The big news in the Bitcoin world is that there are several Bitcoin-mining ASICs (custom chips) already shipping or about to be launched. Avalon in particular has been getting some attention recently. Bitcoin mining moved long ago from CPUs to GPUs, but this takes it one step further. The expectation is that very soon most mining will be done by such specialized hardware; general-purpose machines will be too inefficient to be profitable. This development raises a whole host of interesting questions.
First, if ASIC mining is indeed so much faster than any current method, and can recover the hardware cost in just a few days as claimed, why are the creators selling them, or even talking about them?! Why not keep the existence of these devices a secret and mine as much BTC as possible before the world catches on? After all, the costs of packaging and distribution will only decrease the margins. There are several possible explanations:
- Maybe the buyers are people who have lower electricity costs (which is a significant fraction of total cost of mining over say a two-year period).
- There is a claim in this thread that the creators of Avalon could only raise money for manufacturing by pre-selling the units.
- Another reason from Butterfly Labs, one of the ASIC companies: “we believe the best way to secure bitcoin and help it to succeed is to get as much product out to as many people as widely as possible. That makes it exceptionally difficult, if not impossible, for any individual, corporation or hostile government to destabilize and wreck Bitcoin from within.” This refers to the “51% attack.” Given the speed of the Avalon chip, for example, only about 300 would be needed to exceed the current hashrate of the Bitcoin network!
None of these reasons are fully convincing to me. In fact, one of the companies, ASICMiner, decided to use their machines in-house for mining, and tried to raise capital for manufacturing by selling shares in the company via a Bitcoin stock exchange, with amusing results that you can read about in the same Bitcoin magazine article linked to above.
The second question that comes to mind is: what does a dramatically better mining technology do to the value of Bitcoin? Note that it’s not like the market is being flooded with Bitcoins. All this horsepower doesn’t mean mining is happening any faster — the system automatically and continually adjusts the mining difficulty so that Bitcoins are produced globally at a constant rate. As a side effect, the amount of BTC that a person or entity can mine is determined not by their absolute computing power but by the fraction of global bitcoin mining resources that they control.
My speculation is that the shift to ASIC is good news for Bitcoin’s price inasmuch as it is a signal that people are taking it seriously, just as the price spiked after a hacker demanded ransom in Bitcoin. The fact that these custom chips are useless for any other computing may be particularly meaningful here. News about ASICs has been coming out for the better part of a year; I wonder if it’s possible to see the effect of an individual piece of news on the price.
The “uselessness” brings me to my third point: is there a way to re-design Bitcoin so that the proof-of-work computations have a useful side effect — say protein folding — in essence making Bitcoin mining a massive grid computing project? This has been proposed repeatedly on forums and in blog posts, but my intuition is that it is not possible, that proof-of-work derives its effectiveness precisely from its uselessness. That said, given how much electricity is being spent on Bitcoin mining, it would be good to have a definitive answer.