,
12 tweets,
2 min read
Read on Twitter
There have been several fundamental changes to the incentives in Bitcoin since its launch.
1. We now know that selfish mining is a possibility against PoW coins. 51% honesty isn't sufficient -- depending on the coin, 66% to 100% need to be honest.
Luckily, most miners have been honest so far, but the threat is ever present, esp for low hashrate coins.
Luckily, most miners have been honest so far, but the threat is ever present, esp for low hashrate coins.
2. Satoshi did not explicitly mention pools. That changes the game theory substantially, especially if the miners within a pool are not checking the work of the pool operator. A single bad actor can rope in honest-but-not-tech-savvy actors into a byzantine attack.
3. Satoshi only considered a world where there's a single coin. When multiple coins share the same hash function, the dynamics are different. Hashwars, difficulty raising attacks, and death spirals are example attacks from this domain.
4. Hash rental markets make temporary attacks possible. These did exist in 2009, became unfeasible with the shift to ASICs, and are once again a possibility. We saw hash rental markets used, in conjunction with some of the other points above, to attack BTG.
5. Satoshi's famous "6 block" confirmation calculation assumes a relatively small percentage of attacker hashpower. As the amount of available rental hash and attacker budget changes, it needs to be recalculated.
If a pow attacker has 49%, # of confirms required is infinity.
If a pow attacker has 49%, # of confirms required is infinity.
6. Finally, people are layering applications on top of the chain, eg Omni and LN. If miners are in bed with attackers on these systems, the dynamics change. E.g. LN can be attacked via censorship. If Omni holds more value than block rewards, the mining dynamics are different.
7. Satoshi did foresee, but was wrong on, what happens when transaction fees exceed block rewards. He indicated that the system would operate as normal. This is incorrect, as miners have an incentive to snipe each other's transactions instead of collaboratively building a chain.
Some of these attacks may not arise in practice. Some are fixable. But the original question asked how the incentives are different from what SN foresaw in 2009. This is my quick summary of how the world has evolved.
Oh, let's not forget:
8. SN seemed not to foresee just how much energy would be spent on mining, and how much value would leak out of the system to the power companies. That changes the dynamic.
8. SN seemed not to foresee just how much energy would be spent on mining, and how much value would leak out of the system to the power companies. That changes the dynamic.
Ok 9. SN did not seem to consider the inherent centralization of industrial scale mining. Mining rig manufacture, access to rigs, and access to cheap electricity are inherently limited. This leads towards a small set of industrial miners.
Overall, the world is a different place than it was in 2009. It's a testament to SN that his tech is still around. The original question asked about changes to miner game theory and SN's assumptions. I chronicled the ways in which the original ideas have had to be modified.
