OK, I deleted my earlier posts as I realized upon further reflection they were mostly using this thread as a soapbox for venting my frustrations with the process around ProgPOW and a “governance” process that is mostly powered by Twitter.
I’m going to stop contributing to this thread but for the record, I don’t buy into the argument that ASICs are good for network security, especially not before a transition that should lower the cost of renting soon to be obsolete hardware. I also believe the arguments regarding the rentability of GPU mining are credible only if one supposes an infinitely elastic supply of rented GPU power.
If anyone actually tried "putting in a credit card into AWS and renting 51% of ethash mining power, I doubt they would be able to get their hands on 100K+ idle p3dm.24xlarge instances. Not to mention that the credit card charges would fail, and way before getting close to making a dent in hashrate, the giant spike in demand would trigger various Amazon fraud systems and an prompt investigations.
To prempt the claim that the mining profits would subsidize the attack, a quick back of the envelope calculation is that renting GPU mining power from a perfectly elastic hypothetical AWS would recover about 2-4% of the costs. I am frankly surprised by the extent to which GPU mining ethash on AWS is unprofitable even if my rough calculation are wrong by an order of magnitude.
The strongest argument against ASICs is that any comparison of network security prior to transition to PoS should take into account the differences in expected depreciation of value of special purpose mining hardware vs general purpose GPUs.
At one extreme, you have a hyper optimized efficient special purpose hardware solution that may have been developed in secret, held by a handful of actors that have enough of an edge to squeeze out many other professionals.
At the other extreme, mining is done on general purpose hardware that anyone can get access to. It’s a race to the bottom with an equilibrium close to converting electricity to ETH. Some big players have economies of scale advantages, but they are milder compared with special purpose hardware, and in any case theu don’t squeeze out non-professionals that paid for the hardware anyway and may even be mining it a loss relative to the cost of electricity (e.g., kid in parents basement doesn’t mind running his GPU when he’s not playing games because he’s not the one paying for electricity).
It is not enough to believe that ASIC miners are altruistic. For the argument to work one has to suppose they are so altruistic that they would not even sell their hardware to non altruists.
Just suppose honest ASIC miners don’t attack the network themselves but instead sell off all their hardware right before their transition? Who would buy it from them, at what price? Would buyers plotting a 51% attack pay more or less than buyers plotting to scrap the devices for raw metal?
Finally, the argument that GPU manufacturing is centralized (e.g., only AMD, Nvidia being credible players) is also quite weak. GPU hardware is a competitive market and it doesn’t make sense for these companies to keep designs with big performance gains to themselves just to gain an edge in cryptocurrency mining. What matters is the decentralization of access to the hardware that computes the PoW, not the centralization of manufacturing. If centralization of manufacturing mattered then one could point back to the big semiconductor fabs like TSMC and are investing $15B to build new process fabs. For centralized manufacturing to be a problem, it would have to be economically interesting for AMD/Nvidia (or the fabs) to hold back innovative designs, and keep manufactured hardware to themselves instead of selling it to the public. Even if it was short-term profitable for these companies to do a 51% attack (and it isn’t), the collateral damage to their brand and reputation would be significant compared with brandless miners.
All of this makes the risks from custom hardware significantly higher than general purpose hardware.
FWIW, the strongest argument I’ve seen against ProgPOW is that the debate is “too divisive” and not worth it before the transition to PoS which is what we should be focusing on. That argument seems to discount the heightened risk right before the transition.
Finally, though I agree with the original poster regarding that this is a threat Ethereum should take seriously, I am somewhat put off by the blatant appeal to the authority of Vitalik, overly hostile style, ad hominem attacks, and assumptions of bad faith against the other posters here.
@CryptoBlockchainTech, I sense you mean well and are as frustrated as I am but ad-hominem attacks do your argument more harm than good. Even if you’re right that the other side is not just misguided but disingenuous and downright evil. The heat doesn’t make those on the sidelines of the debate any more likely to see the light. They come across what seems like an angry person full of spite and resentment personally attacking those civil ASIC apologists. You can’t blame them for deciding they prefer to be on the side of those with good manners.
Oh and I recognize that complaining that someone else is being too aggressive is somewhat ironic coming from an Israeli person. I’m told we Israelis breathe too aggressively