51 thoughts on “Bitcoin: a waste of energy

  1. I had no idea that crunching numbers would consume this much energy. That must be a lot of CPU cycles.

  2. I’m not a computer engineer, but I think you will find that the specialised chips include CPUs.

    And, obviously, you have to crunch an awful lot of numbers to burn up that much energy: that’s designed into the system.

  3. @Trevor
    Yeah, me too. I know large computers generate lots of heat and require large air conditioning units to assist with temp control which is a bit like heating an oven to 60 deg and using an aircon to maintain it to 40deg. I’ve heard some graphics cards for typical home games computers can chew up something in the ballpark of 500Watts and even northwards of that. But I would have thought that typical modern day electronic currency and financial transactions and recording would also consume a significant fraction of the energy Bitcoins use anyway?
    I can’t see this point resonating with many people in general as an argument against Bitcoins and I’m sure there are plenty of valid concerns regarding the currency, but I’m certainly open to be convinced.

  4. A nice and very stimulating article.

    I’m surprised you didnt manage to work other problematic quasi currencies into your article especially given its Floriade season.

    You should be proud (I am serious) of all the hysterical reactions you generated. Much better than being ignored or preaching to the quasi-converted/fellow travellers alone (e.g. on this blog site).

    One very minor rejoinder. Bitcoin and the responses do once again raise the problem of
    a. what ‘value’ is exactly,
    b. the strange interaction between accumulation of value, and the ideas of wealth and debt especially when you get madness such as the (endless) Greek impossible trap
    c. how people are ‘entitled’ to a precise wage obsessively managed to the nth degree
    d. how we assign value to anything to the nth decimal place when values are so rubbery,

    We take for granted we understand value and hence bitcoins but any delving reveals the confusion/misunderstandings still largely underlying this outgrowth of human social interactions which those with smarts exploit to their benefit and too often someone else’s loss.

    John does your new book tackle this mess – perhaps as a primer for dummies like me – given earlier posts indicating it looks at the crazy sources of property rights claims which seems intimately linked to how we value things?

    I would really appreciate this as it might help to explain how we can justify how we exchange and value stuff – not just the obvious of solid relatively immutable material substances but also:

    – transient material – food being a classic which has different values depending where it is, how old it is, how well its maintained, how fashionable it is, how hungry we are, how much we have been brainwashed by MacDonalds adds, its temperature (MacDonalds chips are remarkably vile once cooled.

    – memeplexes which can be patented such as designs

    – how stuff loses value and then gains value as it becomes rarer or more obsessed over by people with lots of loot (Antiques Roadshow illustrates the confusion here in respect to how some art which is actually easy to reproduce becomes invaluable – original paintings – whereas other crafted art which is less easy to reproduce – exquisite glass – has relatively little value.

    – memeplexes which arent or cant be patented in a cost effective way like house designs – or which the creators dont want to lock up because its counter productive – the vast body of academic knowledge in journals, or philosophy texts

    – “natural resources” which noone created but which can be claimed by virtue of prospecting rights.

    – irreplacable ecosystems whose value in toto is infinite but which economists want to internalise within a system whose mechanisms they dont really seem to have quite grasped either.

    – trademarking of names by people smart enough to see an openning for rentier profits now exploited by others who would think the idea daft – like ugg boots.

    Overall our normal valuation system, aside from the fact that it works after a fashion, seems even crazier a system than bitcoin so maybe there something more to be said here.

  5. @John Quiggin
    How long a laptop takes to generate a coin is irrelevant because nobody does it that way anymore. It’s done using highly specialized, and efficient AISC chips.

  6. @Chris
    and surely the energy input isn’t anything close to the inverse ratio of speed it takes for Laptop vs AISC chips (unless it takes quite a few of them) as the heat generated would be virtually impossible to sink out.

  7. Troy Prideaux :
    and surely the energy input isn’t anything close to the inverse ratio of speed it takes for Laptop vs AISC chips (unless it takes quite a few of them) as the heat generated would be virtually impossible to sink out.

    The best ASIC miner has 4000 Mhash/J. The best laptop is about 3 Mhash/J. The ASIC uses over 100 times less energy!

  8. The payment system part inside bitcoin – ie. moving value from party A to party B without requiring a trusted intermediary and the associated blockchain ledger is pretty neat. Lots of banks / financial companies are busy co-opting that technology outside bitcoin. To actually run that system takes a fairly trivial amount of computing power. In the beginning the power required to run the bitcoin network was just incidental – just run the software in the background on your laptop and it would add trivial amounts of power usage.

    BUT what happened was when the price of bitcoin shot up (from a few cents to hundreds then thousands of dollars), people started throwing computing power at the system to try and mine more and more bitcoin, or at least increase their personal market share of the new bitcoins created by the system. The system generates a fairly fixed amount of new bitcoins every 10 minutes or so, as a substitute for inflation/central bank issuing, and as a reward for verifying the transactions on the network. The new bitcoin reward is meant to be the incentive for people contributing computing power to the verification process, so there is computing power available to actually run the network It’s a little bit like Lotto though – the bitcoin reward for verifying the last 10 minutes or so of transactions only goes to the computer that solves the crypto problem – everyone else trying to solve it gets nothing. So if you throw MORE computing power at it, you essentially buy more tickets to be the winner of the bitcoin lotto that 10 minutes.

    However, there is a protocol inside bitcoin that adjusts the difficulty of the cryptographic calculations in response to the amount of computing power thrown at it – ie the system adjusts so it cant be “forced” to make new bitcoins faster just because someone threw a supercomputer at it. (again the ACTUAL computing power needed to effectively run the network is pretty low)

    So we arrive at the situation John describes where the computing power required to “verify” the network transactions is insane, because of the ongoing computing arms race as people throw more power at the system to get the bitcoin rewards and the amount of computing power required keeps increasing in response, and is now insanely in excess of the actual need to simply run the transactions part of the network.

    The currency part of bitcoin is dumb.

  9. @John Quiggin

    Sort of – Bitcoin has the concept of “difficulty” built into it – it makes things harder as more computing power is devoted to the system. Remember all the computing power is only there to verify the transactions happening on the network – Party A sending value to Party B.

    Difficulty self adjusts so that you can’t throw a supercomputer at the system and get all the new bitcoins. Difficulty has gone through the roof since the price went up:

    http://www.coindesk.com/data/bitcoin-mining-difficulty-time/ (change the graph to “all”)


    Bitcoin is a strange beast because it is a very promising payment system (the blockchain) stapled together with a very dumb currency. (plus the whole system is stuck in a libertarian community which doesn’t realise this)

  10. @Chris
    Chris, the time required for one laptop to mine a bitcoin is relevant, as it demonstrates in a form that simple folk can understand how much computing power is required, even if no one actually does it that way.

    It’s similar to how I could say that the Saturn V rocket used for the first Apollo landing on the moon soft landed 154 times as much mass as the Semyorka rocket used for the first lunar probe soft landing just three and a half years earlier. However, despite the comparison being hopefully interesting, no one would ever use 154 Semyorka rockets to send a crewed mission to the moon. For one thing, the astronauts would have to be cut into pieces.

  11. I don’t know John I think your missing some of the key differences between bitcoins and paper money.

    Would it change your perspective if all bitcoin generation was done from renewable power sources? Have you looked at the size of Facebook’s servers, power and water consumption lately.. Facebook have a huge investment in renewable power but at the end of the day it’s all for cat pictures and advertising… What do you think is worse for the environment cat pictures on the internet or bitcoin generation?

    While it’s good your thinking about power consumption I think you should be more worried about Australia’s new coal mines and the environmental effects they will cause http://www.greenpeace.org/australia/Global/australia/assets/The%20Critical%20List.pdf

  12. @John Quiggin
    People apparently use blocks of GPU cards now, which go thousands of times faster than standard chips if programmed correctly. Whilst these are fairly greedy energy wise — if you have solar panels, then using the extra energy in sunny times might be worthwhile given the price to sell back to the grid is getting so low.

  13. @Shane

    That’s an interesting comment. I mean where you write;

    “Bitcoin is a strange beast because it is a very promising payment system (the blockchain) stapled together with a very dumb currency. (plus the whole system is stuck in a libertarian community which doesn’t realise this)”

    This roughly accords with my (very untutored) perception of Bitcoin. Having said that, does Ethereum show any promise of using the smart idea part and jettisoning or fixing the dumb idea part of Bitcoin? Ethereum as I understand it is using block chains for contract making.

    “Ethereum is a cryptocurrency platform and Turing-complete programming framework intended to allow a network of peers to administer their own stateful user-created smart contracts in the absence of central authority. It features a blockchain-based virtual machine that securely records and incentivizes the validation of transactions, i.e. code executions, made through a cryptocurrency called Ether. This “crypto-fuel” or digital asset is what is used to pay for the decentralized computational instructions hosted in the smart contracts deployed on Ethereum’s blockchain.” Wikipedia.

    It is unclear to me how Ethers (the currency unit) are issued. This below does not really explain it.

    The currency unit of Ethereum is the Ether, used to pay for computational services on the network.

    “To finance development, Ethereum distributed the initial allocation of Ethers via a 42-day public crowdsale, netting 31,591 bitcoins,[12] worth $18,439,086 at that time, in exchange for about 60,102,216 Ethers.

    Ether is divided into smaller units of currency called finney, szabo, shannon, babbage, lovelace, and wei (named after Wei Dai, the creator of b-money). Each larger unit is equal to 1000 of the next lower unit.[13] In practice, however, the developers encourage the use of ether and wei. Wei is the base unit of implementation and can not be further divided.” – Wikipedia.

    Disclosure: I have not dabbled in Bitcoin or Ethers in any way. If I don’t understand investments I don’t make them… and that’s not mentioning the several other reasons which mean I rarely invest… including the fact that I have little spare money and even less that I feel like risking.

  14. @John Quiggin

    Is Bitcoin mining the most worrisome waste of energy under our current economic system? My guess is that it would come a very long way down the list. How much energy do we waste by having millions of inefficient internal combustion engine automobiles instead of electric-powered mass transit? Why do we strain out a gnat but swallow so many camels?

    How much of our economic consumption is really necessary for humans? For example, one could make a very good case that no healthy adult ever needs to consume any drink but water. Maybe we could add cow’s milk as a food and hydrating electrolyte drinks for temporarily dehydrated people. Why do we need to manufacture any other drink at all? Is it not all a pointless waste of energy? No healthy adult ever actually needs any other kind of drink. Soft drinks, alcoholic drinks, coffee, tea and anything else are all totally superfluous and wasteful. Why have them in our economy? We would mostly be much healthier without such drinks. This is just one example I can pick out of the thousands of irrational things that the “rational agent” homo economicus does and wastes a lot of energy doing. Bitcoin does not stand out as especially irrational or wasteful given the rich field of human economic irrationality and wastefulness.

  15. As Shane says above, the value of bitcoin is in the transactions rather than the currency — although the currency and transactions are symbiotically linked.

    When I give a $10 note to a shopkeeper how much energy do we use before it is a number in their bank account? I have no idea how it compares to verifying a bitcoin transaction, but it isn’t zero.

  16. Ikonoclast :
    Interesting paper;

    Indeed. I’ve learnt a lot about Bitcoin in this thread and I can see the energy concerns.

  17. Interestingly high frequency trading is going to have similar problems. It isn’t exponential like bit coin, but all of the traders need to be near the trading centre to reduce latency. This means that more and more power needs to be directed to a specific location, and that the consumers of it are willing to pay for that. But ultimately it will reach a physical limit.

  18. Has anybody in the world done reliable, peer-reviewed calculations of how much total energy is used daily by;

    (a) world bitcoin mining operations;
    (b) the world’s gambling operations (every casino, every tote, every on-line gambling site);
    (c) the world’s stock-markets and bank transaction systems);
    (d) brewing, chilling and transporting the world’s alcoholic drinks (as an example of completely unnecessary and resource-wasting activity).

    Comparisons of these sorts would give us an idea of how concerned we really need to be about bitcoin operations in the greater scheme of things.

    My guess is Las Vegas alone probably over-matches world bit-coin mining in energy use.

  19. @conrad

    Unfortunately the capital investment required to use excess solar makes it unlikely to pay off. I did the numbers recently, and even buying second hand mining hardware would take a year to pay off, assuming no other costs (ie, not paying for the electricity). I gave up at the point where the 5c/kWh I get for feeding the grid started to look unrealistic as a return on mining.

    Where it makes sense is a combination of cheap access to the hardware, cheap labour and cheap electricity. Which seems to mean global south countries with subsidised or stolen electricity.

    You can get a 4kW 5Th/s machine for about $AUS4000 delivered (second hand), so you need 5kW of PV (at ~4kWh/kWday) to power it and hope your 18Eh/day gets you more than the $1 of grid feed-in tariff and $1 of time-value-of-money.

    I’m not willing to spend $4000 to find out whether that would actually work out in my favour.

  20. @Moz the Solar Geek
    Also, it’s really hard to find power consumption figures for bitcoin mining gear. It’s almost as though the manufacturers don’t want to say, or perhaps the users don’t care.

  21. @Moz the Solar Geek
    If you did run these “mines” (?) on 100% renewable energy that you own the generation of, then any return is potentially measured against pure capital investment. Of course, there is the likelihood of the requirement for continual upgrades to both the processing capacity and energy generation to power that capacity as the “difficulty factor” (?) rises, but it is an interesting concept with significant risk I suppose… maybe that’s why it’s labelled as “mining” 🙂

  22. @Troy Prideaux John’s point is that Bitcoin mining will not be done with renewable energy — in any existing grid with a mix of renewable sources with more or less fixed capacity, marginal increases in demand are likely to be met from increasing the output of coal or (better) gas power stations.

  23. Troy Prideaux :any return is potentially measured against pure capital investment.

    I thought the lost income from not feeding that power into the grid also had to be taken into account. Bitcoin seems like a classic gamble – maybe I’ll make $1000 this year if I successfully mine a coin, or alternatively I can guarantee $350 by selling the power directly. In reality most miners join collectives where they share a pool of winnings based on how many hashes they contribute, so it’s a slightly simpler question: can I sell my hashes for more than $2/day?

    Note that I’m assuming that I can passively cool the miner when that’s almost certainly not true. I’m also ignoring the difficulty of matching the miner to the instantaneous available power, the opinion of the ATO and a few other things. FWIW I understand the ATO regards the capex as personal spending, and the income as earned rather than won (viz, it’s taxable).

    Per Lomberg, seeing it as an investment in “making the world better” means it should go to Bush Heritage or Doctors Without Borders rather than buying solar panels and bitcoin mining hardware.

  24. @Tom Davies
    I’m not suggesting it will be done with renewable energy. My post was more of a conceptual hypothetical thought bubble:) I can totally appreciate that the reality will be a significant reliance on fossil fuel based power generation.

  25. As I see it, Bitcoinism is essentially a self-serving religion akin scientology so bitcoins are virtual religious supplies. If the Catholic Church was using communion wafers that required a ton of CO2 to produce they would be called to account.

  26. Bitcoin miners make this very clear when they talk between themselves about how to make the most profit from mining it.
    Then they talk openly about how they depend on subsidies to make bitcoin mining worth it. Either they live in one of the few places with cheaper gov-subsidized electric energy or, in most cases, they mooch “free” (oh the irony) electricity from somewhere, their parents, university dorms, etc.
    Bitcoin is not profitable in most countries, if you pay your own energy bill.
    This radical difference in discourse from when they talk between themselves to when they talk to outsiders is pretty telling.

  27. From comments, a lot of bitcoin miners use cheap hydro. But hydro (unlike wind and solar) is effectively in fixed supply, so this just means more coal-fired power used somewhere else.

  28. This bitcoin project could end up as a “real world” demonstration of LtG, especially the over-shoot phenomenon, if I understand the mining correctly.

  29. Donald Oats :
    This bitcoin project could end up as a “real world” demonstration of LtG, especially the over-shoot phenomenon, if I understand the mining correctly.

    The worry is, it’s looking increasingly like we might see a real “real world” demonstration of LtG before a Bitcoin demo.

  30. (Seriously, to complain that BitCoin is not as wasteful as other unrelated activities is just what-aboutery, isn’t it? Bitcoins energy use is an interesting thing to discuss)

  31. @Tom Davies
    Indeed, Tom. If I saw someone about to pee on an electric fence I’d inform them of the fact and not hold back on account of how people could be urinating on higher votages elsewhere.

    And if someone did pee on an electric fence on account of my not warning them, I don’t think they’d be very satisfied with my excuse that I was saving all my “warning energy” for something more serious.

  32. Unless I’m missing something, the BitCarbon website misses an essential point – the number of bitcoins produced every 10 minutes halves every 4 years. 25 bit coins are produced every ten minutes at the moment but this will taper off to about 1 an hour in 30 or so years. If a single bitcoin is worth $1 million by then ( unlikely ) it would mean that they have become extremely useful, as a currency etc. 1$ million per hour to secure an extremely useful network doesn’t seem like much to me.

  33. The Bitcarbon website also makes a rather stupid assertion that:

    ‘For example, if the value of Bitcoin is $1,000 and going up, maybe it is rational to spend $2,000 per Bitcoin in electricity.’

    Why would you spend $2000 to mine a $1000 bitcoin when you could just buy as many as you like at $1000 ?

  34. @Moz the Solar Geek
    There’s no single number as it depends on your hardware and how you run it. I have a top of the line GPU card in my workstation and it is quite greedy and can use around 300w when really chugging through numbers, although I’ve never seen how it goes on bitcoin mining as its a work computer (presumably similar). This is similar to two high end multi-core Xeons, which are enough to heat my office. So the answer is lots of power at present.

  35. @Joe

    “By that metric, all chips are specialised.”

    By what metric, Joe? An FPGA/ASIC does whatever you program/design it to do. It can act as the central processing unit (cpu) of a machine, or it can perform an auxiliary function.

  36. @Joe

    “[…] this will taper off to about 1 an hour in 30 or so years. If a single bitcoin is worth $1 million by then ( unlikely ) it would mean that they have become extremely useful, as a currency etc. 1$ million per hour to secure an extremely useful network doesn’t seem like much to me.”

    It doesn’t seem like much because you’re not including any transaction fees. These are currently averaging about 1 bit coin per hour, or 4% of total revenue to the miners.

    But they’re designed to ramp up in response to (based on a very quick glance at the formula) increase in velocity of the currency, general increase over time in the volume of smaller and smaller purchases, increase in priority/speed with which people require their transactions verified.

    Given the value of all bitcoins in existence in your 2045 scenario is well over $20 trillion – or 1.5 times the total current balance sheet of every bank in the US, or about 300 times the size of Visa and Paypal put together.

    It’s fairly safe to assume the transaction fees in this scenario are going to dwarf the $1 million / hour in mining subsidies you’re referring to.

    That is the true cost of “securing the network”. That is the figure you should be basing CO2 emissions calculations on. (It is also the actual reason why miners will continue to mine long after their overheads outgrow the face value of the subsidies they receive).

    Even if at a conservative minimum they were to simply keep pace with the tapering off of mining subsidies, that would represent 25 bit coins per hour in 2045, as compared to one per hour now in 2015.

    ie. $26 million per hour, or $228 billion per year in total mining revenue.

    And what does mining solely consist of? Consuming electrical energy.

  37. @conrad

    I was talking about power consumption of specific devices as (not) included as part of the description when people are selling them. The link I posted was the first I found that had power listed, and it was at least the tenth I looked at. My point is that the people doing the mining don’t seem to care about power consumption, with the possible caveat that some of them seem to care about how much cooling they need to supply.

    They’re very definitely not sold as “1Th/s at 2215W”, which is what you’d need to feed into a calculation of how much money you’d make (or lose) when mining.

    I do wonder whether blockchain can be made to work environmentally, because it seems at my superficial level that inherent to the idea is that the hashes are ridiculously expensive to generate. I recall hearing that if a collusive group can produce 50% of the hashes they can corrupt/re-write/take over the blockchain they’re targeting. If true, that means that to be secure you need to be burning a lot of energy, otherwise you’re vulnerable to anyone willing to burn more.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s