Hacker News new | past | comments | ask | show | jobs | submit login
Bitcoin Mining Uses $15 Million's Worth Of Electricity Every Day? (forbes.com/sites/timworstall)
28 points by peter123 on Dec 4, 2013 | hide | past | favorite | 44 comments

Previously debunked: the actual number is 1 or 2 orders of magnitude lower. https://news.ycombinator.com/item?id=6829860

I was curious so here's my napkin calculations:

Assume a block erupter is the base level to get into mining now. The network hash rate right now is 6,751,767.65 GH/s. Let's round to 6,600,000GH/s. Block erupters are 330MH/s and take 2.5w. The network would need 20m of them. That would be 50MW, or 1,200MWh / day. PG&E charges me $0.15 per kWh. So running the network would be $180,000 / day.

Yeah there's a lot of assumptions in there, but that's nowhere near $15m.

Isn't that still very expensive?

Well that's a very subjective question. Compared to keeping an ounce of gold under your mattress, yes. Compared to Western Union's daily operating costs, no.

The natural equilibrium is where it costs a little less than $1 to make $1 worth of bitcoin. The system isn't quite at equilibrium now because of price uncertainty and the lag time to order new mining equipment, but the invisible hand is working on it.

Keep in mind that the equilibrium is where the total cost to make $1 of bitcoin is $1 -- not where the electricity cost is $1. The relative weight of electricity vs. amortized hardware cost will vary, of course.

If you're doing accounting (backward-looking financial reporting) you should look at amortised cost. If, on the other hand, you're making a decision about whether or not to mine, then you care only about marginal (cash) costs.

If you already paid for your hardware (or already committed to paying for it), then there's zero additional cost, unless (i) there are significant maintenance costs or, more likely (ii) you could recoup some of the cash by selling the equipment if you stop mining. I'm ignoring time value of money, and also the drop in the resale value of the equipment. (Deliberately avoiding the work 'depreciation' here as the accounting meaning the every day meaning confuse the issue.)

This is going to be interesting since the volatility of bitcoin price in USD is so high. If you have +15% one day and -30% two days after, this completely destroys your forecast if and when mining is still profitable.

It should be, otherwise it would be easy to attack the network

A Block Eruptor ASIC (BE100) uses 10W/GH/s, a newer chip like Bitfury uses 0.9W/GW/s, the Block Eruptor 2 chip uses less than 0.2W/GH/s. It's decreasing rapidly.

if its about $150K a day - let's compare that to other world-wide power wasters:

- Old type base load power stations run with coal or other fossil, nuclear e.a. 24h/day but only contributing energy a few days a month (high estimate). IMHO they are only running because it's such a great business for the large power supplier conglomerates (otherwise they would have long replaced them with more efficient, quick starting solutions) - rough guess 2-3 are wasting comparable electric energy per day and there are 1000+ of those power stations around the globe.

- 100M+ people having their office computer(s) running at least 8h/5days-week but rarely using them or using computers 10-20 times more powerful than their applications are requiring. Alternative approach - using operating systems that require the latest, high powerful computers just to display some glossy windows, but are solely used for text processing e.a. - rough guess 3-5M of those equal to the amount of energy used for BitCoin production.

with these two examples alone you already have at least 400-500 times the energy used for BitCoin production each day just puffing away into hot air and I'm sure there are plenty more examples available.

We are wasting energy (and depleting limited resources) on a very large scale every day and only because it's such a great business today (not calculated at true cost) seemingly nobody needs to care about tomorrow.

It would be interesting to compare this figure to Western Union's electricity usage from their network infrastructure and all their brick-and-mortar locations, and see which of the two supports more transaction volume per kWh expended.

If this $15M figure were true (and it probably isn't), at today's going rate of about $1K USD/BTC, miners as a whole are losing $11.4 Million in electrical costs. (assuming 25 BTC are paid out every 10 minutes, and ignoring miners fees. The value of all new mined coins is only 3600 BTC per day, worth about $3.6 Million USD.)

That $15M figure is wildly speculative, and assumes an average wattage per gigahash that's more in line with the state of mining as much as 2 years ago, when GPUs dominated mining.

ASIC units (like my 14.5GH/s from two BFL units which pull less than 50W together) are far more efficient than the quoted 650W per gigahash, by up to two orders of magnitude, which would push the average for the whole network far below break-even.

KNC units are pretty close to 1W per gigahash.

Exactly my point. This $15M figure is completely wrong.

Ok, so $3.6M is our upper limit in terms of electricity costs... and I'd wager the electricity costs are close to the total payout. That's always been the trend in mining; the reward normalizes to the cost.

That's always been my frustration with the bitcoin network. It is a perfect zero sum game, so improvements in hashrate and increases in power consumption are not actually productive, just part of an arms race. Whoever consumes more electricity wins, assuming the miner operates in the black by some margin.

Edit: That said, the explosion in the BTC/USD exchange rate has probably thrown that balance way out of whack for the moment.

You have a good point, over the long run the payout and cost should reach equilibrium because of the difficulty adjustment. However, that's assuming the mining hardware is commoditized and widely available (everyone has the most energy-efficient hardware), and that the difficulty adjustment happens instantaneously rather than every week or two. I bet that there's some room for arbitrage in both of those assumptions...

Not quite a hard upper limit – I suspect there's a fair bit of mining going on using electricity that isn't being paid for by the person running the rig... (At the very least that single ~2.5W USB ASIC I can see across the office here…)

I don't know why blockchain.info keeps that statistic up even though it's obviously very wrong.

At least it is just electricity. Nothing like the ten's of thousands of lives that were lost in the initial gold (and silver) mining rushes in South America.

Or 12 year old's who work 60 hours a week in a mine to gather Coltan with a pickaxe in the Congo. http://bloodinthemobile.org

On http://realtimebitcoin.info i'm assuming that running 19mh/s is consuming 1 joule. That number is a few months old now and reading on the bitcoin wiki i see that the average is probably closer to 120mhash/j. Updating it now...


"Electricity consumption is estimated based on power consumption of 650 Watts per gigahash..."

My BFL Jalapeño does 7.6 GH with ~70W, plus a couple more for the RaspberryPi. That's two orders of magnitude lower than these reported figures, and massive scale ASICs are even more power efficient.

Seems nobody does investigative reporting anymore. In fact, not even "common-sense-math reporting..

May I take this opportunity to introduce you to a paper that me & my colleague from UCL have recently published.

Part II at http://arxiv.org/abs/1310.7935 - "The Unreasonable Fundamental Incertitudes Behind Bitcoin Mining"

We have tried to address the same issue of the excessive electricity bills associated with Bitcoin mining. What we have outlined in our paper are ways to optimise the underlying SHA256 hashing algorithm so as to mine Bitcoins faster & in a more efficient manner. We have managed to achieve a 38% improvement against traditional mining techniques.You can imagine the impact as it could now potentially allow miners to save millions on electricity bills.

Betteridge's Law seems to apply here. http://en.wikipedia.org/wiki/Betteridge's_law_of_headlines

Bitcoin mining could use any amount of electricity given that the block difficulty adjusts to the prevailing hashrate.

If more people pile in to mining then the difficulty increases and total $$$ spent on electricity will increase. Conversely if mining hashrate declines then difficulty is decreased to compensate.

I'm not sure how the difficulty adjustment would cope with a severe and abrupt change in mining hashrate (take it to the extreme and imagine the hashrate drops to 1/1000th of what it is now, there are going to be a lot of unconfirmed transactions out there).

Is more or less electricity being used now that the serious mining is done by ASICs / fpgas and not GPU clusters?

Interesting question.

For the same hashrate then less electricity is going to be used given that ASIC mining is far more efficient that GPU mining.

But the overall hashrate is still on an ever upward trend.

I guess it depends on how many people have quit CPU and GPU mining altogether, or whether there are still lots of forgotten machines still burning away contributing very little (in comparison).

The cost of electricity used by miners should never exceed the miners' revenue (assuming most miners only mine when it's profitable to do so), which is currently fixed at approximately 3600 bitcoins per day, or $3.6M at the present cost of $1000 per bitcoin. Factor in the all the expensive ASICs miners have to purchase and it's likely a lot less than that.

It costs money to make any other currency as well, and process the transactions, right? So why is it such a big deal?

But the miners are also enabling and protecting the bitcoin network so that transactions can take place.

As such, how much does it cost to run the US military everyday to back up the US dollar? Then comparing the value of actual commercial transactions that take place, I wonder what is more efficient? I haven't done the math.

Why would the military be needed to back the US dollar? What's backing the Hong Kong dollar, the Canadian dollar, or the Brazilian real?

Their armies and alliances. Would be interesting if altcoin networks engaged in attacks on each other, haha.

Your argument (or the sense I can make of it) works on the assumption that in the absence of an army any country will be automatically conquered and the money repudiated. That assumption is undermined by the growth of trade, but is not totally unwarranted.

What's backing the Somali shilling? It's not even approved by any governing power in Somalia.

If I buy Somali shillings (S), and I make a forex transaction, then the other party has my $ and I have their S. However, I can really only spend those S in Somalia, but, I still don't want to buy more S than what it costs for some other country (including my own) to invade it and make those S worthless by replacing or nullifying that currency in that country. (in fact with cryptocurrency, such an exercise is not unfeasible) So the bigger the army or more alliances it has, the more expensive it is to invade, and the more S I can buy, knowing it is protected, and, the more it can appreciate in relation to other currencies. Likewise a country's army should grow in relation to its economy so it becomes slightly uneconomical to invade (unless the country is a belligerent one.)

It's not a pretty thought exercise, but theoretically, large holders of a cryptocurrency can devalue it by flooding the market with them. That makes them targets for people or groups that might want to do that, and targets for other groups who want to protect them for those groups. (however in that respect, flooding may be counterproductive if it helps with the spread and awareness of a currency, in fact if I wanted to ensure the success of a crypto - I'd do that - get them to as many people as efficiently as possible... reverse-hoarding, rather than think solely about appreciation for my own benefit) It's not just about stealing crypto, it's about debasing one, so people choose another altcoin that is more reliable or appreciative (well that's the intention anyway.) A good film plot. Likewise if I am a large Bitcoin holder, and also a large Litecoin holder, I might want to dump, or constantly flood Litecoin, so that Bitcoin looks like a better currency for hoarding, or accepting.

Also, if one crypto network can attack another one, then it'd look bad if it did (really bad), but at least I know it can win such battles. As a thought exercise, this may be one reason why Litecoin may end up China centric - it has an army to back it up (if the online battle was from another country-centric crypto) - and lots of Chinese holders that want to see it appreciate and used.

I don't understand your first paragraph. Could you provide a more specific or particularized answer to "what's backing the Somali shilling"?

As a side point, the shilling appreciating in relation to other currencies would have to be driven by those other currencies depreciating against the world, or by printing press productivity falling dramatically; there is no authority behind the shilling, and they are printed by anyone who wants to print their own, so their value in trade is generally the cost of printing them (quite low).

I have no idea about Somalia and their money-printing operations. I also have no desire to purchase anything from that country, or visit it. I certainly wouldn't want to buy their currency as an investment if it can be printed by anyone, or some army can come in and make those notes illegal, or the country has no economic development to make the demand for those notes increase over time. So what's backing that currency is a number of things, with "backing" being a synonym for "making it valuable or useful in the long term to buy things from and in Somalia." Having a country be physically protected from invasion being a large factor in that equation. Also, not only an army, but more importantly a naval force that can protect trade routes from pirates (doesn't Somalia have its own fleet of pirate ships?? lol.)

The other thing to evaluate in a currency is the amount of it held by foreign countries. If all of a sudden China were to dump its US dollars, I'm sure the value would go down (but what would they buy instead?) Yet at the same time, China has an interest seeing the US survive and/or thrive to a certain extent so those dollars remain valuable and/or useful.

In fact, it seems the Chinese are going crazy about Bitcoins and Litecoins and it makes more sense for many to hold them than US dollars.. possibly because it allows them to make their funds available offshore and also pay bills locally.

As a side note, I've heard the Chinese have official printing plates for US dollars... another indicator about how important ubiquity is, in contrast to artificial scarcity to back a currency. If there's a lesson to be learnt about bitcoin, I think it's that artifical scarcity can also alienate a lot of people (even if it motivates bitcoin holders) and drive up the value of other crypto currencies, and if it was distributed more evenly from the outset, if today's reality would be any different. So my prediction is that there'll be a number of competing and complimentaty crypto currencies emerging supported by groups that were early adopters (or buyers) of them. If I were a fortune 100 company, it might make more sense to buy up say Litecoins on the cheap now, and then take payments in them later. (a bit off-topic, but useful to think about).. why give a free-ride to bitcoin holders? (unless the top execs are holding them.. or more likely, they'll be buying Litecoins now, esp Chinese ones)

This is one of the reasons why Proof of Stake was proposed as a more efficient alternative to Proof of Work while keeping the incentive for miners to keep their rigs validating transactions.

Other cryptocurrencies such as Peercoin and Novacoin use both systems, but will shift to PoS only at some point in the future.

What is important is that a society has reached a level of sophistication where discussions about relative costs of handling different money can and are possible

it's an impressive leap from cutting doubloons into 8 pieces

How does the cost of ASIC mining compare to the cost of fiat currency? What about compared to the quantitative easing?

Does anyone consider bitcoin to be a waste of thousands of people's time and effort?

Bitcoin might fail but it's not a waste. It's inspired a whole wave of alternate crypto-currencies and gotten the public excited in general for them. And wallet and exchange service and a general ecosystem of services have sprung up that could (and do) support any virtual currency. That's all value that can't be lost at this point.

As far as wasting people's time and effort, my hunch is that most people are devoting their hobby time to this and in the process learned a little bit about economics and computer hardware. That time would have probably been spent on computer gaming or aimless web surfing anyhow.

Bitcoin may end up a big flop[1] in the end (who knows) but without trying it there's no way to know, so expending all of this time and effort isn't wasted.

Similar to scientific experimentation; the outcome, regardless of success or failure, is useful in itself.

1. For the majority, there are obviously some people for whom it has been a roaring (financial) success.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact