With the Astor Testnet currently live since May 2019, Alexander Tsankov's ECIP-1049 proposal to change Ethereum Classic's proof of work algorithm has some real evidence that this could be the potential solution for negating future 51% attacks. A good time to discuss scenarios around what Ethereum Classic implementing SHA3 would actual entail.
Table of Contents
- What would be the biggest challenge?
- SHA-3 ASIC miners should be easier to produce
- What is the trade off of adopting SHA3?
- Potential Roadmap for SHA3
- Final words
The inevitable replacement of GPU miners with ASIC miners stands out as the biggest challenge. Basically, you lose all the people and organizations that mine ETC as GPU only and you would also lose the hypothetical ETH hashrate that would seek to migrate somewhere after ETH2 launches POS.
What you can gain, however, are two things:
- ETC will be mined by SHA3 ASICs that anyone can purchase at a competitive price from multiple vendors (not just Bitmain).
- ETC might become the apex predator of SHA3.
Let's unpack that second point first.
If you take a look at Bitcoin Cash (BCH), it does have more hashrate than most coins. Yet because it uses the same algorithm as Bitcoin (BTC), it is vulnerable. A single large BTC mining pool has more hashrate than the whole BCH network, and can thus organize a 51% attack on BCH.
Ultimately, all blockchains with the same hashing algorithm are competing for the same miners. You could argue that it is in Bitcoin's best interest to do the following:
- Dedicate a mining pool to mine empty blocks on BCH to effectively kill it (no transfers will be possible).
- BCH miners will realize that they are not getting any payouts and the chain is dead, so they need to point their ASICs to mine BTC. Note that economically they do not have a choice - otherwise their ASICs are just expensive bricks.
- After BCH is dead the mining pool returns to mining Bitcoin. BTC has effectively killed the competition, resulting in price going up and the security of BTC increases even further.
What about ASIC resistance?
The problem with so-called ASIC resistant hashing algorithms is that you either have ASIC monopolists or it is impossible to achieve. If only one firm has done the right research and development, then they can manufacture ASICs and they sell them for as much as they want. Even if an ASIC costs $100 to manufacture they can sell it for $50,000 and you pretty much have to buy it even if it is just 1.1x more energy efficient - such are the rules of the game.
If such ASIC doesn't exist yet, somebody can repurpose their GPUs from other coins for the same attack. e.g. Monero, 0xBTC, Callisto and some other GPU coins can team up to kill ETC through the same 1-2-3 algorithm.
Which leads us back to point 1, the only way to be secure is to be the apex predator in the world of hashing. So for SHA-256 you have to be BTC. For cryptonight you have to be Monero and so forth. The rationale behind ECIP-1049 is that there are no significant SHA-3 mined coins yet, and ETC could secure this top position.
ETC will be mined by SHA3 ASICs that anyone can purchase at a competitive price from multiple vendors (not just Bitmain).
The thought that SHA-3 ASIC miners will be available at a reasonable price, simply comes down to a hypothesis that there will be healthy competition and not a monopoly as there will be various companies producing them. Let's critique this hypothesis and discuss evidence for and against it.
Producing an ASIC requires two things:
- R&D. You need to come up with ASIC design.
- Capital for placing manufacturing orders once you have the design.
The only way to make ASICs accessible is to encourage competition: use simple, ASIC friendly algorithms. It will push research and development costs down to a point where a lot of companies can compete.
The problem of the current algorithm Ethash is that the R&D aspect is so expensive that it creates a monopoly. Yet if we compare that to SHA-3, the R&D for SHA-3 ASICs is literally open source and fairly licensed. This means it will not require the same amount of investment to produce ASIC miners for SHA-3, and a healthy competition between manufacturers will ensure the price is reasonable.
The downside is the loss of positioning of being the blockchain that every ETH miner will switch to once Ethereum implements PoS. The thinking When ETH goes PoS all their miners will switch to ETC, has started to become an unproven meme - but even so it is highly likely.
However, the bigger question is to ask which strategy will ensure that ETC remains apex predator within a given mining method?
- Keep Ethash with the hope that ETH miners will move to ETC and will stay loyal to ETC chain. Hope that they stick around and don't move to another GPU-minable coin and abandon ETC overnight when a new coin comes out and thus make ETC vulnerable.
- Cause short term havoc and move to SHA-3, but ultimately emerge as the SHA-3 apex predator chain and become so dominant within this niche that nobody can 51% ETC. Miners who purchase SHA-3 ASICs will have to stay committed to mining ETC, which creates a positive hashrate feedback loop that BTC currently enjoys.
There is not really an obvious answer, it is not a mathematics question or a computer science question. Many question marks exist around a migration that would need to have well researched answers before any serious consideration. Such as: how do you coordinate the migration? How do you estimate how much hashrate SHA3 ETC will have? How much will ASICs truly cost? How long will they take to be delivered?
One thing is for sure, it would be very silly to attempt to make a migration before SHA-3 ASIC miners are available on the market.
PoW chains all compete within a separate niche, and this niche is defined by hardware. Today launching a SHA256 coin is stupid - bitcoin will kill you. Launching a GPU coin also does not make much sense today. If ETC was 51% attacked then so will you. The only reason something like Callisto or Ubiq were not 51% attacked is because currently nobody cares enough to make them a target.
Right now, I think 0xBTC is the largest SHA3 coin, and that has a very tiny hashrate. In terms of available hardware, there are some FPGA miners but no ASIC's yet. Which is why we believe the best course of action would be to do what political science people do. Come up with polls and proceed with research, ask all the ETC communities in all countries, speak to miner farms and talk to miner pools and also home miners.
Here is what we would propose in terms of a rough roadmap, one which we think Tsankov would agree with:
- Launch SHA3 testnet. ✔️
- Educate people on the trade offs. (In progress)
- Have a serious go/no-go discussion with the whole community.
- If the decision is to make the migration, then there needs to be enough ASIC manufacturers ready to commit, make and ship ASICs by date X. Then ETC will switch to SHA3 at block Y (which should be date X + 3-6 months). If by date X there are no ASICs shipping then the plan would need to be abandoned.
I hope you enjoyed reading this article discussing SHA-3 for Ethereum Classic, if you would like to learn more Rick recently had Alex on his Digital Fortress podcast where they discussed many of the same topics.
Future is bright!