Hacker News new | past | comments | ask | show | jobs | submit login

ASICs have a hardware hash/watt race issue that GPUs (with memory hard algos) don't have (older GPUs are actually more ROI efficient). GPUs are also easier for a wider range of people to get with a much lower cost of entry. I'd argue that GPUs are still a better solution than ASICs, but this is an age old battle full of opinions.

Bitcoin hardware isn't worthless for any other application, any other sha256 based network works just fine (see BCH). The problem there is that it is just ripe for 51 attack because there can only be one top coin on each algo/compute layer. BTC = ASIC, ETH = GPU, Monero = CPU. The rest of them are all interesting datapoints on https://www.crypto51.app/




What's that attack cost supposed to represent? Just the electricity?

Seems like a misleading comparison, to get into a position to be able to do this you'd need significant investments in specialized hardware for the likes of Bitcoin and Ethereum. I've seen estimates of multiple billions of USD. And keep in mind that should the attack be discovered, which with coins running on open ledgers seems likely sooner rather than later, the price is going to tank, trust in Bitcoin will be broken and your special-purpose hardware will likely massively lose value. You'd have successfully destroyed billions of your own money.

On the other hand, the real cost of attacking some smaller coins may be even lower than that, because botnets are free or the electricity may simply be stolen, which happens a lot and can easily be done in less developed countries where the utility companies don't have sophisticated meters keeping track of where it all goes in the neighborhoods.


The site details the cost to attack using rented 3rd party compute. For some of the coins, there is enough compute out there that can simply be rented with no upfront capex/opex involved on the part of the person doing the renting.

The problem with that is that the rental market is an open bid supply/demand market. The second you start to rent out enough hashrate, the rental price also increases. That isn't factored into the numbers.

You are correct that the cost of capex/opex for ETH/BTC is in the billions, which is also what makes them so secure and attacking the network would also destroy the network. It is a brilliant feedback loop.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: