When Satoshi created the Proof-of-Work (PoW) algorithm for Bitcoin, he envisioned a distributed, widespread network of individual miners validating transactions and being rewarded with new bitcoins.
However, this vision has not come to pass as mining, at least for bitcoin has become "centralized". How did this come to be, and what will the future hold? Let's briefly go over the history of bitcoin mining. Then we will really get into the juicy controversies and battles going on in the crypto mining space.
The Early Days:
The first bitcoins were mined by multi-core CPUs (central processing units). Anyone with computers that had decent specs was able to participate in mining.
GPUs Started to Change the Game:
It turns out that using the CPU to mine is highly inefficient, because a CPU was made for general purpose.
The GPU (graphics processing unit) is designed to render video, and has a much faster processing speed.
Here's a quick primer on the difference between CPU and GPU:
You can think of this analogy. Using the CPU is like using a companies executives to do the physical labor, while using the GPU is using laborers to do the physical labor. Here's a more detailed explanation of GPU vs. CPU:
In Oct 2010, the code for GPU mining (for Bitcoin) was released to the public. Within months, we saw the hashrate jump in orders of around 1000 - from around 10 gigabytes to 10 terabytes.
Note: It should be noted that the hashrate was rising also because the price of bitcoin was rising. The rising price encouraged more miners, or miners to be more productive. With rigs that are mining bitcoins faster and faster, the hashrate adjusted rapidly (Bitcoin's code established this mechanism).
This also increased demand for AMD and NVDA graphics cards, driving up the price of these products to the chagrin of many gamers.
FPGA is Even Faster, but Harder to Set Up:
Soon after the introduction of GPU mining, field-programmable gate arrays (FPGAs) became the next best way to mine bitcoin and other Proof-of-Work cryptos in mid 2011.
Mining began to scale once FPGAs were modified for the purpose. The biggest draw to this hardware was the fact that it used three times less power than simple GPU setups to effectively accomplish the same task. In cryptocurrency, for a brief moment, FPGAs were the best thing that’d happened since sliced bread. (thenextweb.com)
The thing is, FPGA by design is meant to be tweaked for specific purpose after the purchase. So FPGA-mining was really not for the hobbyist as CPU and even GPU-mining would be. Still, many entrepreneurial miners did shift to FPGA, which subsequently increased the hashrate another two-fold.
ASICs Becomes Standard:
In 2013, a Chinese computer electronics company Canaan officially produced the first ASIC semiconductor for bitcoin mining. It's first product was call Avalon, which was first shipped out in January 2013.
Extending the above analogy...
If CPUs are the executives, GPUs the general laborers, then FPGAs were very adaptable laborers who can be trained for specialized tasks. Then ASICs are the expert specialists you only use for a very specific role because they were born for that role.
With the introduction of ASIC mining, Bitcoin's hashrate surged again. Individual miners were being pushed out and replaced by mining pools as well as industrial-size mining operations (mining farms).
The innovations did not stop after the first ASIC chips. In fact, the earlier chips operated about 60 GH/s, while some of the ones in 2018 are claiming 14 TH/s. That's almost 250 times faster since the first ASICs chips in 2013.
You can say that Moore's Law was at work. But after 50 years, many technologists are calling an end to this phenomenon - where processing efficiency has been almost doubling every year. The improvement in ASIC chips therefore will become less and less significant, though there will likely be other ways miners improve on efficiency.
We might be nearing the end of any significant improvements in efficiency based purely on the processor. "Colwell said that for the Defense Department, he uses the year 2020 and 7 nanometers as the "last process technology node." (Scientific America). Meanwhile, some of the latest chip developers are claiming to have 7nm transistors already.
Centralization (Mining Pools and Suppliers):
The advent of GPU and ASICs led to centralization in the form of mining pools, larger mining farms and of hardware producers.
Mining Pools and Farms: As the mining hashrate surged, mining pools were formed to achieve a better chance of getting the right hash. Joining such a pool essentially gave the miner more frequent rewards, albeit at a much lower rate because the rewards are shared. So a participant might get a small weekly reward instead of the whole block reward once in a blue moon.
The same concept of scale applies to mining farms, except in this model, one company owns all the hashing power.
Today, the distribution of hashrate shows that there are only a handful of big players in the mining space, not the well distributed network that Satoshi envisioned.
Hardware Producers: Even if ASIC did not come on to the scene, GPUs would have been dominated by just a handful of companies, such as Advanced Micro Devices (AMD) and Nvidia (NVDIA).
ASICs chips are also dominated by a few leaders such as Bitmain and Bitfury. Now, the profitability of ASIC mining has caught the eye of Samsung. We should expect this Korean electronics company to become a disruptive competitor to the dominant crypto mining chip makers today.
Bitcoin.com reported: Samsung Electronics’ “semiconductor division…accounted for about three-quarters (73%) of total operating profits, leading the company to a record high.” Samsung explained:
Demand for the semiconductor division increased due to sales of system LSIs [ASICs] for flagship smartphones and demand for virtual currency mining chips.
Chip production has always been somewhat centralized because the scale and risk requires deep pockets. Add the immense price volatility of Bitcoin to the risk, and you can see why only a handful of companies were able to take on this task.
Note: Think about the margins for these ASIC chip producers. The margin has to be high enough so that it is more profitable to sell you the rig then to mine for themselves. This means, customers of these ASIC rigs are already at a disadvantage.
I think IF price eventually becomes less volatile, especially to the downside, AND the improvement on speed becomes less and less significant, you will see chip producers pivot into miners themselves.
Controversy and Drama:
Miners vs. Coders (Governance Scaling Issue)
Bitcoin was created with an adversarial environment in mind. However, there are some tensions that could cause existential concerns for Bitcoin, especially when the size of the network scales up.
Everyone is talking about scaling issues these days, but scaling is NOT just about speed and efficiency. It is also about governance and how decisions are made at the protocol level. The first bitcoin community was mostly Libertarian so we can assume they achieved consensus much faster than today, where the community is much more diverse.
People in this space have different incentives. Some of the Bitcoin "core" coders are still Libertarians, while the miners are mostly just business operators in it for the profit. You can already see that they will have different opinions on important upgrades to the network.
Example: Segwit was an upgrade proposal intended to improve the speed of Bitcoin transactions. To make a long story short, the adoption of segwit would hamper ASIC miners' profitability if not completely put them out of business. Hardware is not so easy/fast to upgrade as software - many see this tension as a classic case of software vs hardware development that has always existed.
Because of this rift, the segwit upgrade has been slow in adoption. As we can see on the chart below, we are barely at 40% adoption after a year since the proposal.
Note that segwit was a soft-fork. So, despite the contention, the only trouble for bitcoin users is that they have to select whether they want to use segwit or legacy wallets.
However, this rift lead to another controversy. Many believe that Jihan Wu, co-founder of Bitmain, an ASIC chip manufacturer, supported Roger Vers in the hard-fork of Bitcoin Cash with the intent to replace Bitcoin "Core". This attempt was known as the "flippening", but did not result in any real damage to Bitcoin Core's network. The result was one network that has the legacy Bitcoin with Segwit slowly adopted and another network that has a larger block-size and faster block-time.
There are a lot of accusations at the moment, and it is tough to know exactly what is going on behind the scene. It is a highly profitable business and therefore becoming very competitive. This is the wild wild west, and I can only imagine the big dogs doing whatever they can to maintain market share. But even though I side with the accusers, I don't blame the likes of Bitmain because mining is a business after all.
Still, here are some articles that can catch you up on some of the latest accusations of foul play, and of course a rebuttal:
Here's a good episode of the Unchained Podcast by Laura Shin in an interview with David Vorick of Obelisk (Siacoin ASIC hardware manufacturer) who accuses Bitmain of playing dirty. Bitmain of course has its rebuttal.
Alts vs. ASICs and Secret Mining
In the podcast above, Vorick accused Bitmain of mining Siacoin themselves before shipping the mining hardware. This would be a profitable step to take because the manufacturer would be able to capture the juiciest mining profit before the algorithm adjusts.
SIA (Can't Beat Them, Join Them) : ASIC mining has penetrated many Alt networks by 2018. Siacoin (SIA) was suppose to be ASIC-resistant with its BLAKE hashing algorithm. But alas, Bitman was able to create an ASIC for it. As noted in the podcast, the community dealt with ASIC by saying "if you can't beat them, join them". After a hard-fork, SIA is now only mined with ASIC.
Note: ASIC-resistance is a relative description. Being ASIC-resistant can simply mean ASIC only improves speed 2x instead of 100x. All proof-of-work (POW) algorithms are susceptible to ASIC improvement to a degree but small improvements does not warrant mass production. Basically, if a cryptocurrency uses a proof-of-work algorithm, ASIC mining will likely be part of its ecosystem if it is not already.
XMR (Battle to the End!): Vorick also claimed that there has been secret mining in Monero (XMR) since 2017. Today, ASIC hardware does exist for XMR's cryptonote algorithm. The XMR community was indeed dealing with a surge in the hashrate in 2018.
This community dealt with the issue by constantly hard-forking with tweaks to its hashing algorithm. Riccardo Spagni, Monero's lead developer tweeted about Bitmain's ANTMinerx3: “WILL NOT work” for Monero, since Monero's core development group (CDG) is going to perform regular updates of the hashing algorithm."
We can see the hashrate dip in April 2018, after the fork to XMR v.7. Note that part of the surge in hashrate was due to price going up in late 2017, but the fact it kept climbing and holding up suggests that price was not the only factor. So far, Monero has staved off ASIC, but is now going to have to constantly upgrade, a burdensome feature unless the dev community is centralized.
VTC (Can't Touch Us - Yet) Another alt against crypto is Vertcoin (VTC). So far, it has been able to avoid the surging difficulty/hashrate caused by faster mining hardware. We can see from the chart below that the hashrate followed the bubble in late 2017, but also receded when the bubble popped. There was no "ASIC-induced spike".
We noted that ASIC was able to specialize for the BLAKE and Cryptonote algorithms that SIA and XMR used. VTC uses an algorithm call Lyra2REv2. We can't say that this is going to be ASIC-resistant because we can't say that Bitmain or any large manufacturer has put resource into trying.
XMR is ranked 12th and SIA 40th, with market caps of 1,871,109,513 and 273,917,165 respectively. VTC is ranked 147 with a market cap of $41,442,935. If VTC grows, I am sure it will become more vulnerable to ASIC mining. (as of 8/6/2018)
In April 2018 ,Equihash was ASIC resistant because its purely memory oriented. According to coinguides.org, here is a list of coins using the Equihash algorithm:
- Bitcoin Gold (BTG)
- ZCash (ZEC)
- Bitcoin Private (BTCP)
- Komodo (KMD)
- ZenCash (ZEN)
- ZClassic (ZCL)
- BitcoinZ (BTCZ)
- Hush (HUSH)
- Zero (ZER)
- Bitgem (BTG)
- Zelcash (ZEL)
Soon in May, this algorithm also fell to ASIC mining as lamented by VTC's tweet:
The current state of mining is fueling an experiment on governance. ASIC mining is disruptive to proponents of decentralization, which makes up a big portion of the crypto community especially the developers side. But there are some like Vorick of Obelisk who sees ASIC as an inevitable standard. How will these communities deal with ASIC can reflect on the strength of its community. Even bitcoin had a critical split, which revealed a weakness in its having a diverse decentralized development community. Fortunately for BTC, its community was big enough that even after the split, the "Core" developer community was still the largest and possibly the most talented in the crypto space. How will the smaller but growing projects fare?
From a practical, and tactical point of view, the tension here is between software and hardware developers. If software developers have good governance, they can overcome hardware specialization because software devs will have better turnaround time in upgrades. But in larger networks, it is hard to come to consensus, and the astute manufacturer will take advantage of the large networks first. Don't believe in claims of ASIC-resistance from coins that have not been subject to this attempt.
I would like to leave you with this 25-min talk by Andreas Antonoupoulos on the software vs hardware dev issue. This might seem like a tangent from the ASIC mining issues, but it gives a great perspective on the current drama between developers and miners.