Love this format. Walking through selfish mining step by step is exactly how it should be taught.
One thing that surprised me when I first dug into the original Eyal & Sirer paper from 2014: the profitability threshold for selfish mining isn't at 50% hashrate like most people assume. They proved it's actually around 33% (1/3 of total hashrate). Below that, the selfish miner wastes more blocks than they steal. Above it, they start earning disproportionately to their hashrate share because they can strategically orphan honest blocks.
The intuition is wild when you think about it. At exactly 1/3 hashrate, a selfish miner earns the same as honest mining. But at say 40%, they're earning more than their "fair share" of 40% of rewards. The honest miners are subsidizing the selfish miner's orphaning strategy without realizing it.
What made it even more interesting was the follow-up work by Nayak et al. in 2015 on "stubborn mining" variants. They showed that if the selfish miner also controls some fraction of the network's connections (so they can sometimes win block races even without finding the next block first), the threshold drops below 1/3. That's the part that actually worried people, because network topology advantages are a lot easier to get than raw hashrate.
Looking forward to Pt. 2 where I'm guessing you'll formalize the private chain strategy. The Markov chain model for this is elegant once you set it up.
Love this format. Walking through selfish mining step by step is exactly how it should be taught.
One thing that surprised me when I first dug into the original Eyal & Sirer paper from 2014: the profitability threshold for selfish mining isn't at 50% hashrate like most people assume. They proved it's actually around 33% (1/3 of total hashrate). Below that, the selfish miner wastes more blocks than they steal. Above it, they start earning disproportionately to their hashrate share because they can strategically orphan honest blocks.
The intuition is wild when you think about it. At exactly 1/3 hashrate, a selfish miner earns the same as honest mining. But at say 40%, they're earning more than their "fair share" of 40% of rewards. The honest miners are subsidizing the selfish miner's orphaning strategy without realizing it.
What made it even more interesting was the follow-up work by Nayak et al. in 2015 on "stubborn mining" variants. They showed that if the selfish miner also controls some fraction of the network's connections (so they can sometimes win block races even without finding the next block first), the threshold drops below 1/3. That's the part that actually worried people, because network topology advantages are a lot easier to get than raw hashrate.
Looking forward to Pt. 2 where I'm guessing you'll formalize the private chain strategy. The Markov chain model for this is elegant once you set it up.