Another Doomsday Scenario for Bitcoin

Despite the constantly increasing concerns in Bitcoin’s community on the digital currency’s growth limit, there still appears to be some room left to worry about other subjects. The latest addition on the list of things that could kill Bitcoin is the upcoming “halving” event, expected to take place in July this year. This concern popped up on the Bitcoin development mailing list, as Luke Dashjr suggested Bitcoin Core should prepare a hard fork to lower the difficulty retargeting time, allowing the protocol to quickly adapt to a potentially lower network hashrate.

Block Rewards and Difficulty

To understand what is being suggested here, it is required to first understand some of the basics of Bitcoin. Most users know that Bitcoin has a finite supply of 21 million coins. These coins are slowly released over time via a pre-determined schedule to Bitcoin miners. At the moment, less than 6 million Bitcoins are still available for miners. Each time a miner “finds” a block, the miner is currently rewarded 25 newly created Bitcoins. This amount decreases over time at moments according the pre-determined schedule. More specifically, the reward miners get decreases by half after every 210,000 mined blocks, or roughly once every four years. On November 28, 2012, the first halving took place, lowering block rewards from 50 to 25 coins per block. During the next halving in July the reward will drop to 12.5 coins per block.

Bitcoin’s difficulty algorithm tries to make sure that each block takes about 10 minutes to be created. To do so, the algorithm sets a new target every 2016 blocks, or about every two weeks. A target is nothing more than a condition that new blocks have to meet, but set in such a way that the total computational power of the network will need 10 minutes on average to meet this condition.

Bitcoin’s retargeting period of two weeks is relatively long. Some altcoins even retarget every minute. Until now, this has never been much of a problem. The only noticeable thing is that slow retargeting combined with a constantly increasing network hashrate has caused blocks to be mined a bit faster than 10 minutes on average historically. It is now being argued that the long retargeting period could also cause a problem at the coming block reward halving.

The Situation

Because blocks don’t take exactly 10 minutes to be created in reality, we can only estimate the exact moment of the next reward halving. We, however, know for certain that it will take place at block 420,000. We also know that there will be a difficulty retargeting at block 419,328 and at block 421,344. To put it differently, there will be at least 9 days left to the next difficulty retarget at the moment of the block reward halving.

The main concern is that this period, and block creation times in general, will be extended significantly due to a sudden fall in hashrate at the moment of the block reward halving. This is based on the assumption that mining will not be profitable for a potentially significant part of the network after the halving takes place. This is not a strange thought knowing that block rewards are currently worth $1.5 million per day. In the event of a block reward halving, miner income per day thus automatically drops by half (unless the price of Bitcoin magically doubles at the same time). Since mining requires electricity, of which the costs remain the same before and after the halving, the operating expense ratio will fall for each miner. This doesn’t mean half of all miners will shut down as well, but it is likely that the most inefficient miners (when expenses exceed income) are pushed out. This may lead to some of the hypothetical scenarios below.

Scenario: Small Drop in Hashrate

After the halving event in 2012, the average network hashrate fell by about seven percent in the first week following the halving. This is a trivial amount considering that current daily swings in the total network hashrate can be around 15-20 percent. It is a reasonable assumption that the drop in average hashrate will be bigger this time around, especially given the fact that the price of Bitcoin doubled in the months prior to the 2012 halving. A rising price creates room for more miners (leaving more income to be shared), but it takes some time to get an install the equipment. If block rewards are slashed when prices are rising, this lag will soften the impact for the already present miners. Unless the price doubles again in the coming months, a moderate drop in hashrate is the more likely event.

Scenario: Moderate Drop in Hashrate

In this scenario, the total Bitcoin hashrate would drop by 30-35 percent on average shortly after the halving. This scenario could already cause some significant problems, because it would increase the time for new blocks to be created up to 15 minutes. This situation would last up to two weeks, because there would still be 1,344 blocks left to the next difficulty retarget and the number of blocks per day would drop to 96. It also creates further downside risk, because miner income per day is slashed by more than half as a result. The daily reward before the halving would equal $1.5 million worth of Bitcoin. The halving brings this down to $0.75 million per day, and the extended time to mine a block further reduces this to $0.5 million per day until the difficulty retargets. This equals a drop of two thirds in miner income per day.

Lastly, the price of Bitcoin could also be facing pressure as extended block times would further limit Bitcoin’s transaction processing capabilities. Bitcoin would temporarily process 1MB of transactions every 15 minutes rather than every 10 minutes, equalling a reduction of the current block size limit to less than 700KB. If the price drops by just 10 percent, then the combined effects of this drop, the bloc reward halving and extended block times could (at least temporarily) lower miners’ income by around 70 percent per day. This may trigger the next scenario as well.

Scenario: Death Spiral to the Bottom

This scenario may be triggered by either one of the following two events. The first one is a worst case sudden drop in hashrate of 50 percent in line with the reward halving. This is not very likely, because this would only happen if all miners are about equally efficient. In reality, economies of scale ensure that the biggest mining farms are also the most efficient ones, and that inefficient miners will be a minority group in terms of total hashrate. The second event relates to the previous scenario spiralling out of control.

If the hashrate drops by 50 percent on average following the halving, then block times will double as well. Effectively, this would temporarily reduce miner income by 75 percent rather than just 50 percent. This would last for at least 2.5 weeks until the next difficulty retarget at a slowed rate of 72 blocks per day. This initial effect is pretty close to how a moderate drop in hashrate might play out. It is also not hard to see how this could escalate further, as more miners may be triggered to switch off their equipment if their income per day drop by 70-75 percent. If panic would strike subsequently, the price of Bitcoin and the total network hashrate could get trapped into a death spiral. A dropping price would continue to add more pressure on the miners, and miners leaving the network would continue to add pressure on the price of Bitcoin. The blockchain may get “stuck” at this point, making this a doomsday scenario for the coming halving.

The Fix

Avoiding the previous scenarios is actually quite easy. It only requires a minor change to the difficulty retargeting period as suggested by Luke Dashjr. The quicker the difficulty adapts to the new situation, the less opportunity there is for the new situation to spiral out of control. Of course, this change is equally “simple” as raising the block size limit to 2MB. A hard fork is required to do so, decreasing the odds of a fix being implemented in time. The window to implement the hard fork would be just a few months, while the minimum safe distance for a hard fork on the block size limit is considered to be more than a year. Furthermore, miners may not agree to the new retargeting period because it potentially also hurts them economically. Block times have historically been below 10 minutes due to slow retargeting, providing a small additional subsidy for miners. Shortening the retargeting window takes this away.

Altogether, it is hard to see a fix being implemented on time, so for now the community will be left with something other than the block size limit to worry about.