to do distributed currency, but there are certainly less bad ways.
  - grey-area 5y ago
    Sure, they could have made better choices, but the decentralisation is really a fundamental problem that can't be designed away. If you want it decentralised it's going to be significantly slower and more expensive than centralised networks.
    - hwillis 5y ago
      More computationally intensive, yes- but computationally intensive may mean "running on smartphones in the background for .1% reduction in battery time" or "a medium-sized country". Increased cost is certainly not so straightforward. If you're sufficiently distributed, using already-existing hardware, with spare compute on extremely efficient devices, it could conceivably cost less. Even if a centralized server farm would be doing an order of magnitude less math, smaller processors use an order of magnitude less power to do that math. Bitcoin proper has a central ledger- every miner needs to hear about your transaction to verify it. Lightning is a clumsy way of reducing how many actors need to be notified of your transaction. Better currencies include stuff like that as first class. It's all in the name of getting closer to a constant-number verification scheme that is closer to competitive with the O(1) of registering a transaction with a bank. Centralized credit/debit cards exist so that a big wealthy firm can say yes, this person has enough money for this transaction and I will guarantee the transaction by paying for it even if they can't. What I would really like to see is distributed, automatic guarantees: when you buy something at a coffee shop, people running validators on wifi will pick it up and use their staked currency as insurance (hedged by the system) that they know accounts who trust this particular account, and the transaction is valid. More people staking and trusting this account means more trust by the larger system, which then only has to validate aggregated transactions. Anyone announcing themselves at a validator plugs in at a given level of aggregation, all of which have different staking/network/latency/storage requirements. Unlike off-chain transactions (eg lightning), the system is guaranteed at every level. Any given transaction will still be validated dozens or hundreds of times. En bloc it will probably use tens or hundreds of times more energy than the server farms powering VISA. I'll be honest, I'm okay with that. I really like the idea of having a bank account that isn't tied to a company.
- MomoXenosaga 5y ago
  My country handles 5 billion electronic transactions a year (thanks to COVID cash practically died last year). Fast and cheap although to be fair not decentralised. To this day I have no idea what cryptocurrency was supposed to solved.
idreyn 5y ago
But it's worth clarifying that the number is correct — Bitcoin is currently using 1 Mwh per transaction.
- fsflover 5y ago
  Except there is no connection between megawatts and transactions.
  - foepys 5y ago
    Because blocksize is fixed and blocks are consistently full, it's possible to estimate the next block's energy usage per transaction. All blocks are currently full as miners have an incentive to collect all transactions available, ordered by their attached transaction fee. If you mean that the same amount of power would be used whether the blocks are full or empty, you are technically correct but in practice it's not relevant until blocks are consistently not full.
  - Balantio 5y ago
    Does BTC makes sense if there would be no transactions? If transactions are critical for everyone, wouldn't that mean that keeping bitcoin mining alive is a fundamental part of transactions? We could argue that fiat keeps databases running (if we ignore physical money) and bitcoin is keeping blocks mining active. I don't think it is wrong to say megatwatts per transactions. We could ignore this completly and say 'the baseload of just keeping btc running is x megawatts per hour' and that would just ring the same bells. I'm pondering if we could also say something like "btc itself as a cryptosystem motivates actors to consume megawatts per hour due to the interest in btc and the current fiat<>btc exchange value"?
- chabes 5y ago
  That is not correct. Two glaring gaps in that concept... First, energy use is not tied to transactions. An empty block uses the same amount as a full block. And second, the transaction count we’re discussing is the count of settled base layer transactions. This doesn’t include the majority of transactions of value: those that occur off chain or through second layer transactions. An infinite amount of off chain and L2 transactions only need a single on chain transaction to settle.
hwillis 5y ago
You are correct, but for the wrong reason. Bitcoin's energy consumption[1] does not track with the number of transactions[2]; energy consumption has significantly increased while the number of transactions has not. What the energy consumption does track is the price[3]: mining creates coins, so higher price means more mining. The current ~4% inflation rate will not significantly decrease incentives for decades. In a sane world, the energy use and price would be tied to the actual usefulness of the currency. They are not. In roughly 100 years the energy use will be set by transaction fees and the number of transactions, but currently it's only proportional to speculative will. When transaction fees incentivize mining, the incentive to mine will certainly be much lower. Arguably more secure, too- the block reward creates a separate incentive for large players to dominate mining, which lowers the diversity of miners. It's not at all clear that the absolute energy consumption will be lower than it currently is. [1]: https://digiconomist.net/bitcoin-energy-consumption/ https://digiconomist.net/bitcoin-energy-consumption/ [2]: https://ycharts.com/indicators/bitcoin_transactions_per_day https://ycharts.com/indicators/bitcoin_transactions_per_day [3]: https://www.coindesk.com/price/bitcoin https://www.coindesk.com/price/bitcoin
piva00 5y ago
Isn't the network already saturated with transactions though? To the point where it was needed to implement something like Lightning Network to compensate for that?
- CyberDildonics 5y ago
  It was never needed. What was needed was to increase the max block size. Instead, new people took it over and kept the throughput to the rate of a 56k modem. Now for the cost of a single transaction you can pay for enough hard drive space to hold the entire chain and enough bandwidth for a billion transactions. This is like building a sidewalk instead of a freeway then saying you need to build a network of gondolas over it that just plunks groups of people down at different places on the sidewalk to move people.
  - piva00 5y ago
    Ah, yes, I'm very aware of that fact, my use of "needed" here was in the sense it was needed after all the ones making money from transaction fees took over and decided to not increase the block size.
UncleMeat 5y ago
But there can't really be 2x the transactions (or, at the very least there cannot be 100x the transactions). Transactions/block varies but there is a fundamental limit on how many bits can fit into a block. Transactions are the thing that matters. Simply securing the network achieves nothing if people cannot move money. That's why people measure the network with transactions/joule.
davewritescode 5y ago
Total energy / total transactions is a completely appropriate metric to use. If you could meaninufylly scale the number of bitcoin transactions and could amortize the high fixed costs of running the blockchain I'd agree with you but as it stands right now, you can't. Yes, I'm aware there's off chain solutions such as lighting but it's unclear to me whether those solutions are viable long term.

grey-area 5y ago

Unfortunately the network can’t handle more transactions. Decentralised, fast, cheap: pick two.

hwillis 5y ago
Bitcoin picked centralized, slow, and expensive. Most mining is done in giant warehouses, and the whole point is that more computing power results in zero faster processing. It remains to be seen if there are good ways to do distributed currency, but there are certainly less bad ways.
- grey-area 5y ago
  Sure, they could have made better choices, but the decentralisation is really a fundamental problem that can't be designed away. If you want it decentralised it's going to be significantly slower and more expensive than centralised networks.
  - hwillis 5y ago
    More computationally intensive, yes- but computationally intensive may mean "running on smartphones in the background for .1% reduction in battery time" or "a medium-sized country". Increased cost is certainly not so straightforward. If you're sufficiently distributed, using already-existing hardware, with spare compute on extremely efficient devices, it could conceivably cost less. Even if a centralized server farm would be doing an order of magnitude less math, smaller processors use an order of magnitude less power to do that math. Bitcoin proper has a central ledger- every miner needs to hear about your transaction to verify it. Lightning is a clumsy way of reducing how many actors need to be notified of your transaction. Better currencies include stuff like that as first class. It's all in the name of getting closer to a constant-number verification scheme that is closer to competitive with the O(1) of registering a transaction with a bank. Centralized credit/debit cards exist so that a big wealthy firm can say yes, this person has enough money for this transaction and I will guarantee the transaction by paying for it even if they can't. What I would really like to see is distributed, automatic guarantees: when you buy something at a coffee shop, people running validators on wifi will pick it up and use their staked currency as insurance (hedged by the system) that they know accounts who trust this particular account, and the transaction is valid. More people staking and trusting this account means more trust by the larger system, which then only has to validate aggregated transactions. Anyone announcing themselves at a validator plugs in at a given level of aggregation, all of which have different staking/network/latency/storage requirements. Unlike off-chain transactions (eg lightning), the system is guaranteed at every level. Any given transaction will still be validated dozens or hundreds of times. En bloc it will probably use tens or hundreds of times more energy than the server farms powering VISA. I'll be honest, I'm okay with that. I really like the idea of having a bank account that isn't tied to a company.
MomoXenosaga 5y ago
My country handles 5 billion electronic transactions a year (thanks to COVID cash practically died last year). Fast and cheap although to be fair not decentralised. To this day I have no idea what cryptocurrency was supposed to solved.

idreyn 5y ago

But it's worth clarifying that the number is correct — Bitcoin is currently using 1 Mwh per transaction.

fsflover 5y ago
Except there is no connection between megawatts and transactions.
- foepys 5y ago
  Because blocksize is fixed and blocks are consistently full, it's possible to estimate the next block's energy usage per transaction. All blocks are currently full as miners have an incentive to collect all transactions available, ordered by their attached transaction fee. If you mean that the same amount of power would be used whether the blocks are full or empty, you are technically correct but in practice it's not relevant until blocks are consistently not full.
- Balantio 5y ago
  Does BTC makes sense if there would be no transactions? If transactions are critical for everyone, wouldn't that mean that keeping bitcoin mining alive is a fundamental part of transactions? We could argue that fiat keeps databases running (if we ignore physical money) and bitcoin is keeping blocks mining active. I don't think it is wrong to say megatwatts per transactions. We could ignore this completly and say 'the baseload of just keeping btc running is x megawatts per hour' and that would just ring the same bells. I'm pondering if we could also say something like "btc itself as a cryptosystem motivates actors to consume megawatts per hour due to the interest in btc and the current fiat<>btc exchange value"?
chabes 5y ago
That is not correct. Two glaring gaps in that concept... First, energy use is not tied to transactions. An empty block uses the same amount as a full block. And second, the transaction count we’re discussing is the count of settled base layer transactions. This doesn’t include the majority of transactions of value: those that occur off chain or through second layer transactions. An infinite amount of off chain and L2 transactions only need a single on chain transaction to settle.

hwillis 5y ago

You are correct, but for the wrong reason. Bitcoin's energy consumption[1] does not track with the number of transactions[2]; energy consumption has significantly increased while the number of transactions has not. What the energy consumption does track is the price[3]: mining creates coins, so higher price means more mining. The current ~4% inflation rate will not significantly decrease incentives for decades. In a sane world, the energy use and price would be tied to the actual usefulness of the currency. They are not. In roughly 100 years the energy use will be set by transaction fees and the number of transactions, but currently it's only proportional to speculative will. When transaction fees incentivize mining, the incentive to mine will certainly be much lower. Arguably more secure, too- the block reward creates a separate incentive for large players to dominate mining, which lowers the diversity of miners. It's not at all clear that the absolute energy consumption will be lower than it currently is. [1]: https://digiconomist.net/bitcoin-energy-consumption/ https://digiconomist.net/bitcoin-energy-consumption/ [2]: https://ycharts.com/indicators/bitcoin_transactions_per_day https://ycharts.com/indicators/bitcoin_transactions_per_day [3]: https://www.coindesk.com/price/bitcoin https://www.coindesk.com/price/bitcoin

piva00 5y ago

Isn't the network already saturated with transactions though? To the point where it was needed to implement something like Lightning Network to compensate for that?

CyberDildonics 5y ago
It was never needed. What was needed was to increase the max block size. Instead, new people took it over and kept the throughput to the rate of a 56k modem. Now for the cost of a single transaction you can pay for enough hard drive space to hold the entire chain and enough bandwidth for a billion transactions. This is like building a sidewalk instead of a freeway then saying you need to build a network of gondolas over it that just plunks groups of people down at different places on the sidewalk to move people.
- piva00 5y ago
  Ah, yes, I'm very aware of that fact, my use of "needed" here was in the sense it was needed after all the ones making money from transaction fees took over and decided to not increase the block size.

UncleMeat 5y ago

But there can't really be 2x the transactions (or, at the very least there cannot be 100x the transactions). Transactions/block varies but there is a fundamental limit on how many bits can fit into a block. Transactions are the thing that matters. Simply securing the network achieves nothing if people cannot move money. That's why people measure the network with transactions/joule.

davewritescode 5y ago

Total energy / total transactions is a completely appropriate metric to use. If you could meaninufylly scale the number of bitcoin transactions and could amortize the high fixed costs of running the blockchain I'd agree with you but as it stands right now, you can't. Yes, I'm aware there's off chain solutions such as lighting but it's unclear to me whether those solutions are viable long term.