Like many things with computers, scales can be deceivingly large.
I want to preface this with a note that I haven't done even napkin math on this, so there's a good chance my numbers are WAY off
All of the CPUs in the world probably won't be able to match even a handful of ASICs. It's probably the same with GPUs.
Each generation of ASICs is hashing about 4x faster with roughly the same power usage as the last. So after a generation or 2, you quickly run into the same sorts of issues where there is a good chance that all of the Antminer S3s in existence won't be able to hold a candle to a medium sized current-gen install.
And to mount an attack like that, you'd need to buy and power a monumental amount of these devices. The current hashrate is roughly 30,000,000 TH/s The current generation of Antminer can do roughly 14 TH/s
You'd need a million current-gen antminer S9's running at 1.3 gigawatts of power to get 50% of the hashrate (and from my extremely rough guesses, 4 million S7s, 12 million S5s, etc... Each generation also multiplying the power needed by some amount as well). And that kind of buying, manufacturing, and power generation isn't going to be easy to conceal.
Not to mention that evidence of a 50% attack will basically end bitcoin, making all of the money, time, and power spent on that attack truly worthless as you are left with a million completely worthless SHA256 hashing machines.
If you are able to pull off that kind of hashrate, you might as well just be a good player at that point, as you would be making $50,000 every 10 minutes at current prices.
> If you are able to pull off that kind of hashrate, you might as well just be a good player at that point, as you would be making $50,000 every 10 minutes at current prices.
You’re missing my point. If there is a pile of unused unprofitable hardware (because energy efficiency is too low given the mining rewards), then you can’t use it to make an honest profit. That means that it has very little value to honest miners and might be sold for a non-crazy amount of money. A dishonest buyer could use it to mount a 50% attack for a few million dollars of electricity, but I bet a dishonest miner planning such a thing could arrange to benefit in an amount far exceeding a few million dollars.
(I bet that the actual major cost would be the connections involved in acquiring the power, not the power itself. But there are many, many GW of working-but-idle non-base-load power plants in the world. You can’t buy their power at rates that make honest mining profitable, but short-term dishonest mining may be an entirely different story.)
Some people may attach considerable value to the destruction of bitcoin if nothing else.
But you missed the fact that the difficulty is increasing at a rate that makes old un-efficent hardware mostly useless, even in big numbers.
Even ignoring the power required, you'd need 67+ MILLION AntMiner S3s, and those still are just barely profitable if you are able to get power for $0.03 per KWh, and are still sold for over $100 a piece.
I'd be willing to wager that even if you were able to buy every single S3 that exists you wouldn't even be close.
So... what happens if the price of Bitcoin stabilizes and then drops enough such that even the most recent generation of miners are no longer profitable if all of them are running?
In that case Bitcoin might be caught in a "big freeze". Bitcoin takes about 2000 blocks to adjust difficulty, so if it becomes worthless overnight, the amount of hash power could drop enough that no new blocks will ever be found, and without any new blocks the difficulty will never adjust.
So one player would need to buy the vast majority of all miners, and even then they would only be capable of doing a double spend attack on a dead blockchain.
I want to preface this with a note that I haven't done even napkin math on this, so there's a good chance my numbers are WAY off
All of the CPUs in the world probably won't be able to match even a handful of ASICs. It's probably the same with GPUs.
Each generation of ASICs is hashing about 4x faster with roughly the same power usage as the last. So after a generation or 2, you quickly run into the same sorts of issues where there is a good chance that all of the Antminer S3s in existence won't be able to hold a candle to a medium sized current-gen install.
And to mount an attack like that, you'd need to buy and power a monumental amount of these devices. The current hashrate is roughly 30,000,000 TH/s The current generation of Antminer can do roughly 14 TH/s
You'd need a million current-gen antminer S9's running at 1.3 gigawatts of power to get 50% of the hashrate (and from my extremely rough guesses, 4 million S7s, 12 million S5s, etc... Each generation also multiplying the power needed by some amount as well). And that kind of buying, manufacturing, and power generation isn't going to be easy to conceal.
Not to mention that evidence of a 50% attack will basically end bitcoin, making all of the money, time, and power spent on that attack truly worthless as you are left with a million completely worthless SHA256 hashing machines.
If you are able to pull off that kind of hashrate, you might as well just be a good player at that point, as you would be making $50,000 every 10 minutes at current prices.