A new research paper argues that bitcoin mining pools can only co-exist peacefully if individual miners regularly switch pools and if pools erect sufficient defenses against attacks.
If pools don’t exhibit either of those properties, then they are liable to attack each other, the report finds.
The paper’s authors, Aron Laszka, Benjamin Johnson and Jens Grossklags, investigated the long-term effects of competition between bitcoin mining pools in a draft version of a study titled, ‘When Bitcoin Mining Pools Run Dry‘.
The final version of the paper will be presented at the Workshop on Bitcoin Research at this year’s Financial Cryptography and Data Security conference in Puerto Rico on 30th January.
The paper uses game theory to look at the impact of attacks between two theoretical mining pools. One pool is deemed more ‘attractive’ than the other, whereby myriad factors like fees, technical structure or public relations savvy find it well-regarded among miners and investors.
How the game works
Pools may spend resources on “productive” or “destructive” investments.
In the former, a pool would add to its overall computing power, trying to beat competitors. A destructive investment, in turn, could come in the form of a distributed denial of service (DDoS) attack on a rival that reduces that entity’s computing power.
The authors devised a game consisting of two players: a bigger mining pool and a smaller pool. In each round of the game, the pools can either attack each other or not. Over time, the size of each pool should be affected by the attacks, as miners choose to “migrate” from one pool to another.
The game was set up to discover two things, according to the paper:
“We study two important questions: the conditions under which the mining pools have no incentives to launch attacks against each other, and the conditions under which one mining pool is annihilated by the attacks.”
Among the findings were signs that pools were capable of cohabitation if individual miners swapped mining pools often and each pool had in place defensive measures that ensured an attack would be too costly for the other side to mount.
Peace can also be achieved between rival pools if either or both pools showed a low level of attractiveness. In practice, the authors write, this means that neither pool would attack the other if they were both busy trying to entice miners to join them.
Figure (a) shows if miner migration rate and the unit cost of attack are high, then peace is possible. Figure (b) shows peace is achieved if either or both pools has low attractiveness. Souce: Laszka et al, 2015.
When mining pools attack
When miners are actively migrating and defensive counter-measures aren’t in place, then the likelihood of an attack increases.
Additionally, the attack will probably come from the less attractive pool. This is because the less attractive pool has more to gain from an attack, whereas the more attractive pool has less reason to change the status quo.
The other scenario in which an attack takes place is when one pool is highly attractive to miners and the other is not, again causing the less attractive entity to strike and ostensibly drive miners to process blocks in their corner of the network.
The attack scenarios contain a silver lining for the bitcoin network, the authors say, as such actions can act as a regulating force that keeps hashing power distributed.
They noted:
“While attacks are generally harmful to the bitcoin ecosystem, they have positive effects in this context, as they prevent one pool from growing too large.”
Figure (a) shows an attack is likely to happen if migration rate is high and unit cost of an attack is low. Figure (b) shows that an attack is likely if either pool becomes highly attractive. Source: Laszka et al, 2015.The authors highlight two earlier studies that examine competition dynamics between mining pools.
An earlier paper by the same authors, plus Marie Vasek and Tyler Moore, used game theory to look the short-term effects of mining pool attacks. In the short-term, large pools are likely to be both targets and aggressors, that paper found.
Another paper, by Vasek, Moore and Micah Thornton, found 7% of all mining pools had experienced at least one DDoS attack between 2011 and 2013, and that big pools were more likely to be attacked than small ones.
Both papers were presented at the inaugural Workshop on Bitcoin Research held in 2014.
White flag image via Shutterstock