re: ASICs > GPU
ASICs are inevitable. ASICs are your friend. Embrace them.
1. There is no such thing as an ASIC proof PoW algo. Given enough time and financial incentive someone will figure out a way to make a purpose built rig that outperforms GPUs.
ASICs are inevitable. ASICs are your friend. Embrace them.
1. There is no such thing as an ASIC proof PoW algo. Given enough time and financial incentive someone will figure out a way to make a purpose built rig that outperforms GPUs.
2. The only practical way to keep ASICs off your network is to continuously manually change the algo, which bricks current ASICs, with the hope of discouraging future ASIC development. This comes with centralization & third party risk. Who decides when and how it changes?
3. ASICs enhance the loose game theory that backs bitcoin's security model. Attacks become much more expensive since the ASICs are useless after. GPUs can mine on other chains after an attack, ASICs cannot.
4. This all operates under the assumption that the chain is the dominant chain for a specific algo. Bitcoin cash for instance doesn't get the benefits of ASICs since it has such a low hashrate compared to bitcoin. Its existence even hurts bitcoin's security model a bit.
5. Until they switch their PoW algo, BCH represents a small threat to bitcoin. Since miners have a second option, albeit a poor one, for their SHA256 ASICs. For this reason, a bitcoin miner will eventually attack BCH to force a PoW change, it's a matter of when not if.
6. Through this logic, the highest valued highest hash GPU chain is substantially more secure than all the other GPU chains below it. The cost to attack them is significantly cheaper, and the malicious miner can easily switch back to the more valuable chain after.
7. The few chains that already have ASICs available for their PoW algo(and are the dominant chain on that algo) have a substantial lead in terms of security/immutability/censorship resistance than any chain that doesn't.
8. Stakeholders and devs should encourage ASIC development and try and get as many manufacturers onboard as possible. Use easy to ASIC algos rather than ASIC "resistant" algos. In the beginning ASIC development is often centralized, getting over that hurdle asap is crucial.
