What is the cost of the bitcoin system running?

[zby]

Anyone seen some good estimates of the cost of sustaining the bitcoin system?  That is mostly the cost of mining.

[1713963683]

1713963683

[1713963683]

1713963683

[cbeast]

Are you doing a cost/benefit analysis? If so, there are many things to factor.

[Jack of Diamonds]

Converted to a common denominator (Radeon 6990) the total hash rate equals about 28,000 of those GPUs.

In that scenario the total power usage of the entire network is 10,500,000 watts at any given moment

(There are very many diverse GPUs, CPUs and maybe even some ASICS the network too so that's just a very rough estimate)

[zby]

Converted to a common denominator (Radeon 6990) the total hash rate equals about 28,000 of those GPUs.

In that scenario the total power usage of the entire network is 10,500,000 watts at any given moment

(There are very many diverse GPUs, CPUs and maybe even some ASICS the network too so that's just a very rough estimate)

Thanks!  Continuing this rough estimate - if we take the price of MegaWattHour to be about $100 (from http://en.wikipedia.org/wiki/Cost_of_electricity_by_source) - that means about $1000 per hour.

[Rassah]

Well, it's about $1.60 a day for my dual ATI 5830, 550MHash/s system...

[saqwe]

Converted to a common denominator (Radeon 6990) the total hash rate equals about 28,000 of those GPUs.

In that scenario the total power usage of the entire network is 10,500,000 watts at any given moment

(There are very many diverse GPUs, CPUs and maybe even some ASICS the network too so that's just a very rough estimate)

wow let's all get on a 10 MegaWatt Boat
/get 28,000 6990's on a boat

World's Biggest Wind Turbine Generates 10 MW And It Floats!
http://www.treehugger.com/files/2010/02/norway_to_build_worlds_most_powerful_offshore_wind_turbine.php

[FreeTrade]

Converted to a common denominator (Radeon 6990) the total hash rate equals about 28,000 of those GPUs.

In that scenario the total power usage of the entire network is 10,500,000 watts at any given moment

(There are very many diverse GPUs, CPUs and maybe even some ASICS the network too so that's just a very rough estimate)

Thanks!  Continuing this rough estimate - if we take the price of MegaWattHour to be about $100 (from http://en.wikipedia.org/wiki/Cost_of_electricity_by_source) - that means about $1000 per hour.

Or about $750,000 per month. It shows a lot of faith in BTC that we're putting this much into maintaining it. It's more funding than a lot of startups get.

[piuk]

Anyone seen some good estimates of the cost of sustaining the bitcoin system?  That is mostly the cost of mining.

http://pi.uk.com/bitcoin

Total miners revenue   $28,042.91
cost per transaction   $4.56   
% of transaction volume   8.24 %
Hash Rate   10,199.79 GH/s   
Electricity Consumption *   159.12 megawatt hours
Electricity Cost   $23,867.52

[Red]

Awesome link. That is a staggering amount of electrical cost. The site shows a theoretical net operating profit per day. Market value of mined coins minus electrical cost.

However, has anyone totaled up exactly how many Dollars/Euros/etc. actually flow into exchanges each day? I'm assuming electrical costs must be paid in some national currency. Is there enough actual incoming cash to pay all the mining bills?

[Revalin]

We only see trades, not deposits and withdrawals, so it's hard to say how much fiat is actually put into the system daily.  However, any time there's not enough influx of fiat to cover what the miners dump, the price will fall.  With falling price, miners lose incentive and drop out, which causes the power consumption to fall.  In the long run this balances out.  The daily currency trades are a lot greater than the power bills, and I expect they're really not that strong a force in the market...  Mining follows price, not the other way around.

And yes, that's an awful lot to keep the network running.  50BTC/block is an lot of payout for a currency that doesn't have a very strong economy, but it's propped up by wild-eyed speculation (but less every day, fortunately).  After block 210,000 the reward drops to 25BTC, thus cutting the mining costs in half, and similar drops occur in the future.

So while it's absurd now, the mining cost will fall in the future (resulting in less power spent mining), unless the BTC starts getting some wider acceptance, in which case greater payouts to miners are justified.

[Red]

We only see trades, not deposits and withdrawals, so it's hard to say how much fiat is actually put into the system daily. 

Pity, that would be great information to know.

Mining follows price, not the other way around.

The fascinating part is how fast the number of miners increases to eat up any potential profit in mining!

So while it's absurd now, the mining cost will fall in the future (resulting in less power spent mining), unless the BTC starts getting some wider acceptance, in which case greater payouts to miners are justified.

It is fascinating what the "cost of distrust" is.

Technically speaking, if everyone were to trust any single peer to keep accurate records of all valid transactions and settle the consensus. (What every peer has to do now anyway.) That would be the optimum in electrical efficiency.

If everyone was to trust one of a few dozen peers, all of whom distrusted all of the other few dozen peers, then that would be a few dozen times less electrically efficient for the same behavior.

By requiring every peer to duplicate all the work of every other (distrusting all other peers) we create a system that is maximally electrically inefficient.

Mining competition based on price seems to optimize convergence toward maximum electrical inefficiency.

[Rassah]

By requiring every peer to duplicate all the work of every other (distrusting all other peers) we create a system that is maximally electrically inefficient.

Mining competition based on price seems to optimize convergence toward maximum electrical inefficiency.

The rule is that mining cost will likely equal bitcoin price, efficient or not. So even if there were just a dozen trusted peers doing the mining very efficiently, they would likely spend as many resources as possible on mining as fast as possible, until the cost of their efficient mining matched the value of the bitcoin or transaction fees they produced.

[Revalin]

Pity, that would be great information to know.

I'd love to know!  Only MtGox knows, and they're not telling.

Technically speaking, if everyone were to trust any single peer to keep accurate records of all valid transactions and settle the consensus. (What every peer has to do now anyway.) That would be the optimum in electrical efficiency.

PayPal, Pecunix, Flooz, Beenz....  No thanks.  I want to own my money, and I don't want corporate slimeballs skimming off of my commerce.

Decentralized cryptocurrency has been done before, but the double-spend problem kept it from being adopted for over a decade.  It's inefficient, but mining is a great solution for that problem.  I'd love to just be able to trust everyone to not double-spend, but people suck.  

It is fascinating what the "cost of distrust" is.

It's also a clever way to handle the initial distribution of a decentralized currency.  "I'll issue it all to myself and distribute it fairly to everyone" doesn't work.  The usual centralized-currency method (fixed exchange rates to something with redeemable value) doesn't work.  Making people perform proofs-of-work and getting security against double-spends in the process is pretty cool.  Can you think of a better way to issue a decentralized currency?  

For now I'm glad to see the exchange rate falling; it'll decrease the energy wasted on mining and by extension the amount we're subsidizing miners for every transaction.  That will also improve when the mining rewards drop.

So sure, it's expensive, but so are central issue skimmers, and for that price we get a solution to several otherwise hard problems.

[Red]

The rule is that mining cost will likely equal bitcoin price, efficient or not. So even if there were just a dozen trusted peers doing the mining very efficiently, they would likely spend as many resources as possible on mining as fast as possible, until the cost of their efficient mining matched the value of the bitcoin or transaction fees they produced.

Actually this is called the iterated prisoner's dilemma problem. If the dozen peers did the mining cooperatively, they could reap much higher profits that mining competitively.

Your rule is that prisoners are unlikely to cooperate even if it is in their own self interest. Mining pools show this isn't true. Competing mining pools show it is!

It's also a clever way to handle the initial distribution of a decentralized currency.  "I'll issue it all to myself and distribute it fairly to everyone" doesn't work.  The usual centralized-currency method (fixed exchange rates to something with redeemable value) doesn't work.  Making people perform proofs-of-work and getting security against double-spends in the process is pretty cool.  Can you think of a better way to issue a decentralized currency? 

[edit] fixed the link!

Actually, yes I can. I have a concept discussion thread here if you are interested.

[Revalin]

Actually, yes I can. I have a concept discussion thread here if you are interested.

I am!  Please fix your link so I can read about it.