All right, I am having a difficulty illustrating this.
I see the bail-out applicable in a scenario where a mining corporation that does lets say 25% of all BTC HR drops of the map because of connectivity issues.
Bail-out comes in where this mining corporation needs to move all of its resources to get the total BTS HR up to where it was.
Without it the attacker has a much lower entry barrier.
I've also edited the last post prior to you posing, sorry...
I still fail to understand your idea of a bail out. What exactly happens in a bailout? Is the mining "corporation" given money? Are they given miners to rebuilt their hash rate? I don't understand what a bailout is or how it would work in the Bitcoin context.
Yes, I agree that 25% lower hashrate would be (generally speaking) a 25% lower barrier to entry. I guess I just don't see how an attacker could easily get that 25% lower hashrate, and whether those resources spent taking down those mining pools (or mining corporations, as you call them) might be better spent on simply adding more hashing power to their own attack.
I am not even considering a maliciuos entity developing their own ASICs (or whatever the next technology is) - this usually involves significant r&d and setup costs. The attacker would be better off simply purchasing the same technology that "good" miners are currently using - the most effective, currently available technology. The community upgrading to the next technology simply means the same technology is available to the attackers. Keep in mind that any technology cannot cost more than the expected profit from mining over some reasonable time frame (for ROI purposes.
Why are you not considering a malicious entity developing their own ASIC? That's by far the cheaper route to do a 51% attack (currently).
A 51% attack via GPUs would require around 50,000 of them (20TH/s at 400MH/s per GPU). Assuming you could find 50,000 video cards for sale that do 400MH/s for $200, it would cost $10M. And that's not including the costs for the basic systems to host them (say, 5 per rig, so 10,000 rigs would need to be built), the cost to rent out a datacenter to house them, the costs to set them up and keep them running, electrical costs, etc. I'd say you're easily looking at $15M, maybe $20M.
Now, a 51% attack via ASICs, with the attacker developing his own, would cost, say $2M in R&D, plus the costs of the chips themselves. People here have speculated that the Jalepeno might use a single chip, and that the cost of the chip would be around $10 in bulk. So, we'll go with that. 3.5GH/s for $10, plus, say, another $40 per chip for the board and other electronics, so $50/3.5GH/s. To get to the requisite 20TH/s, you'd need 5,715 of these chips. At $50 each, that'd be another $0.3M in costs.
So, you can see now why it is important we introduce ASICs into Bitcoin mining. It is to protect against other ASICs mining.
I agree with you that changing from GPU's to ASICs does not help prevent an attack using the same devices though.
You mean if Bitcoin activity in general was blocked?
This sort of case has been discussed, but more in the context of forced governmental restrictions. However, the data could still be encrypted to bypass such restrictions, and unless the restricting government planned to block all encrypted data, they would have no way of knowing what is Bitcoin traffic vs what is other traffic - encrypted traffic all looks very much the same.
Encrypted traffic does not all "look very much the same". Before you can encrypt, you must handshake--this is typically a dead giveaway. Even if a handshake was unnecessary, the bitcoin data stream is open to the public and can be monitored, so it would be excessively easy to find encrypted traffic matching the same patterns. The only significant way to combat this is to use steganography or excessive padding to the point where the real data may only be a small portion of the padding. And even in the latter case, it may still be obvious because there are not many protocols that do this sort of thing.
Right now, someone could develop their own ASIC and EASILY have and hold 90% of the hashrate with a small number of devices (say, 200 devices @ 1 TH/s each). Once ASICs are in the wild, and the hashrate is pumped up to 500 TH/s (my estimate of where it will be after BFL finishes its current preorders), it would take 5,000 of those same devices @ 1TH/s each to gain the same 90% advantage over the Bitcoin network.
So yes, it would make it more difficult and expensive for an attacker to stage a disruption using ASICs.
Why would you need 5,000 TH/s to attack 500 TH/s? If anything, spending a few million to develop your own ASICs and using significantly less power and space seems a lot better than buying 50,000+ GPUs and the power and space requirements that go along with that. Bitcoin is quickly going to approach the spot where there is no profit to be had in mining which means sales of ASICs will stall, especially after the award halve. Who is going to buy specialized hardware to not see a return? GPU mining is only so popular because most everyone already has one. Once ASIC power is the norm, the network can only support so many ASICs per price of a bitcoin. Either transaction fees have to start rising significantly, or the value of BTC has to keep going up for the network to be continually better protected.
PS - The price from existing ASIC manufacturers should probably drop significantly after they recoup the startup costs.
Ok, I was unaware of the specifics of encrypted traffic. Someone had mentioned it would be possible to continue using Bitcoin via encryption, so I was merely passing that statement along. Apparently, that statement was wrong.
I was just using the 90% as an example. You could change it to 50% and redo the maths if you like. I agree that spending money on ASICs to attack the network would be much cheaper and easier than spending money on GPU's to do the same, which is why I believe it is imperative that the Bitcoin network start using ASICs.
