Bitmain’s Overt AsicBoost firmware update is good for bitcoin
Proof of work philosophy is the world's greatest choose your own adventure story.
Bitmain has announced that it has released a new firmware update for the BM1387-based Antminer models. It's popularly known as "overt AsicBoost", which improves the energy efficiency of bitcoin mining by tweaking some of the mathematical functions around mining.
So more hashing power with less energy consumed.
Overt AsicBoost is overt because the boost is clearly displayed in block headers. Basically, you can look at the blockchain to see which blocks were mined with overt AsicBoost, and so which pools and mining outfits are using it.
- If you want this to be good for bitcoin: Look at it as a way of increasing hashing power, which is good for bitcoin, or so they say.
- If you want this to be bad for bitcoin: Look at it as a way of reducing the energy required to mine blocks, which is bad for bitcoin, or so they say.
Why isn't everyone using it?
The main reason is because AsicBoost is patented. The other main reason is because they might have been in secret.
AsicBoost is one of a number of patented techniques for improving mining efficiency, but it's a major one, able to improve the total cost of hashing by about 20%. This makes a huge difference in the world of tight mining margins.
But there are (at least) two AsicBoost methods. The other one is called "covert AsicBoost", which produces similar results by manipulating the merkle tree but doesn't display the boost in block headers and so is much harder to detect. Covert AsicBoost also has other downsides. It's easier to perform with a low transaction count, which means there's more of an incentive for covert AsicBoost miners to mine empty blocks, and some variations also interfere with other aspects of bitcoin.
Covert AsicBoost is also patented, but being covert and all this is a bit tricky to enforce.
"We have strong reason to believe that some manufacturers of mining equipment have been secretly using this covert method to evade detection by the patent holder and gain unfair advantage over others but not revealing what they were doing, and without making it known to the patent holders," said (some of) the AsicBoost creators on 1 March 2018.
And in the grander scheme of bitcoin, covert AsicBoost also means giving one party a potentially unexpected advantage over others which might accelerate the consolidation of hashing power and the centralisation of bitcoin and other PoW cryptocurrencies.
All these reasons were enough to elicit the release of overt AsicBoost under an open licence regime, the "Blockchain Defensive Patent License". The general objective was to re-level the playing field by letting everyone access AsicBoost equally, remove the advantage being enjoyed by the less ethical covert AsicBoost miners and prevent hashpower from centralising faster than it needs to. Plus, the old patent was proving unenforceable anyway, so there probably wasn't a whole lot to lose.
It's worth taking it all with a grain of salt. There are a number of different perspectives around the issue. Bitmain says it holds the ASICBOOST patent in China, has been building ASICBOOST functionality into mining machines since 2017, but that despite allegations to the contrary it hasn't been using it until now, even though it could have been.
Plus, there's an argument to be made that maximising hashing power is exactly what a miner should be doing. Some might see tricks like covert AsicBoost as an "unfair advantage" that promotes centralisation, while others might see it as laudable design that improves bitcoin's security.
In a similar vein of thought, Bitmain arguably has a moral obligation to ensure its customers' equipment is running as efficiently as possible, and so should have been using AsicBoost all along no matter the consequences.
But, if Bitmain can single handedly decide that a huge portion of bitcoin's hashing power should not be using AsicBoost, then is bitcoin really decentralised?
It's also worth considering that Bitmain's farms and mining pools have collectively been controlling nearly 50% of bitcoin's hashing power, so assuming it wasn't already using AsicBoost it could have bumped its control to well over 51% any time it wanted. But it didn't... so Bitmain made the executive decision to keep bitcoin decentralised?
There are no right answers to any of these questions, and the details have never been conclusively settled.
Proof of work is goofy that way.
An ongoing shift
The release of overt AsicBoost under a more open patent only happened recently, on 1 March 2018. The first overt AsicBoost-mined block was popped out a few weeks later, on 24 March by a Slushpool miner using Halong Mining gear.
Bitmain is the world's largest mining company, but it's still just one mining hardware firm among many to flip the switch on overt AsicBoost now.
"Initially, we decided against activating this mathematical function in mining hardware produced by us, largely because of the legal uncertainty surrounding the use of AsicBoost," Bitmain said in its statement. "As an organization, we didn’t want to violate patent laws or act in any way that was untoward. Instead, we continued to focus our efforts on R&D and building the industry’s most efficient mining chips. It’s worth noting... our ASIC chip BM1387 has the capability built into the chip technology."
"In March 2018, we noticed new manufacturers of hardware advertising the use of “AsicBoost” and soon after that we noticed blocks being mined on the blockchain using AsicBoost. To give users of our mining pools the option to mine using AsicBoost we added support for it on both BTC.COM and Antpool."
Bitmain cites two reasons for its implementation of overt AsicBoost:
- Mining is increasingly difficult and competitive, and old gear depreciates quickly. "We recognize that cryptocurrency mining technology is rapidly evolving, innovating, and improving, and hardware depreciates rapidly. By activating AsicBoost, our customers' hardware will increase in effectiveness while the future hardware we make continues to be the industry gold standard," it says.
- "By activating this upgrade widely, we are making the Bitcoin network stronger than ever before," Bitmain explains.
Of course, the answers just raise more questions with no right answers. Bitmain's customers might be wishing it activated AsicBoost months ago rather than deciding to do the noble and honest thing on their behalf, and at their expense.
Not that it was the noble thing, if you follow the other line of reasoning. If AsicBoost means more hashing power, and more hashing power makes bitcoin stronger, then surely the correct thing would have been for Bitmain to activate AsicBoost as soon as possible, rather than leave bitcoin weakened just to satisfy its own selfish desire to do the ethically correct thing?
It's all part of the endlessly self-contradictory nature of proof of work.
More energy consumption is good because that's the cost associated with the production of hashing power, so therefore high energy requirements are what makes bitcoin strong and resistant to attack. But AsicBoost, which increases hashing power without a commensurate increase in energy consumption functionally means bitcoin mining consumes less energy. So in that respect it actually makes bitcoin weaker. But we might as well ignore that inconvenient truth and just simplify it as more hashing power equals more protection.
The great thing about the consistently paradoxical nature of proof of work philosophy is that no matter what happens it can always be justified as being good for bitcoin.
Both more energy consumption and less energy consumption are good for bitcoin. Miners have an obligation to produce as much hashing power as possible because that's good for bitcoin, but when they don't then that's also good for bitcoin. And all while the bitcoin powers that be – of which there aren't any – voluntarily enforce bitcoin decentralisation which just goes to show that bitcoin is so decentralised that it doesn't matter if it is centralised, so therefore it's not.
Whatever Bitmain's firmware update means, it's by definition good for bitcoin.
Disclosure: At the time of writing the author holds ETH, IOTA, ICX, VET, XLM, BTC, ADA