Cost per transaction on Bitcoin?

[irukandji]

Hi all, I am ding some research and and trying to work out the possible raw cost of a transaction on Bitcoin. I guess I would assume blocks were full. I am interested in the raw cost.
Does anyone know of any resources that could help?
Thank you

[1714946772]

1714946772

[mocacinno]

Hi all, I am ding some research and and trying to work out the possible raw cost of a transaction on Bitcoin. I guess I would assume blocks were full. I am interested in the raw cost.
Does anyone know of any resources that could help?
Thank you

That would be close to impossible because you have no clue about the amounts and types of ASIC's currently mining.

You can estimate the current network's total hashrate tough:
Hashrate = Difficulty * 2**32 / 600

So

Hashrate : 19298087186262 * 2^32 /600

Hashrate = 138.141.088.900.586.584.146 h/s
Hashrate = 138.141.088.900.586.584 kh/s
Hashrate = 138.141.088.900.587 Mh/s
Hashrate = 138.141.088.900 Gh/s
Hashrate = 138.141.089 Th/s
Hashrate = 138.141 Ph/s
Hashrate = 138 Eh/s

Now, you can estimate what would be a reasonable mix of asic's that would result in a total hashrate of 138 Eh/s, look up their power draw, multiply by what you think would be the average power price... Add the cost of the ASIC's, S&H, shelves, cooling, wiring, human labour,...

You could, just for fun, estimate the BARE MINIMUM.... A HUUUUUUUUGE underestimation tough!
Bitmain's latest ASIC is the S19 Pro, 110 Th/s, $2407 (not including customs, S&H, PSU,...). It draws 3500 Watt at the wall.

https://www.statista.com/statistics/263492/electricity-prices-in-selected-countries/
Tells us china's power price is 8 cent/Kwu. But let's assume most mining farms are located at places where they pay 4 cents/kwu (i've heared a rumour they're usually locating these things at places with excess power)

You'd need 1.255.829 S19 pro's to reach 138 Eh/s.
They'd draw 4.395.401.500 Watt at the wall.

So, each day, they'd use 105.489.636 Kwu (105 Gw). With a cost of 4,25 Million USD.

So, the absolute, ultimate minimum energy cost for 144 blocks would be 4.25 Million USD. That's ~$30.000/block.
This does not include the fact you'd have to buy 1,25 million ASIC's, and these calculations are for the newest asic on the market, and only paying 4 cents/kwh!!!!

[irukandji]

So, the absolute, ultimate minimum energy cost for 144 blocks would be 4.25 Million USD. That's ~$30.000/block.
This does not include the fact you'd have to buy 1,25 million ASIC's, and these calculations are for the newest asic on the market, and only paying 4 cents/kwh!!!!

Wow. Thank you so much for taking the time to help me. I appreciate it.
Now i need to find out how many tx can fit in a block. I'm sure that won't be too hard.
Thanks again.

[mocacinno]

So, the absolute, ultimate minimum energy cost for 144 blocks would be 4.25 Million USD. That's ~$30.000/block.
This does not include the fact you'd have to buy 1,25 million ASIC's, and these calculations are for the newest asic on the market, and only paying 4 cents/kwh!!!!

Wow. Thank you so much for taking the time to help me. I appreciate it.
Now i need to find out how many tx can fit in a block. I'm sure that won't be too hard.
Thanks again.

Also, for this part of your question, there's a huge difference between the amount of transactions that *could* theoretically fit into one block, the amount of transactions on average (all times) or the amount of transactions on average (using a sliding window).

I've just quickly used https://coinb.in/#fees without actually overthinking this... And it seems the absolute minimum tx size would be ~130 bytes. I'm assuming this is without witness data.
The maximum amount of tx data in one block (without witness data) is still 1Mb, so one block could theoretically fit ~7700 transactions. But these transactions would be 1 input - 1 output, all segwit transactions, nothing special, block 100% full. This is far from realistic.

If you look at https://blockstream.info/, you see that the average amount of transactions in one block is ~2500-2700.

Do realise that a lot of assumptions were made in my last 2 posts... I mean: you have no idear about the mix of ASIC's that currently make up the network, you have no idear about the power prices these asic farms pay and you're estimating the amount of transactions in a block... So your end result is basically a guesstimation....

[BitMaxz]

Wow. Thank you so much for taking the time to help me. I appreciate it.
Now I need to find out how many tx can fit in a block. I'm sure that won't be too hard.
Thanks again.

It always depends on the transaction size if how many transactions can be fit per block.

According to cryptoslate

3,500 transactions
Assuming an average transaction occupies 570 bytes of data, then a block can contain approximately 3,500 transactions, given the 1 MB limit.

But look at the chart from blockchain, the average transactions per block contains around 2k.
- https://blockchain.com/charts/n-transactions-per-block

[Wind_FURY]

There are some similar website that discuss it such as
1. https://digiconomist.net/bitcoin-energy-consumption (about footprint rather than raw cost)
2. https://www.cbeci.org/ (only shows estimated electricity used by all bitcoin miner)

But don't forget that it doesn't actually reflect cost/transaction since block size aren't affected by cost/transaction or hashrate.

Digiconomist has posted very biased articles against Bitcoin. There's truth in the facts presented, BUT they have spinned them that "Bitcoin/POW is wasteful" for the most part, without considering the fact the that POW is also "a value" in itself.

OP, be careful with your next assumptions without doing your research on Proof of Work, and it's function in Bitcoin.

[Carlton Banks]

Do realise that a lot of assumptions were made in my last 2 posts...

you forgot the most basic observation of all:

there is no relationship between the number of transactions in a block and the Bitcoin network hashrate


so there is no reason to ask the question in the OP, the answer tells you nothing of any meaning whatsoever, and so you entirely wasted your (and their) time

[stompix]

https://www.statista.com/statistics/263492/electricity-prices-in-selected-countries/
Tells us china's power price is 8 cent/Kwu. But let's assume most mining farms are located at places where they pay 4 cents/kwu (i've heared a rumour they're usually locating these things at places with excess power)

Nobody buys tons of brand new gear and hosts it as 8cents per kwh, basic ROI would be in 10-12 months not including other costs or downtime or many other aspects.
It would be pretty safe to say 4 cents is the above average, there are some miners here (on the mining board I mean) that have dirt cheap almost free energy  

So, the absolute, ultimate minimum energy cost for 144 blocks would be 4.25 Million USD. That's ~$30.000/block.

Well, in theory, the absolute minimum would be 0, since there are a few fairytales that miners only burn excess energy that is anyhow wasted  
If we take into account the 2 cents per kWh in China in Ordos due to coal power plants and dirt cheap energy in Sichuan, and we again believe(!) the majority of the global hashrate is focused there you could almost split that sum. The minimum is debatable, one thing we know is the maximum, the miner's costs will never manage to exceed the value of the daily reward+fees buy coins.

As for the transaction part, it's a bit of a useless metric, it's there to approximate things but one tx can have 100 outputs and it means 100 people have received money or there can be 100 tx with1 input/output with the same result, so if we speak about costs we must speak in terms of utility, rather than the number of tx I would focus on outputs.

Bottom line, I have to agree with Carlton Banks

[mocacinno]

--snip--

you forgot the most basic observation of all:

there is no relationship between the number of transactions in a block and the Bitcoin network hashrate


so there is no reason to ask the question in the OP, the answer tells you nothing of any meaning whatsoever, and so you entirely wasted your (and their) time

True, but that doesn't mean you can't do a very raw estimation about how much power is burnt per transaction when using a short window at current diff if you assume full blocks... I mean, the OP already gave the assumption of full blocks, and eventough the diff retargets every 2016 blocks, it's possible to do estimate the hashrate when knowing the diff at a certain point in time. It's also possible to calculate the average transaction size in the last couple of weeks/months and see how many would fit in a block (on average).

However, the mix of ASIC types/vendors and the average power price are a big unknown... I assumed 4 cents/kwh was cheap, but apparently it was still to high... I live in a country where i pay ~30 cents/kwh

I, for one, would be interested to know how much the electricity cost per transaction given full blocks and current diff would be... It's interesting to the point that the current block reward covers most of these costs right now, but in a couple halvings these costs have to be covered by an increase of the fees, or by more efficient hardware, or by a lowering of the total network hashrate or a change in POW algo or a combination thereof...

[Carlton Banks]

so there is no reason to ask the question in the OP, the answer tells you nothing of any meaning whatsoever, and so you entirely wasted your (and their) time

True, but that doesn't mean you can't do a very raw estimation about how much power is burnt per transaction when using a short window at current diff if you assume full blocks...

We can calculate worldwide banana shipments for the same period of time, and  but it would be similarly fruitless (in b4 "banana is not a fruit", I did say "fruitless" )


There is precisely _zero_ power "burnt per transaction", it's a very easy calculation that does not require estimation. There _is_ an amount of power consumed per block solution computed, and the transactions included in the body of the block are completely irrelevant to this, they are not hashed as part of the solution, and (to demonstrate this further) need not be included at all (and frequently are not)

I mean, the OP already gave the assumption of full blocks, and eventough the diff retargets every 2016 blocks, it's possible to do estimate the hashrate when knowing the diff at a certain point in time. It's also possible to calculate the average transaction size in the last couple of weeks/months and see how many would fit in a block (on average).

However, the mix of ASIC types/vendors and the average power price are a big unknown... I assumed 4 cents/kwh was cheap, but apparently it was still to high... I live in a country where i pay ~30 cents/kwh

I, for one, would be interested to know how much the electricity cost per transaction given full blocks and current diff would be... It's interesting to the point that the current block reward covers most of these costs right now, but in a couple halvings these costs have to be covered by an increase of the fees, or by more efficient hardware, or by a lowering of the total network hashrate or a change in POW algo or a combination thereof...

or by an increase in the BTC exchange rate. Or by a reduction in the hashrate. There are, as you allude to, several parameters to mining economics (not least that lightning will be much more reliable/mature by then, and miners can expect significantly higher quantity of blockweight to fill, as open/close commitment transactions can more effectively fill the maximum 4 Mweight units in a block, hence more fee income will be available to miners in this way also)

So the calculation you're suggesting doesn't provide a useful model for predicting transaction fees, not least because there is no relationship between transaction fees and the hashrate. You are only serving to further underline the fact

[mocacinno]

--snip--
We can calculate worldwide banana shipments for the same period of time, and  but it would be similarly fruitless (in b4 "banana is not a fruit", I did say "fruitless" )

The banana analogy could work aswell: if you assume full containers being shipped from whatever country they import bananas from, how much fuel is used per banana given that a ship only transports bananas and doesn't carry other containers aswell?
Same thing: sure, a container can be shipped empty aswell, it's not the banana that has the biggest impact on the fuel consumption (eventough the banana does have a certain volume and weight, so it does have some impact, but i'm  ignoring this for the sake of the argument) but that's why there's an assumption of full containers. The main reason why containers are shipped from country x to my country is to transport bananas. If there was no need for bananas to be transported, there would be no containers and the ship would remain in it's harbour, so calculating the fuel usage per banana is something you could do, it makes economic sense to set the minimum price level of a banana in a shop, eventough empty containers can be shipped, and the banana's impact on the fuel usage is low (at least for the sake of this argument).

Same here: a miner could mine empty blocks, but their main function is to include transactions... So, eventough there's no 1<=>1 relationship between the transactions and the power usage, it's still feasible to assume the main reasons why the blockchain exists is to store transactions in an immutable way, and the mining process uses power, so you could estimate the amount of power used to make 1 transaction immutable, even without a direct correlation .

There is precisely _zero_ power "burnt per transaction", it's a very easy calculation that does not require estimation. There _is_ an amount of power consumed per block solution computed, and the transactions included in the body of the block are completely irrelevant to this, they are not hashed as part of the solution, and (to demonstrate this further) need not be included at all (and frequently are not)

That's why i'm working under the OP's assumption of full blocks. I realize there is ~zero power per transaction used (actually, a merkle tree IS being created, the merkle root has to be included in the block header, so some cpu cycles are being used, using a tiny amount of power), but there IS a lot of energy being burned by creating sha256d hashes of the block headers untill a header is found whose sha256d hash is under the current target. The OP asked about full blocks...

--snip--

So the calculation you're suggesting doesn't provide a useful model for predicting transaction fees, not least because there is no relationship between transaction fees and the hashrate. You are only serving to further underline the fact

I'm not building an exact model here... I never said i was searching for a formula that gives an exact sollution to the question of energy used per tx in a block in the longest chain. I don't know why the OP asked the question he asked... Maybe he's interested in future fees, maybe he was just interested, maybe he wanted to use this number for something else..
I made it very clear that it is a raw estimation.

[Carlton Banks]

So, eventough there's no 1<=>1 relationship between the transactions and the power usage, it's still feasible to assume the main reasons why the blockchain exists is to store transactions in an immutable way, and the mining process uses power, so you could estimate the amount of power used to make 1 transaction immutable, even without a direct correlation

Bitcoin's design means either you're protecting every previous block (and the transactions appended to them) from subversion, or none of them

The miner can:

• mine a new block on the longest chain with the most work -> every tx in every block becomes equally immutable corresponding to how many blocks/work they are buried under
• or mine their own chain -> none of the Bitcoin tx's become any more or any less immutable

There is no option to choose which transactions are conferred with hashpower protection, or by how much hashing power they are protected by. So what's the point of measuring such a thing?

[mocacinno]

So, eventough there's no 1<=>1 relationship between the transactions and the power usage, it's still feasible to assume the main reasons why the blockchain exists is to store transactions in an immutable way, and the mining process uses power, so you could estimate the amount of power used to make 1 transaction immutable, even without a direct correlation

Bitcoin's design means either you're protecting every previous block (and the transactions appended to them) from subversion, or none of them

The miner can:

• mine a new block on the longest chain with the most work -> every tx in every block becomes equally immutable corresponding to how many blocks/work they are buried under
• or mine their own chain -> none of the Bitcoin tx's become any more or any less immutable

There is no option to choose which transactions are conferred with hashpower protection, or by how much hashing power they are protected by. So what's the point of measuring such a thing?

Because, at the end of the day, the miners get confronted with an energy bill, a writeoff of their equipment, labour cost,... Most miners do add transactions from their mempool to the blocks they're trying to solve. If no miner would add transactions to the blocks they're trying to solve (and solve the block), the mempools would fill and no transactions would get confirmed anymore (except coinbase transactions) and bitcoin would become useless. I agree that the protocol allows empty blocks, and there is no 1<=>1 correlation.

So, in the end, if you knew the exact amount of active ASIC's, the exact power price of each miner and all other costs, you could calculate how much money is being spent by the miners in a certain timeframe and you can count how many transactions end up in the chain with the most work during this same timeframe. So you can calculate the cost/tx in that timeframe.

Is this number usefull? I have no idear... But it IS something you can calculate...

[Carlton Banks]

So you can calculate the cost/tx in that timeframe.

I know.

Is this number usefull? I have no idear... But it IS something you can calculate...

in my posts above, I am outlining why I consider it to be not useful. You are replying to the posts in which I make the case that the number is not useful. Hope that helps.

[HCP]

There is precisely _zero_ power "burnt per transaction", it's a very easy calculation that does not require estimation.

Just to play "pedant's advocate"... that's not exactly true... A user (or service) will burn power constructing a transaction, they'll burn power broadcasting a transaction, other nodes will burn power receiving/relaying/storing that transaction.

One could say that this is the "hidden" cost of Bitcoin... granted, it's probably tiny compared to the costs of power being consumed by mining pools etc... but it's not zero   It's like the "cost" of email.


As for how you would go about estimating what this cost actually is...

[PrimeNumber7]

<>
there is no relationship between the number of transactions in a block and the Bitcoin network hashrate

I have to disagree with you on this.

There are many alt(shit)coins out that that have very low transaction fees that also have very few transactions confirmed per hour. I believe the reason for this is the lack of security altcoins provided compared to bitcoin. Specifically security provided by the miners (hashrate) that is well distributed among many entities and geographically.

The risk of a transaction being double spent after two confirmations is near zero (this is not something that happens), while many altcoin blockchains, including 'major' altcoins (defined by market cap) have seen 51% attacks involving transaction double spends much deeper than the equivalent of two confirmatoins.

[pooya87]

There are many alt(shit)coins out that that have very low transaction fees that also have very few transactions confirmed per hour. I believe the reason for this is the lack of security altcoins provided compared to bitcoin. Specifically security provided by the miners (hashrate) that is well distributed among many entities and geographically.

you are confusing a couple of topics here. we are talking about bitcoin not altcoins and if you want to bring up altcoins then their lack of transaction/block is mainly because of their lack of utility. those people who own these altcoins are gamblers who will also store it on exchanges where they gamble with the price. in fact most of them don't even know or care what security and 51% attack is. if they did they wouldn't have bought those coins in first place.

[LoyceV]

So, the absolute, ultimate minimum energy cost for 144 blocks would be 4.25 Million USD. That's ~$30.000/block.
This does not include the fact you'd have to buy 1,25 million ASIC's, and these calculations are for the newest asic on the market, and only paying 4 cents/kwh!!!!

For a back of the envelope calculation that's not bad. The block reward is more than double your result, and is also used to pay for hardware.
Based on basic economic principles for a free market, I expect the raw cost per block in the long term to be close to the reward per block, which is the sum of the block reward and transaction fees.