Mining centralization: the nuclear option
347 sats \ 9 comments \ @machuPikacchu 25 Apr bitcoin
TFTC had a great article/podcast about the current state of mining centralization [1]. We've always had the nuclear option of changing the PoW algorithm to reset the mining industry.
Algorithms that support ASICs (like SHA256) are desirable for many reasons, but if we have to switch every 15 years or less then we need to get creative.
One idea could be to create a family of hash functions and rotate randomly between them each difficulty adjustment. The hash of the block at the next difficulty adjustment could determine which hash function will be used for the next difficulty epoch.
If we have a family of 10 hash functions, then the industry would have to produce 10 different ASICs. Big mining companies would have to support 10 different sets of hardware and only 1 is being used at a given time. It would disincentivize big mining operations and hopefully encourage more smaller and hobbyist miners.
  1. https://www.tftc.io/bitcoin-mining-pool-centralization/
write
preview
related posts
147 sats \ 6 replies \ @harrr 25 Apr
Big mining companies would have to support 10 different sets of hardware and only 1 is being used at a given time.
Sounds like that only makes mining more expensive, decreases profit margins, and leads to big miners growing while small miners are priced out.
Btw, one ASIC could do more than one algo.
reply
0 sats \ 5 replies \ @machuPikacchu OP 25 Apr
Maybe you're right, but let's consider what it takes to run a big operation now.
Large data centers housing thousands of a given ASIC. All of them are running at the same time and therefore you have high density of computation in that building. If they had to support 10 models then at any given time 90% of their hardware is idle. It becomes hard to justify having that as a business.
On the other hand an average person could buy a handful of a given ASIC and have it sit idle in their house until it can be used. They can be opportunistic with earning sats.
If you combine 10 ASICs into one then it would be less efficient than an ASIC optimized for a single hash function right? Then those who hash with the optimized hardware will have the advantage. That seems like it would favor plebs.
reply
16 sats \ 4 replies \ @harrr 25 Apr
If you combine 10 ASICs into one then it would be less efficient than an ASIC optimized for a single hash function right? Then those who hash with the optimized hardware will have the advantage. That seems like it would favor plebs.
You may lose some efficiency, but the basement dwelling pleb won't be able to mine 90% of the time, so they lose 90% efficiency.
reply
0 sats \ 3 replies \ @machuPikacchu OP 25 Apr
Then aside from hoping the pools adopt Stratum V2 what's the solution? Change the algo once every few years and repeat this process? Is there another way?
reply
10 sats \ 2 replies \ @harrr 25 Apr
Solution to which problem, fearmongering? If there is a real problem, miners can switch to a different pool whenever they want, and they are incentivized to do so as their revenue depends on bitcoin being trustless.
reply
0 sats \ 1 reply \ @machuPikacchu OP 26 Apr
The vast majority of mining equipment comes from a country whose government takes pride in censorship. That country is also expected to bring online significantly more energy production in the coming decade. They can make it difficult to change the pool that equipment mines on and most people can't be bothered to reverse engineer manufacturer firmware. Do you not see a danger in this trend continuing?
reply
0 sats \ 0 replies \ @harrr 26 Apr
Aw man. Have ASIC manufacturers sell 10 different ASICs and increase their revenue! That was your proposal to solve your first problem. Now you conjure a different problem, dude, that's just moving the goal post. I'm done here.
reply
35 sats \ 0 replies \ @nym 25 Apr
good idea
reply
0 sats \ 0 replies \ @SaloonCat 25 Apr
This was a great episode, I hope it receives the attention it deserves. I'll just draw out one part that struck me from the transcript at your link:
Matt Now, you know, that needs to be on the roadmap in the very short term because we're in such a bad place right now. And I you know, I think the Bitcoin community needs to start thinking about what the long term plan is here if miners don't start taking these options.
Marty Yeah. What hash function would you would you prefer to use?
Matt Honestly, just leave SHA256, but use a 64-bit timestamp. Let's just fix the timestamp issue. Keep SHA256. Who cares? And well yeah, we'll just brick all the miners and replace them with new ones.
https://rumble.com/v4r6o42-501-bitcoin-mining-pools-are-critically-centralized-with-matt-corallo.html?start=2936
The next questions sensibly included the concern that it could lead to another monopoly/dominance issue, given the timing and the announcement of a new 'mining system'. At any rate, a cartel forming doesn't look as likely as some healthy competition.
reply