Bitcoin Is Broken | Hacker News Reader

Bitcoin Is Broken

hackingdistributed.com

32 pointsjulien8 comments

I've read the paper. The title is alarmist. "selfish" mining seems to work, pools should implement it as fast as possible. When a majority of them do, the problem cancels out. The given solution suggest propagating all chains and choosing a random one to mine on. Their proposed solution is immediately exploitable, as a "selfish pool" could disseminate as many competing forks as possible to dilute the processing power of the "honest" system then collaborate on a single chain.

That was my thinking as well. All pools should behave selfishly. If they all behave this way, then there's no incentive for the miners to move towards a single pool, therefore avoiding centralization. Am I missing something?

Yes, the strategy works better if you have a lower-latency network at your disposal. What it creates is a winner-take-all situation where the mining pool with the best optimized network, some function of latency-vs-cost, gets a disproportionate reward even when all miners are pursuing this strategy.

There is an incentive to move to the largest mining pool, since that has the greatest chance of discovering a new block and gaining a selfish advantage.

I haven't read the paper, and I'm certainly no expert in game theory, but I don't think we have to worry too much about this. The assumption seems to be that selfish miners will collude to control the network. However, it still seems to be in their best interest to compete. If the coin starts losing real-world value due to a bad controlling group, rational miners will stop helping the controlling group mine.

In addition to spending time/money/resources on mining hardware and power, miners now have to spend time/money/resources on finding and implementing better strategies. This just shows that there's another variable.

(I could be, and probably am, wrong.)

I think their attack doesn't work.

It relies on "publishing their newest lbock" before the others discover it. However, there is no way to know when the others found the newest block. Since they just try some Hash IV (afaik), it is well possible they discover the next block within seconds of the first. So there is an incentive to publish their latest block as soon as they found it.

Also, what I read here and what I agree with, selfish mining seems self defeating which reinforces the first point. The second selfish miner pool should publish its newest block just before the first selfish miner pool, pushing the equilibrium of "latest block release" to ASAP.

Any comments?

What happens when the honest miners fluke it and solve a block early, before the dishonest miners decide to publish theirs...?

Wouldn't this ruin any and all gains the honest miners have, nulling all the blocks they have mined 'ahead'...

I think in the long run (repeat the process hundreds of times), there would still be an advantage to the selfish miners.

Edit: Here's a simulation written by Dan Kaminsky. http://pastebin.com/PfBhLc7r

Reply on news.ycombinator.com