Yesterday I published a post, claiming that the current power requirement is mining bitcoins is 3GW, or about one power station. This post was sourced against data from the Bitcoin community but I have been told via Twitter that this number might be erroneous because it assumes use of old technology, and that actual power requirement for mining is rather in the order of 50MW. (Note: an intro to bitcoin mining is here, and a more simplified version is here, and it is also explained in this video lecture)
In the following I will assume that the current requirement of the mining pool is 3GB as this higher figure weakens my argument, so please keep in mind that it is possible that the actual power requirement of the pool is 50MW which would make an attack even easier.
(UPDATE 20/Nov 11:30 – apparently the correct number to use is 50MW, the number quoted on blockchain.info has been off by a factor of 50x for the better part of the year)
What does one need to attack the bitcoin infrastructure? There are various ways of doing that, one of them being controlling the majority of the mining pool. What does someone need have to control the majority of the mining pool?
- The same hardware mining power as the rest of the pool (slighly simplified, the same number of ASICs)
-
The same electric power as the rest of the pool (ie the aforementioned 3GW)
-
An Internet connection
Point (1) is currently an issue because of availability of ASICs but this will not be the case forever (and someone determined enough can probably build his own), point (3) is trivial, and point (2) is having control of a power station (or 1/50th of it, as the case may be). Someone who has all of those can control the mining pool – maybe not in the sense that they can permanently steal someone’s bitcoins (I am not sure about this, comments would be appreciated) – but certainly in the sense that the bitcoin infrastructure would disappear as long as the attack lasts. If an economy is actually relying on BTC, rather than using it as a secondary payment system, even a short attack of a few days is probably all it takes to bring it down pretty conclusively.
So whilst it is not trivial to pull off, any government (or any sufficiently powerful criminal organisation) has access to those resources, and therefore could disrupt the bitcoin infrastructure if they desired to do so. Note that this risk actually increases over time when the mining reward goes down and there are fewer miners online to protect the system (this is described in more detail here).
So somehow we have the worst of two worlds here: on the one hand we spend a significant amount of resources on an ongoing basis to protect the system, and on the other hand, everyone willing to spend the same level of resources even just for a couple of days can bring the entire system down.
Ultimately, as it is much cheaper to destroy the system than to protect it there is a significant risk that it gets destroyed. The only protection against this is the power of a state – but if one is ready to rely on this one might as well get for a less resource-hungry, trust-based central bookkeeping system.
