How Do Seasonal Fluctuations Really Affect Bitcoin Mining?

There has been much ado made recently over supposed seasonal mining fluctuations on the Bitcoin network. 

The narrative goes that, as the rainy season in China comes to a halt around the August to October timeframe each year, cheap and abundant hydroelectricity dries up. This forces many inefficient late-model miners to shut off or move elsewhere to find more affordable, accessible energy — creating migratory or nomadic miners, if you will. 

The narrative also claims that the network sees a significant drop in hash rate and difficulty at yearly intervals roughly matching this seasonal decline in Chinese hydroelectric generation. This certainly appears to be the case now, in fall 2020, as many speculate the recent loss of about 48 exahashes per second (Eh/s) (30 percent of the network’s total hash rate) is due to just this phenomenon. But does the data support this for other years? 

And what about the recent Bitcoin difficulty adjustment at block height 655,200, one of the largest drops in Bitcoin’s history? Clark Moody’s dashboard shows the block difficulty experienced a 16 percent drop based on the aforementioned loss of network hash rate.

Bitcoin Block Production Rate, Difficulty Adjustment And Hash Rate 

The Bitcoin protocol is finely tuned and optimized for certain predictable outcomes. The way the network arrives at these desired results is through a series of carefully designed system rules and guidelines that were crafted into the free and open-source software upon its creation. 

The Bitcoin timechain is a series of blocks that verify, group and order transactions based on a preset series of rules. One such rule is the fact that blocks are added to the chain at a programmatic rate of approximately once every 10 minutes, six blocks per hour and about 144 per day. 

Block difficulty is generally proportional to the amount of computational work miners need to generate to produce a block. The Bitcoin Genesis block had a difficulty of 1. Yesterday, the block difficulty was 19,997,335,994,446. And, at the time of this writing, the block difficulty is 16,787,779,609,932. This means that today, it is about 16.7 trillion times harder to discover a block compared to the first block. Block difficulty is a unitless Bitcoin network metric. 

In order to maintain 10-minute block production rates with an ever-changing amount of miners and hash rate being produced on the network, the software programmatically adjusts block difficulty  every 2,016 blocks, or roughly once every two weeks, commonly referred to as a “Bitcoin block difficulty epoch.” This difficulty adjustment algorithm elegantly maintains an average block production rate, even with wildly fluctuating network hash rates. Over time, as more miners have tried their luck on the network, block difficulty has automatically adjusted upwards to compensate and stabilize block production rates.  

On the chart above, difficulty is seen as declining every so often after a reduction in hash rate, and increasing as hash rate goes up. If blocks are minted at a rate faster (or slower) than once every 10 minutes, on average, that would mean more (or less) computing power is being pointed toward Bitcoin than the difficulty threshold can accommodate. As more or fewer miners work toward the chain of blocks, the block difficulty target number will be changed to compensate, ensuring blocks are created at a rate of about one every ten minutes.

While we can clearly see the difficulty decreases on the linear chart, the above logarithmic chart makes the hash rate and difficulty drawdowns less perceivable. Historically on the Bitcoin network, block difficulty has trended upward and block difficulty reductions are rare. This is in part due to increasing mining equipment efficiency and effectiveness. 

There have only been a handful of months over the past decade in which the block difficulty ended at a value lower than when it started. The relentless growth is even more apparent in charts that illustrate an average Bitcoin network hash rate by month and year. There has not been a month, year-over-year, in which the Bitcoin network hash rate went down.

Accounting For Seasonal Fluctuations 

So, now that we have established that the Bitcoin hash rate, over long enough timeframes, is aggressively NgU (Number Go Up), is there validity to the theory that seasonal fluctuations cause significant changes in network hash rate? 

Per the chart above, it appears that the years 2020, 2019 and 2018 all saw average network hash rates trend lower toward the end of the year than they were in late summer and early fall. And what about other years?

Fall 2013

For 2013, the network doesn’t appear to have had a downward difficulty adjustment. This aggressive upward movement may be due to the revolution in ASIC effectiveness that was occurring during this timeframe. 

Fall 2014

For 2014, there were some difficulty adjustments downward during the late November timeframe. However, difficulty appears to be trending upward during most of the season. 

Fall 2015

2015 is a similar story to 2013: the network doesn’t appear to have had a downward difficulty adjustment. More and better ASICs were being rapidly developed at this time.

Fall 2016

So, 2016 sees a small difficulty adjustment downward around the October timeframe. Also, it appears that the growth of the network hash rate and difficulty slows down during the same timeframe, however, NgU.

Fall 2017

2017 tells a similar story to 2016: network hash rate growth stalls and difficulty actually adjusts downwards a few different times. This is especially noteworthy, as price was increasing aggressively along these same timelines. However, these fluctuations may not be due to migratory miners. Along these same seasonal time frames in 2017, some major miners were forking off the network and manipulating hash rate to pursue other avenues. 

Fall 2018

The fall of 2018 may show the most obvious seasonal trend of difficulty and hash rate declining rapidly across the network. It’s important to note that during these timeframes, the price was also falling from all-time highs. A significant percentage of the network, almost half, went offline seasonally. Yet, due to the elegance of the difficulty adjustment algorithm, the peer-to-peer network continued churning along. 

Fall 2019

The fall of 2019 doesn’t show as significant of a decline as the 2018 season does, but it does show a few significant difficulty adjustments downward and hash rate reductions. It also shows a similar stunting of network hash power growth during the same time frames. 

Fall 2020

So, this brings us to today. Mining centralization is a common criticism of Bitcoin and the narrative that many Chinese ASICs shut off seasonally appears to be valid for the fall of 2020. Where else would folks have about 48 Eh/s (30 percent of the network, as noted above) sitting idly by waiting for abundant and affordable energy? This works out to be about 3 million Antminer S9 ASIC mining units. 

So, where does this dropoff rank among previous difficulty drops in the October and November time frames? 

With chain data we can actually see that 2011 had the largest downward difficulty adjustment and an even larger month-over-month drop, occurring over a few different difficulty adjustments. This November 2020 difficulty drop is the largest we have seen in recent years caused by apparent seasonal fluctuations from enterprising Chinese hydroelectric miners. 2012 to 2015 did not see any difficulty drops between these months. 

How do these seasonal difficulty and hash rate fluctuations stack up to the entirety of the block history on the Bitcoin network? The histograms for both the difficulty adjustments and hash rate changes offer some insight: 

Is This A Bitcoin Mining Death Spiral?

When the network hash rate or price of bitcoin goes down, there is always much speculation on the possibility of what is colloquially referred to as the “Mining Death Spiral.” The claim goes that if the price drops low enough, miners decide to shut down and the network loses a sizable percentage of its hash rate. This would force some miners to liquidate their earnings, pushing market prices lower and furthering this vicious feedback loop until a death spiral ensues. For a network like Bitcoin, this would result in miners shutting down, blocks ceasing to be mined and the timechain ceasing to propagate  —  Bitcoin would fail. 

However, as an example, Dogecoin ($DOGE) is still minting blocks, so this doom and gloom theory may not hold true for these types of distributed systems. There are many enthusiasts, fanatics and “miners of last resort” that will maintain some of these systems simply for the sake of maintaining them, and the faith of belief. 

So, let us imagine if 50 percent of the miners stopped mining right at the difficulty adjustment block and shut down; what would happen? 

It would take the remaining half of the network about twice as long to find the 2,016 blocks. This would mean four weeks, or about one month, to get to the next difficulty adjustment point. What would the ramifications be for the network? 

The mempool would begin to build up, transactions would become delayed and fees would increase as folks bid up new, even more scarce block space. This is actually exactly what we have seen for a few days after the hash rate recently dropped off. 

However, the Bitcoin network eventually adjusted difficulty, as it has each time that 2,016th block has come up in the past. This difficulty adjustment algorithm is a very elegant solution to a few different challenges facing the Bitcoin network, and because of these types of solutions, the self-regulating system continues to propagate forward.

The post How Do Seasonal Fluctuations Really Affect Bitcoin Mining? appeared first on Bitcoin Magazine.