The hashrate deep dive 3 — Why hashrate

Ambroid
Ambroid
Jan 6 · 6 min read

Let’s take a step back from the articles I’ve written previously about Litecoin, Bitcoin and DASH. As an ongoing effort to educate both the readers and myself, you’ll have to forgive the ongoing and iterative process of writing these ramblings.

The feedback I’ve received so far has been two-pronged:

  1. More please.
  2. I don’t understand what you are talking about, but more please.

I dove directly into talking about hashrate without even mentioning what it is. In retrospect, expecting the reader to be an expert in Proof of Work (POW) was a mistake.

Let’s remedy this. I’m an engineer, not a mathematician, so apologies if I butcher anything by simplifying.

Hashrate

The completely basic view is that POW is based on solving puzzles to encrypt a block of transactions so that the marvel that is blockchain can keep moving.

That’s entirely wrong.

Checking transaction integrity is a simple operation an old PC can do at a much faster rate than the entire Bitcoin blockchain is doing right now. You have to put restrictions on the propagation of transactions. A regular computer could process the entirety of the Bitcoin transactions in a couple of hours (it’s called synchronizing a node). That’s the simple part. The transaction validation is not work intense, it’s the arbitrary Proof of Work mechanism that is.

Hashing itself is a one-way mathematical function, which processes any data input and spits out a hash. In our case a hash is a string of numbers in hexadecimal scale (ie 16 digits). This is a non-reversible (only feasible one way) operation meaning it’s a proof of something and the same data input always spits out the same hash.

It’s the combination of hashing with difficulty adjustments that’s of interest in Proof of Work. Deployed alone, they are useless. Together, they are the blockchain’s backbone.

Hashrate is a quantification of how many hashes per second a chip (CPU, GPU, FPGA, ASIC) can generate on the given algorithm.

We can increase the difficulty by lowering the target number which is to be hashed. We can lower the difficulty by upping the target number which is to be hashed.

In cryptocurrencies, when your computer mines, it’s generating hashes, which have to have a lower value than the current target number. So in effect, what we get by setting a target is a set of potential solutions. The smaller the number, the less correct solutions. This is the lowest possible (truncated) difficulty hash for SHA256 algorithm: 0x00000000FFFF0000000000000000000000000000000000000000000000000000

If that were the target, any solution below this number is valid and would reap you a block reward. Yes, this means there is more than one solution. Each node knows what the current target is at every point in time (except right after a miner finds a solution and it hasn’t propagated to the rest of the network — a process, which has been sped up immensely).

Just to reiterate: hashing is not directly useful work for the network, it is made so through block reward and transaction fees.

Consult the Bitcoin Wikipedia for a deeper dive.

The end result is a constantly self-balancing system, which attempts to attain an equilibrium given by the defined block time (how often we would like to see a correct hash solution).

Imagine a scenario where the price of a cryptocurrency is flat, miners wish to sell the coins they earn right away, the emission stays flat (we ignore the existence of transaction costs) and ignore the variability in hardware costs and their efficiency.

The only two variables are hashrate and electricity rates. Hardware depreciates at a constant rate that can stay undefined (the rate of depreciation is not relevant to the scenario).

As more miners join the network, the global hashrate grows, making solutions to the target hash more frequent than the optimal block time. And so the target number is lowered to make the solutions less frequent. The difficulty therefore grows. All miners have it harder now (the same amount of rewards are spread out among more workers), which lowers their profit margin. The time comes when the margin for some miners is so low, they might not even make back their initial investment in their investment horizon (remember, hardware depreciates). They continue to mine, trying to recoup at least a part of their investment. More miners join the network, until such a time that the weakest miners with the worst electricity prices are breaking even or are in the negatives.

At this point, they have two options — turn off their machines and wait for better times, or sell the hardware. The funny thing is, the difficulty hasn’t necessarily reached a plateau. Still more miners could be joining the network (buying machines from sad miners in worse locations) as long as they can find sources of cheap/free electricity. We will only reach a difficulty plateau when the locations with oversaturation of cheap electric energy are fully harnessed by cryptocurrency miners. At this point, they will be mining at typical commodity margins, which are bloody low.*

Imagine introducing just one of the variables we ignored to the model (miner hodlers, deflating emission, transaction fees, variability in prices and efficiency of hardware, etc.).

The prices of cryptocurrencies are chronically volatile, making margins wildly different from month to month. Same applies to hardware costs. Many miners choose to hold their coins in expectation of higher eventual prices, paying their monthly expenses (OPEX) out of their pocket. And the emission gets lower over time. Cryptocurrencies typically inflate at deflationary rates, with some coins like Viacoin already almost fully emitted.

Mining’s fate is the same as of every other industry — commoditization. Bitcoin’s security is carried entirely on the shoulders of miners, who are motivated purely through economy in fiat currencies. As long as you have to pay the bills in dollars, nothing else matters.

Now consider this:

Bitcoin has declined, over 80% of it’s price in 2018. At the same time the hashrate grew from 15 to 60 Exahash/s (15*10¹⁸ to 60*10¹⁸ hashes per second). We’re finally seeing the correction in exuberant growth. The price will stabilize and the miners who are struggling are selling their hardware to the more enterprising madmen. This process takes time, as the machines are physically moved from less economically viable places to sources of cheap (usually remote and renewable) electricity.

This is the reasoning for using hashrate and difficulty as the main data points of interest for analyzing the mining ecosystem.

In light of this, I invite you to re-read the first and the second article, hopefully with more understanding and ensuing discussion.

Regarding the first two articles: The original premise behind these articles was to investigate the saying “hashrate follows price”. People use it as a little axiom, without any rigor. From what we’ve seen so far — hashrate follows price at the trough of the bear market and that price follows hashrate at the beginning of the bull market. They can also move in tandem.

— — — — — — — — — — — — — — — — — — — — —

*The intermediate future for Bitcoin is low block rewards (say, inflation of less than 1% yearly, which will happen after the 2024 halving). This means that the economic incentives for miners will lie mostly in transaction fees. Although, users complain about high fees . Miners will have no other incentive but transaction fees (the block reward itself will become minor). At the same time we see a massive effort from Bitcoin developers to keep the fees at minimum through integration of further layers (Lightning) and sidechains (Liquid). Thiese developments risk putting users and miners in opposing camps. Miners want to generate enough revenue to keep their operations profitable, users do not want to pay $10 to transact $1000. As the block reward decreases, either the asset price has to increase, or the transaction fees have to be high enough to support miners in USD terms. I foresee this causing major friction towards the end of the second decade in Bitcoin.

The ultimate future, as portrayed in The Bitcoin Standard by Saifedean Ammous, is Bitcoin becoming digital gold, with emission (transaction costs) becoming fairly stable and mining a commodity-like business. Price will be volatile in the short/medium term but stable/appreciating in the long term.

In the commodity markets, the moment a commodity (lumber, salt, copper) rises in price, producers start producing more. In Bitcoin’s case, miners can increase the hashrate and split the rewards among more players. The price swings in Bitcoin are, however, so severe, that the miners are confused and end up making misled investments. The friction will stem from the fact that at current rate, the commoditization of mining is outpacing the stabilization of Bitcoin price itself.

Thanks to Scott Wehman.

Ambroid

Written by

Ambroid

Euphoric Cryptocurrency Speculator turned Depressed Cryptocurrency Speculator