The electrical cost and impact of crypto-currencies
New Post has been published on https://stockcointalk.com/the-electrical-cost-of-crypto-currencies/
The electrical cost and impact of crypto-currencies
The operation of the main crypto-currencies causes a considerable electrical expense. What consequences for their future?
In 2013 the value of a bitcoin mentioned was 100 euros corresponding to about the average price of September 2013. Today, a bitcoin is worth more than 5000 euros. The value of bitcoin has been multiplied by more than 50. It is doing well!
We will not discuss whether this is a speculative bubble or valuation with a real foundation because opinions are divided on the issue. Two camps oppose each other diametrically. Almost as many as in 2013, there are economists and currency specialists on the one hand, who consider it a “fraud” (a term recently used by Jamie Dimon, the managing director of Investment JP Morgan Chase, here) and that everything will collapse: the price will fall sharply and perhaps become zero. On the other hand, the defenders of cryptographic currencies offer various arguments to persuade us that the value of bitcoin and its sisters is justified. For them, the characteristics of cryptographic currencies – anonymity, irreversibility, decentralization, openness to everyone, pre-set issuance and speed of transactions – make them new currencies without equivalents, which are useful for a greater fluidity of money movements. on the network, and for a better overall economic operation. It will probably take a while before we know for sure which side is right.
The electrical cost of the network
One point is however seriously worrying: the computer network that allows the operation of bitcoin exchanges and their security burns a significant amount of electrical energy. It is evaluated by the specialized Digiconomist site at 23 TWh per year, which corresponds to the annual electrical expenditure of more than 2 million American households, or 0.10% of all the world’s electricity production.
This amount of energy burned by the network is soon to grow. A simple economic reasoning that we will detail below shows that the electrical expenditure of the network is proportional to the course of the bitcoin with an adjustment period of several months for the catch-up investments to take place when the price increases (a kind of inertia). As bitcoin’s price has recently risen a lot – it has increased by more than 8 in a year – the current-period consumption adjustment has not fully taken place and will be done by multiplying the electricity consumption of the current network. by at least two in the next few months. This is a certainty … unless the price of bitcoin falls.
This will lead to a consumption of the bitcoin computer network of at least 40 TWh per year, equivalent to that of a country like Peru or New Zealand or to almost 10% of the French electricity expenditure which was 468 TWh in 2015.
Today bitcoin spends 23 TWh. (Digiconomist).
In 2015, Switzerland spent 62.1 TWh, Portugal 49.8 TWh, France 468.4, Peru 42.9, New Zealand 41.4. [International Energy Agency Energy Statistics 2017 page 60 and following]
Optimists and pessimists
The estimates I mentioned are approximate because it is impossible to know in detail who is spending electricity and how much for the extraction of new bitcoins. See Appendix 2 below which explains the difficulties of this evaluation.
Two camps compete for this evaluation. On the one hand, there may be a little bit of a pessimistic side to the high numbers I mention. Digiconomist website is the most serious representative of this camp which expresses a concern and finally distrust vis-à-vis the cryptocurrencies whose ecological footprint seems unreasonable. Another more optimistic camp arrives at roughly half the size. His most accurate representative is Marc Bevand. He may be interested in demonstrating that this power consumption is not serious because he has had links with the industry that produces specialized chips for this extraction of bitcoins, and is openly an “enthusiastic cryptocurrency”. Anyway and whatever the good numbers, it is certain that there is a significant expenditure of electricity, which, even considering the optimists’ evaluations, establishes that the operation of the Bitcoin network will soon burn equivalent to at least 5% of France’s electricity expenditure, and this expense will increase if the interest in bitcoin crypto-currencies is confirmed and their prices go up! Taking into account the other bitcoin-like cryptographic currencies, it is necessary to double the optimistic evaluation and therefore it is at least the equivalent of 10% of the French production that the cryptographic currencies will consume … while waiting much more.
The reasoning that proves the inevitable expense
To encourage actors to participate in the management and monitoring of the bitcoin network – this allows its operation without central control authority – a system of remuneration was planned from its conception in 2008 by the mysterious Satoshi Nakamoto. The remuneration of the nodes of the network that make it work is made by assigning them new bitcoins created periodically according to a fixed program once and for all: 12.5 bitcoins are issued every 10 minutes. These new bitcoins as well as commissions related to transactions (about 1 bitcoin currently) that are added to the 12.5 bitcoins are not distributed among all the nodes but attributed to a single node of the network following a competition between them . The competition consists in solving a problem of mathematical nature. The problem is that it is more likely to solve it first – so to win the 12.5 bitcoins issued and commissions related to transactions – that we can quickly calculate a function called SHA256. Those who participate in this competition are referred to as “minors”, by analogy with the miners in a gold mine. Of course, they want to increase their probability of winning and have therefore been caught in a race, each trying to have an ability to calculate the SHA256 function as large as possible, representing a percentage as large as possible of the total capacity of the network, since it is this percentage that fixes their income.
At first the calculations of the competition were made using common machines and even on desktops or binder. Some participants quickly realized that graphics cards were more efficient, that is, spent less power to calculate SHA256. So they used them massively. Rapidly, a second step was taken in designing and manufacturing application-specific integrated circuit (ASIC) chips that calculate the SHA256 function and do nothing else. Today, those involved in the SHA256 computing race are competitive only by using such ASIC chips that are manufactured in the millions and are perfected from year to year. The mining of bitcoins has become an industry, 80% of which is located in China.
The total power of the network measured by its capacity to calculate SHA256 is today colossal. Ten billion of billion SHA256 calculations are done every second or so (see here). The electrical cost of this operation is, of course, important and in the medium term – that is to say in a few months – it is equalized with a certain percentage of the value of the bitcoins issued and the commissions associated with the transactions.
The rationale for this statement is simple: when the costs of the mining systems are higher than what they report, they stop using them; when you know how to do it, the operating cost of which is lower than they earn, new “bitcoin mines” are created because there is money to be made. The logic of this system is comparable to that of gold mining: when the price of gold falls, some gold mines are no longer profitable, they are closed; when prices rise, unexploited deposits become interesting and new mines are opened, while some mines that have been closed are put back into operation. These adjustments are not immediate, but in a few months, the relentless economic logic of profit-seeking produces a new equilibrium between the cost of extraction and the gains from it.
The cost of electricity is not the only cost of running the SHA256 function calculation because it is necessary to buy specialized chips, set up the bitcoin mines which are today real factories composed of several buildings and employing workers and technicians by the tens. It is estimated that electricity consumption represents a fairly stable percentage of the operating and depreciation cost of these digital mines in the order of 50%. And so that the electricity spent is after equalization between cost and gain (as for gold mines) of the order of 50% of what bitcoins created and commissions.
Optimists and pessimists argue over this difficult percentage. It is also the disagreement on its value that essentially explains the contradictory figures obtained between the two camps: the optimists use the value 30% or less, the pessimists use 60%, sometimes more. In the end, between the optimists and the pessimists, we, therefore, have a result that varies from simple to double, as we have already indicated.
But whatever the value, it follows from this implacable economic logic that the higher the bitcoin price, the more electricity is spent by those who want to own the bitcoins issued and commissions associated with transactions. For unavoidable economic reasons, the electrical expenditure of the bitcoin network is therefore proportional to the bitcoin price. If bitcoin gains value, the electrical cost of running its network increases proportionally in the months that follow. Similarly, if it goes down, the electric cost goes down.
Precision to be complete that this reasoning is valid only during periods when the emission of new bitcoins is stable. However, the bitcoins issue protocol, set in 2008, stipulated that this issue should be halved every four years. It was halved on July 9, 2016 (from 25 to 12.5), it will be halved in 2020 and will increase from 12.5 to 6.25 bitcoins per 10 minutes. Therefore, the long-term forecast of the grid’s electrical cost should be divided into two parts of the electrical cost once every four years. However, it should also include in this calculation the inclusion of commissions associated with transactions that are added to the bitcoins issued by the protocol and also contribute to the remuneration of minors. These commissions evolve in a complex way, but basically they increase along with the value of bitcoin. They have been introduced by Nakamoto to make up for the doubling every four years and to keep people interested in monitoring and running the network. It can be assumed that, as a first approximation, what Nakamoto predicts will happen, which means that in future electricity cost projections, the doubling every four years can be ignored, offset by the commission system. In the end, we must consider that in order of magnitude, 50% of the miners’ earnings are spent annually on electricity and that this represents and will continue to represent 50% of the value of bitcoins issued annually today, that is, ie 657,000 bitcoins.
Direct control of the calculations mentioned above
Assume a cost of $ 0.05 per KWh (used by Digiconomist) and a $ 5000 per bitcoin price. This gives 657000 * $ 5000 = $ 3,285,000,000 per year of compensation for minors. It corresponds to them the following quantities of electricity according to whether one is pessimistic, optimistic or between the two:
– With 60% [(657000 * 5000) / (0.05)] * 0.30 = 39.42 TWh per year
– With 30% [(657000 * 5000) / (0.05)] * 0.30 = 19.71 TWh per year
– with 50% [(657000 * 5000) / (0.05)] * 0.50 = 32.85 TWh per year
This calculation is very close to the 40 TWh per year cited above as the current annual electricity cost of bitcoin in its pessimistic version and the 20 TWh of the optimistic version.
Note that Marc Bevand in his blog assumes that the average cost of KWh is $ 0.1, so twice as high as in the calculations made here. Today it is 0.125 € (= 0.147 $) in France during off-peak hours.
The calculations made above therefore assume that miners have advantageous tariffs for electricity (1/3 of the French price), which is probably the case because bitcoin mines are established where the price of the electricity is the lowest. We also understand that Bevand’s choice of assuming a high price for electricity leads him mechanically, following the reasoning proposed, to an optimistic (ie, low) estimate of the electrical expenditure measured in TWh per year.
Impossible future
Let’s come to the anticipation of what could happen if the goals of the bitcoin defenders were achieved: to match the dollar or the euro.
Today the value of all the bitcoins in circulation is about 100 billion dollars. It’s a form of digital cash. It is therefore not absurd to compare this value with the value of all outstanding dollar notes that according to the EDF is $ 1500 billion (December 2016 figures). For the euro, we have figures of the same order of magnitude. There are 15 times more dollars in the form of tickets that circulate than dollars in the form of bitcoins. If the volume of bitcoins in circulation becomes equivalent to the dollars circulating in the form of notes, it would be necessary that its price is multiplied by 15. Indeed bitcoins issued represents 80% of bitcoins to be issued when the 21 million provided by Nakamoto have been issued. Only by increasing their unit value of bitcoin can their total approximation in value of the total dollar bills in circulation. Such a multiplication by 15 of the value of the bitcoins (or by 13 if we want to take into account the 20% of bitcoins not issued) is in principle quite possible since a higher increase of the price has just occurred in the last 2 years.
This multiplication by 15 would lead the power consumption of the bitcoins network to 300 TWh per year for the optimists and 600 TWh per year for the pessimists, that is to say in order of magnitude to the annual French electricity consumption.
Let us insist on the fact that the link between the increase in electricity expenditure and the price of bitcoin is of an economic nature and therefore almost automatic if we do not fundamentally change the operating protocol of cryptographic money. Nothing can stop it without a determined will, either of the community that has the collective power but which is not the interest, either States by imposing a control or prohibiting this type of diabolical digital and economic mechanism.
Unclear but certainly serious!
The figures quoted correspond to fairly imprecise evaluations but the orders of magnitude are good and of course, give the dizziness … including when it comes to optimistic figures. They simply show that it is not seriously conceivable that one day the bitcoins in circulation or more generally the cryptographic currencies running on its model could replace the international exchange currencies that are the dollar and the euro.
Note also that we have only considered an increase in the price of bitcoin so that their total equals dollar banknotes what economists call the “monetary base” and note M0. If we had taken into account the value of the money supply M1 which also includes the money present on the current accounts of individuals and companies (M1 is about five times more than M0) what perhaps should have been done for to be able to really say that bitcoin becomes the equal of the dollar, we would have arrived at an electrical expense five times larger, so still much higher than we found. The figures obtained would then be between 1500 TWh and 3000 TWh per year, reaching in order of magnitude half of the annual electricity consumption of the United States (4128 TWh for the USA in 2015).
A shot on the right, a shot on the left
Let us mention two more remarks, one in favor of the pessimistic camp, the other in favor of the optimistic camp.
When we talk about bitcoins in circulation, we estimate them at around 16 million and we say that there will never be more than 21 million because it is in the starting protocol that we will not change no, we forget that some bitcoins are lost because they are on accounts whose keys have been forgotten. Except when breaking the signature system used by bitcoin (ECDSA, based on elliptic curves), these lost bitcoins are definitely: nobody, never will be able to use them anymore, they are exactly like hundred dollar bills that would have been thrown away. in a fire. It is impossible to estimate precisely how many bitcoins have been destroyed, but they may be quite numerous because at the beginning when bitcoin was worthless, we were not very attentive and we did not care to lose. We also lost because of hard disk failures, old machines thrown away without attention, clumsiness in the handling of the keys. If, for example, 1/3 of the bitcoins emitted have been lost, it is necessary to reach M0 or M1 with the value of the remaining ones, resume the evaluations of the necessary electricity and increase them by 50%. I let you do the calculations.
And now a note in favor of the optimists. We have perhaps been a little simplistic with the issuing division by a factor of two every four years (called “halving”). We assumed that the commissions associated with the transactions made up for these losses of income for the miners, which nobody today can really be certain. If this is not the case and the commissions become negligible (extreme assumption in favor of the optimists) then the electric energy necessary to reach M0 or M1 is weaker than that calculated and now depends on the date when one envisages achieve these goals. If we want to achieve them in 2 years no change (because the next halving is in more than two years), if we want to reach them in 6 years, we must operate a division by two consumptions announced, if we want to achieve them in 10 years, it must operate a division by 4 announced results, and so on with couples:
– 6 years / factor 2 – 10 years / factor 4 – 14 years / factor 8 – 18 years / factor 16 – 22 years / factor 16 – 26 years old factor 32 – 30 years factor 64 …
It may be possible to achieve acceptable electricity consumption for decades, but is it really reasonable to believe that the commissions will remain negligible since in the long run, it is they and they alone who ensure the remuneration of minors?
This optimistic vision is not an option for Ethereum since the emission system of ethers is constant (without ever dividing by two). And this difference between bitcoins and ethers is a new source of concern for bitcoins since when the revenue becomes more interesting in ethers than in bitcoin, miners risk giving up bitcoins for ethers and the bitcoin network to be seriously weakened.
The incentive scheme at issue
Let us now ask the question of whether we can change the way cryptographic currencies work to avoid this crazy electrical expense and still allow cryptographic currencies to replace national or international currencies one day.
The bitcoin’s electrical expense is not due to the costs of monitoring transactions and managing its account register (the blockchain). It is due to the mode of distribution of bitcoins issued every 10 minutes, and more precisely to the competition between nodes of the network which designates every 10 minutes the winner of the bitcoins issued and commissions associated with transactions. Other mechanisms based on the idea of blockchains do not use incentive systems and even less competition between the nodes of the associated networks because they do not need them. What is called “private blockchain” which serves, for example, to manage exchanges between a consortium of banks in fixed number are not subject at all to the crazy electrical expenses of bitcoin; they do not need incentive systems because each actor in the network is interested in advance in the proper functioning of the network. Only the willingness to base the operation of bitcoin on a peer-to-peer, anonymous, decentralized, open and extensible network requires the establishment of (a) an incentive system that ensures the existence of volunteers involved in managing and monitoring the network, and (b) a competition between nodes to attribute the incentive. The real question concerning the electrical expense of the Bitcoin network and the networks copying it is that of the competition between nodes of the network for the allocation of the incentive bonus. Can we imagine another type of competition that does not involve the electrical madness of bitcoin?
The question is not new and many ideas have been proposed to otherwise effect the distribution of the incentive. We will not discuss them here in detail because it is enough to know that no replacement system today has been able to fully convince. A system called “proof of stake” has been presented as likely to replace the proofs of work of bitcoin. The Ethereum network, which is the largest cryptographic currency network behind Bitcoin, was planning to replace the evidence of work it uses today with evidence of stakes. From month to month, the planned Ethereum transition from one system to another has been postponed. Only a partial replacement is now envisaged and for the moment pending. Today, everyone is wondering if the choices made for the distribution system of bitcoins and the most important cryptographic currencies is really avoidable.
The explanation could be related to the concept of computational content (see What is a complex object?). By completing their calculations to win the winner’s contest every 10 minutes, the nodes in the network are spending electricity; the calculation produces the solution to a problem (uninteresting but difficult) that is put in the register of accounts of bitcoin. This deposit has the consequence that the register is difficult to falsify. To falsify it and create a register that can replace the current register (and to divert bitcoins from one account to another), it would be necessary to calculate as much as was calculated by the entire network, and therefore spend as much electrical energy than what was spent by the entire network, at least over the period one wishes to falsify. It is impossible or extremely expensive. The robustness of bitcoin seems essentially linked to this almost absolute impossibility of making a false register, an impossibility that no longer occurs when one gets rid of the proofs of work. Finally, to look closely, it seems that the mode of distribution by competition and proof of work of the bitcoin network could be irreplaceable if one wants to guarantee the overall strength of the protocol.
To blame too much, we risk losing everything
In conclusion, the idea of the blockchain is, without a doubt, very good when it is implemented by renouncing certain characteristics of the bitcoin blockchain and for example by adopting the idea of a private blockchain that does not There is no need for an incentive method. It is the desire to have a system protected by putting a colossal amount of calculations in the register of accounts (and therefore electricity) to make it unfalsifiable that is at the root of the problem. Unfortunately, this almost perfect solidity has a price that is the impossibility of price growth allowing bitcoin to one day become a true competitor of the dollar or the euro.
Between the simple and electrically viable private blockchain systems, and the public blockchain systems, totally decentralized, open, anonymous, and having an extensible configuration of the nodes which seem absurd and finally condemned in advance, it is necessary to choose or invent intermediate systems. To give birth to a new international currency, it will be necessary to give up certain properties of the currency set in motion by the brilliant Nakamoto.
