It's easy to calculate a number for the waste of Bitcoin, but it's not easy to calculate a number for the waste of the awful, archaic systems it could replace. Transferring money is done today with mainframes, paper forms, and middlemen galore. If Bitcoin replaced all that there's no doubt in my mind we'd all be better off.
This is a faith-based assertion, when data is needed. I am heartily skeptical that, were Bitcoin economy as big as the US dollar economy, that it could ever be as efficient as our system of middle-men, paper forms, etc. US Dollars extract transaction-based fees, whereas Bitcoin extracts continuous make-work in order to maintain consistent value. Say credit cards take a 1% fee on a transaction. At $150,000 per day, and a $2 billion market cap, the overhead is a little over 2.5% per year for Bitcoin. (Not that I think these numbers are completely solid, I'm just going off the best estimates available right now. Also, I'm not an economist, I'm just throwing together a napkin-calculation. Critique away!)
Some things are very amenable to solving with technology. Other things are more amenable to solving with people. Only data will tell for sure.
Interestingly, the original Bitcoin paper opened by addressing exactly these concerns: the origin of transaction costs in current payment systems consisting of middle-men (central trusted parties, in more charitable terms), paper forms, etc.
While the system works well enough for most transactions, it still suffers from the inherent weaknesses of the trust based model. Completely non-reversible transactions are not really possible, since financial institutions cannot avoid mediating disputes. The cost of mediation increases transaction costs, limiting the minimum practical transaction size and cutting off the possibility for small casual transactions, and there is a broader cost in the loss of ability to make non reversible payments for nonreversible services. With the possibility of reversal, the need for trust spreads. Merchants must be wary of their customers, hassling them for more information than they would otherwise need. A certain percentage of fraud is accepted as unavoidable.
The emphasis on supporting small casual transactions seems to have created a number of problems for the bitcoin infrastructure. Mt. Gox charges a small % for transactions (rather than a flat fee), a by product of this is it allows people to flood the exchange with a ton of micro transactions without any additional penalty for the volume of orders. Seems like this could be part of the reason they have been having issues.
Doesn't this also impact the block chain in some way? A ton of small transactions can flood the chain, requiring more computational energy to be spent than the transactions are actually even worth.
I'll also admit ignorance here, but it doesn't seem valid to compare overhead as percent of transaction vs overhead as percent of market cap.
Current estimated USD Transaction Volume is ~$70 Million per day[1], so the transaction fees needed to sustain the $150k per day electricity overhead would be ~0.2%.
All of the parameters in this calculation are in high flux of course (and we're not including total cost to mine here, just electricity), but when I've run this same calculation in the past at various times I've gotten similar results.
Note also that mining is profitable enough due to new coin creation that transaction fees in practice are much smaller than this for now (average ~0.01%), but coin creation will gradually taper off over the years.
Making up statistics out of thin air doesn't help anyone. Credit cards charge more than double what you speculate. Neither you nor I have the data necessary to calculate real numbers. The important point is we shouldn't assume Bitcoin is worse just because the waste is more obvious and/or less socially acceptable.
Yeah, my understanding is that miners are not just mining, they're also powering the distributed transaction engine of bitcoin. Is that correct? In which case the power consumption needs to be compared to (per $ or whatever) consumption for Bacs, swift, visa etc
They are doing a make-work exercise which only needs to be time consuming, in order to prevent cheaters. It's mostly throw-away work. The actual computation that needs to happen is pretty trivial. I think it would be possible to fix Bitcoin so that "proof of work" is something that doesn't consume all that electricity, but I'm not sure what it would be.
As I understand it, Ripple's trust system works by selecting multiple third-parties that are unlikely to collude against you, and checking that they agree on a transaction before considering it "confirmed".
I understand that its a make-work exercise, I was querying whether the output is just mined coins, or whether the output also includes verifying everyone elses transactions. My current understanding is that its the latter.
Sure, so I'm not sure which "output" you mean. The output of the computationally-intensive part is really just a magic number that solves a puzzle. Once a miner finds that magic number, they publish a block that has the most recent transactions (which are all verified separately), as well as the special transaction where the miner grants themselves the current mining reward. Other miners see it and verify that all that stuff was correctly computed.
So that whole block (which is maybe what you're calling the "output"), does contain the puzzle solution as well as all the latest transactions. But the actual specifics of the puzzle (which is currently brute-force SHA1 reversing), aren't super important. It could always be swapped out for some other task.
By 'output' I (think) I meant the output we get for the energy used by the miners.
You seem to be saying that The Miners have to verify transactions to claim their reward? When you say "which are all verified separately" - you mean verified by the same miner who publishes the block? If so then the bitcoin network as a whole benefits from their energy use.
On the other hand, I see your point that the make-work is not intrinsically linked to transaction verification. But my understanding is, its the difficulty of the make-work that protects the block-chain from being hijacked by ..er... 'the bad guys', whomever they may be.
So, the whole thing kinda makes sense, and the 'make-work' is like an energy commitment that keeps the whole thing integrated. The assumption being, the total commitment of all bitcoin users will always be greater than the commitment of any single attacker/attacker group.
Nope. The GPUs get plenty hot. The difference is that with Litecoin, your CPU gets hot too! I have not dug into the technical details, but at a glance it looks as if Litecoin motivated research that fairly successfully overcame the GPU-unfriendliness of Scrypt. If anyone here can confirm or deny this, I'd love to hear it.
The CPU doesn't get that hot. On a machine with an Athlon II 240e cgminer uses ~18% CPU vs the GPU that sits at 99% load. scrypt is indeed viable on GPUs, but it depends onto the available memory bandwidth which presumably lessens the impact of FPGA/ASIC if people start to push those as LTC mining solutions.
"By convention, the first transaction in a block is a special transaction that starts a new coin owned by the creator of the block. This adds an incentive for nodes to support the network, and provides a way to initially distribute coins into circulation, since there is no central authority to issue them."
From whats said here https://news.ycombinator.com/item?id=5548503 miners grant themselves a fraction of a bitcoin for each block, and that fraction is what decreases over time.
Wouldn't - if bitcoins was to become mainstream - wouldn't there be a need for bitcoin debit or credit cards though and atms and banks. So realistically we are only limited to comparing it to actual transaction costs.
It's not necessarily waste. My home is heated using an electric furnace. The heat given off by GPU mining offsets the furnace. It's essentially "free" electricity while mining during winter months.
Since you mentioned it - note that heat pumps are like 5 times more effective than gas or resistive heating. Really, given that they tend to have 400% - 500% efficiency, everything else is a waste.
I was trying to find stats on the energy/environmental impact involved in producing a single penny. What I did find was that at times a penny costs nearly double its worth to produce [1].
- many countries have eliminated pennies
- mining will have to continue even after "coins" aren't produced, to verify transactions.
It's more apt to measure power per transaction, against a traditional bank (or a handful of cash, amortized over it's life.) There was a good explain xkcd about the cost of carrying around physical money in your car.
All Bitcoin documentation ever. The purpose of mining is actually not to generate BTC; the purpose is to prevent double spending and the block reward is just an incentive.
The power numbers cited by this article are very inaccurate when considering current mining hardware and software.
"* Electricity consumption is estimated based on power consumption of 650 Watts per gigahash and electricity price of 15 cent per kilowatt hour. In reality some miners will be more or less efficient."
Most serious miners pay less than $0.15/kWh.
An Avalon ASIC does 65Gh/s on 620W (9.5 J/Gh)
A BFL Single FPGA does 800Mh/s on 80W (100 J/Gh)
Even a dual ATI 5970 rig will do 1300Mh/s on 600W (461 J/Gh)
Currently over 1/3 of the network is ASICs (300 * 65Gh/s of Avalons + 7Th/s from ASICMINER) and another 1/3 is probably FPGAs.
The problem is people also mine on dual core CPU's. When looking for total energy consumption you need to include the long tail of horrible hardware because of botnet's / school computers where the miner does not directly pay for energy costs and so does not care about efficiency.
PS: Consider a ten hour 600 mile trip by a 60mpg car and a 6mpg camper. The average fuel consumption is not (60mpg + 6mpg) / 2 = 33mpg, but (2*600miles)/(600miles/60mpg + 600miles/6mpg) or 10.9mpg.
Nobody mines on a CPU anymore, there's absolutely no point. At the current difficulty, you would barely even get a single share. For example, a single share on p2pool is at difficulty 725 now, there's no way a CPU miner can solve that.
Botnets. School computer labs. Kids in their parent's basement.
No rational person who bears the full cost of their actions would mine on a CPU anymore. But 1) not everyone is rational and moreover 2) not everyone bears the full cost of their actions.
Are these ASICS going to be good for anything after mining coins? I can't help thinking they're like a shovel that only works within a goldfield and has no utility (other than as a doorstop) once superseded by faster technology or obviated by the last Bitcoin being mined.
Keep in mind that hashing data and brute-forcing a hash function require different hardware. There are no real applications that brute-force chained SHA hashes.
After the last Bitcoin is mined (and actually much sooner than that), the profit from mining will have shifted away from new coins and into transaction fees.
Mining new coins and processing transactions are the same thing. So the hardware can remain useful as long as Bitcoin stays in use.
(Could it get obsoleted by better hardware? Of course, but that's true of every piece of electronics.)
I don't know enough about the system but is it also possible to mine for "lost" coins?
So someone started out mining and their wallet was corrupted or their hard-drive failed or what-have-you but at the time BTC were worth a few ¢. Is it possible to cryptographically reclaim those lost coins by claiming to have a hash or something and seeing if it verifies? Could one test, like a brute force check, in a separate environment so as not to leak that you're trying to reclaim a coin.
Or is this impossible within the restraints of the system.
[I can't see how it can be mathematically impossible, if a coin was verified and placed in a wallet then there's some "code" that says "I have the wallet with that coin"; perhaps though it's impossible at present to mine for that code successfully in a worthwhile way??]
So not impossible, you _just_ have to guess the owners private key?
Presumably you could in theory brute force that in a million years or whatever. Also if there were an example of data encrypted with that key (is there in the block chain by virtue of the protocol?) then if the private key system were broken - eg by advance in quantum computing or a weakness found in the crypto used - then perhaps it would then be feasible.
The corollary of this appears to be that you'd then be able to take anyone's bitcoins?
I guess what I am confused by is that the transaction processing is much easier than mining (ie it's not going to get more and more difficult to do do over time) so you'd need to be processing a very large number of transactions before you'd need an ASIC. If you're a bank or BTC exchange then it's a no-brainer, but I wonder if more than a small number of these devices will be able to do enough work to justify their power consumption as time goes by.
>I guess what I am confused by is that the transaction processing is much easier than mining (ie it's not going to get more and more difficult to do do over time)
Are you sure this is true?
My understanding is that the proof-of-work requirements will be the same even after no new coins can be mined. Otherwise, the integrity of bitcoin would be threatened.
The transaction fees are won in the same fashion as the mining rewards, so given a high enough level of paying users there should be incentive to run nodes (the transaction fees might even overtake the mining rewards before the last new coins are awarded, if things play out in a bitcoin friendly manner).
But if you are mining, there are fewer and fewer new coins to be discovered over time, so you have to compute more and more hashes in hopes of finding a new one. My (possibly incorrect) understanding is that processing a transaction 1 year from now will be no more difficult than processing one today, whereas in the future you'll have to mine for longer to have the same chance of getting a BTC.
In other words, doesn't the workload for processing each transaction stay the same over time?
From the moment I saw this I immediately thought it's probably wrong, because they think the biggest volume of mining is done with PC's, when in fact it's with ASIC's and FPGA's now, which are far more efficient than PC's.
So that is about 55 million dollars a year to completely secure the network and process all transactions. I am not entirely sure how it will scale, but I predict that it will scale logarithmically with the number of transactions, especially considering that mining equipment will become more efficient.
Meanwhile the United States will spend 797 million dollars [1] this year in simply printing and transporting physical currency, not including all the other obvious costs in maintaining a modern currency.
Yeah they should have linked to the guys actually shipping product... Butterfly said on their FB they are shipping next week apparently. But who knows w/ them
BFL has (edit) not been (/edit) know to ship -but I can confirm that people have gotten their orders!- becuase It's taken some time as they've had problems with getting ASIC's. They supposedly will start sending them out at mid april (15th) which is tomorrow, so I'm going to wait and see what happens. If they do ship then I'll place my order.
They probably mentioned BFL because they're currently accepting orders and also have had more media attention because of the whole scam controversy.
I was just surprised you'd admit it in public, and thinking about what could happen if the school found your ASIC and knew what it was. Of course, you could make a counter-argument that you were entitled to use the school's electricity in partial exchange for your tuition fees :)
In a perfect mining market the total amount of money spent by miners each day (cost of electricity/bandwidth + amortized cost of hardware) will always tend to converge on the value of the mining rewards (minus some profit margin), which for the next 4 years is about 3600 BTC per day (25 BTC/block * 6 blocks/hour * 24 hours/day).
If the value of Bitcoins go up, so does the number of miners securing the network, which makes sense. It's pretty ingenious.
It says the network has about 9 million hosts, which at about 200 watts (ballpark, GPUs can consume a lot of power) would consume about 2 megawatts of electricity. Bitcoin consumes 892.54 MWh/day = 37 megawatts.
Imo the Bitcoin Goldrush is long gone. In mid 2011 you could make a couple of hundred $ per month with a decent Rig that had 3 decent GPUs. Todays numbers are far from that.