Unit 5 Lecture Notes

The amount of energy that Bitcoin consumes can theoretically be estimated from its hashrate, or direct output of the energy consumption of the machines that secure the network.Since there is a wide variety of equipment performing bitcoin mining and no central authority or registry to count the miners, there is no easy way of observing how many of each type of mining rig is produced, and how much energy they are collectively consuming. Various methodologies have been employed to estimate this, and I do not intend to delve into the measurement problems and debates around their accuracy. As I explain below, the exact current number is not important. Various estimates put the range of energy consumption between 1 and 10 GW. Translated over a year, this would be around 8.76 TWh/year to 87.6 TWh/year, or between 0.005% or 0.05% of total global energy consumption. In other words, Bitcoin is somewhere between five-thousandths of one percent and one-twentieth of one percent of total energy production, which of course is an enormous amount of energy. Bitcoin currently consumes as much as an average 350,000 to 3.5 million individual humans consume per year.

However, if Bitcoin continues to operate successfully for another ten years its energy consumption will likely be far greater than even the most exaggerated estimates of today, which makes quibbling about today's Gigawatts pointless. If the Bitcoin network attracts more store of value demand and the mining reward allocated out every ten minutes appreciates in value, more energy expenditure by miners will be justified to mine these coins. If, on the other hand, bitcoin were to collapse in value, the decline in the value of the reward would cause many miners to switch off their mining equipment, thus reducing the demand for electricity.

Figure 7

The current drop in the bitcoin price has brought about a relatively rare situation in the history of bitcoin: The bitcoin hashrate has been largely flat over a three-month period. This is useful in explaining the economic dynamics of mining and how it works. As the price of bitcoin was rising throughout the period of 2017, bitcoin miners worldwide were watching the value of the rewards they earn rise very quickly, while non-miners looked on with envy. In this environment miners would seek to expand operations, while outsiders enter the market, placing orders for miners being delivered. As the price continues to rise, the amount of mining hashpower on the network expands, which leads to the speed of finding blocks to increase. Bitcoin's difficulty adjusts roughly every two weeks, and will react to the rising hashpower trying to solve the bitcoin mining puzzles with raising the difficulty of the puzzle, thus making mining less profitable for miners. As the difficulty rises, the amount of processing power and electricity that needs to be expended to produce a bitcoin rises.

As the bitcoin price began to drop at the end of 2017, bitcoin's hashrate showed no signs of slowing down. This is to be expected, since new investment in new mining equipment takes time to be produced and deployed on the network, so 2018 was when most of the mining ordered by investors in 2017 was being deployed. As the hashrate went up by around five-fold over the course of 2018, and the bitcoin price lost about three quarters of its dollar value, the profitability of miners has been decimated. Reports emerged of miners in China even destroying their equipment, because it became too expensive to operate.

Anybody with familiarity with the mining business knows that over the last few months, only the mining businesses operating with the lowest electricity costs have managed to maintain profitability. Miners who invested in infrastructure connected to grids with prices above 10c/kWh, likely stopped being profitable months ago. At this point it is not clear whether even miners at 5c/ kWh are profitable at current prices and difficulties. This became evident as the difficulty started to fall in October, meaning that many miners were shutting down their miners, which signified not only that the miner is not profitable, but that it cannot even cover its operating expense.

It is unclear where hashrate will go over the coming few weeks, and it will depend to a large extent on the price of bitcoin, as well as the magnitude of new miners coming out of mining hardware producers. So long as the hashrate is falling, we can deduce that the marginal miners, the ones left with the most expensive electricity costs, are turning off their equipment. The hashrate has largely stopped rising around the end of August and has been dropping since October.

Figure 8

Figure 9

Over the next few days, bitcoin's mining difficulty is set to drop by another 13%, back to its levels in July. Whether this will be enough to stop the decline in hashrate remains to be seen. But for now, we have on our hand the first three-month decline in difficulty since July 2015. The only previous times this had happened was in January and February 2013, in January 2012, and in October 2011. Should the difficulty of mining drop another 10% in the next difficulty adjustment period, then it would be the first time since April 2012 that the Bitcoin difficulty is down over a six month period.

While it is definitely too early to start panicking, it must be noted that a drop in bitcoin's hashrate is a reduction in its security and the predictability of its supply. At a very basic level, the increase in the hashrate of bitcoin is what makes it very expensive to try to attack or disrupt the network with mining power. The higher the hashrate, the more expensive it is to solve a block's proof of work and receive its reward.

The cost of securing the bitcoin network cannot understood as a static or fixed amount of hashrate mining the network. The production of processing power worldwide is an ever-growing industry that continues to churn out more processing power at an ever-decreasing cost. A set amount of hashrate today might be too expensive to assemble to attack the network, but in a few years, this amount will become feasible to produce at a much lower cost.

Therefore, bitcoin's security does not just depend on having a high hashrate, but on having an ever-growing hashrate, so that a potential attacker is constantly having to update the arsenal they need to attack the system. More precisely, I view the security to be dependent on the hashpower rising, because that would signify that bitcoin continues to be a profitable use of processing power, which is incentivizing the producers of hashpow er worldwide to dedicate more of their resources to produce miners for bitcoin. This means that a potential attacker would be fighting a losing war against market forces in a vain attempt to accumulate more than 50% of the hashrate behind the network.

The production of processing power is not a secret monopolistic guild which can simply be hired to produce hardware only for attacking bitcoin. There are many companies that produce processing power, many that could produce it, and the more money is to be made in this business, the more companies enter. Should an attacker attempt to buy the hashrate needed to attack bitcoin while the hashrate is rising, they cannot monopolize all hashrate production to attack bitcoin, as the rising incentive to mine the network will continue to attract people to invest in mining it honestly. Thus, as long as the hashrate is rising, any potential attacker faces an ever-drifting hashrate goal to amass, and I would not expect it possible to manage to secure more than half the network's hashrate for attacking it, because the more hashrate is taken away from securing the network, the more profitable securing it becomes, the more investment goes into it, the bigger the task facing an attacker.

If, on the other hand, bitcoin hashrate continues to drop or stagnate for a significant period of time, this would be an indication that little new hardware is being produced to enter into bitcoin mining. The longer this goes on, the less daunting the task of a potential attacker becomes. With out new processing power being demanded and rewarded by the bitcoin network, it is feasible for another bidder to outbid the network for hashing power and muster enough of it to attack the network, particularly when many of the miners are not profitable and have mining capacity unused which they could be persuaded to sell for very little.

Bearing in mind that Bitcoin hashpower increased by five multiples in the space of one year allows us to understand how this threat could unfold. An attacker with very large amount of resources could bid for a lot of machinery while bitcoin's price is stagnant or declining as miners are laying off their machinery. Within a few months they could amass enough hashrate to disrupt the network.

Having said that, just because the attack becomes more likely does not make it likely, easy, or perhaps even possible, at all. Contrary to what many believe, a miner with 51% of the hashrate cannot overturn consensus rules, change the monetary policy, or even confiscate any bitcoiner's coins. These consensus rules are determined by the nodes, and not the miners, and if the nodes were to reject a miner's changes, all the hashrate in the world cannot force the node to accept them. A miner might be able to censor transactions from being included, and potentially double-spend some transactions. The real danger lies in passing off enough invalid transactions to create confusion about the valid chain and cause the network to be split among differing forks with different transaction records, which would likely highly undermine investors' confidence in the network, causing them to sell, which would bring the price down, making honest mining less profitable, and further attacks more feasible. But such attacks are not just expensive to mount, they are more expensive to maintain, as nodes continue to refuse invalid blocks, and the miner needs to continue to solve the proof of work correctly to keep adding the invalid block to the nodes. In the worst case where such an attack succeeds an efficient line of defense is for bitcoiners to raise the number of confirmations they would require to accept a transaction, which would make it harder for users to not rely on transactions recorded on an invalid ledger.

In conclusion, a dropping difficulty does make bitcoin slightly less secure, but not by much. We would require many more months, possibly years, of stagnation or decline in hashrate to begin worrying about whether this compromises bitcoin's security. Even if it does, the kind of attack imagined requires an enormous amount of financial resources, as well as logistic capabilities to pull off, that it is highly unlikely to ever be pulled off. Even if it were, it is not clear it could succeed in anything more than temporary disruption of the network.

This brings me back to the conclusion of the section of my book that deals with how to kill bitcoin. These sophisticated technical attacks on bitcoin are interesting in the same way a Hollywood movie plot is interesting; not because they will likely happen. As I explain in my book, the real driver of bitcoin security is the economic incentive it offers for people to keep it alive, to code it, to mine it, and to hold it. As long as government restrictions over money and banking persist, demand for bitcoin will likely persist, and those who use it will likely continue to find a way to use it. Should governments really want to kill bitcoin, they need to attack that economic incentive for using it. In other words, they need to provide their citizens with a free market monetary system that can compete with bitcoin. Their best bet would be a return to the gold standard, but I would not encourage you to bet on this happening any time soon.