Quote from: NewLiberty on October 03, 2014, 12:30:14 AM

Quote from: TheFascistMind on October 03, 2014, 12:07:55 AM

Quote from: NewLiberty on October 02, 2014, 02:05:00 PM

Quote from: TheFascistMind on October 02, 2014, 01:50:53 PM

Afaics, ignoring decentralized checkpoints should be plausible since the attacker would control the decentralized consensus.

Ignoring centralized checkpoints is not so feasible, since you've got to convince others not to run the reference client.

Applying the decentralised checkpoints isn't based on consensus though. It is a decision each miner may make on their own.
They can also be delivered out of band, so DDoS pfft.
It allows each miner to select which chain they like.

So if BCX forks with TW or other method, that fork ends up back where it started, back in the sandbox along with the little shovels, buckets, and Stoli empties.

There are certain further improvements to this innovation that may yet come, but the rapid response to the only plausible indicated threat (which isn't even all that plausible IMHO) remains an underrated achievement. BCX shares some of the thanks/blame for this forced evolution.

You said to me upthread you like disagreement. So please pardon that I need to point out that afaics miner's choice doesn't resolve the issue that once a 51% fork has run for a while and many users get their transactions intertwined with it, you can't untangle it to revoke it any more, especially given the anonymity with the ring signatures.

Sorry.

(note I wrote this already far upthread)

Thank you for this. I'd missed it so I appreciate the extra effort.

That would be true, if not for the fact that the MONERO DEVS ALREADY DID JUST THAT!

When we had that bad transaction a couple weeks back, remember?
Chain was in contention for 30 blocks, and all transactions replayed from the two chains except the coinbase ones (which in that case there weren't any to speak of) and the chain was reintegrated within that period (30 minutes or so). So not even a fork in the end.

Dagnabbit I was hoping to learn something with this disagreement.

Maybe they won't be as good next time, but they have had some practice already, and I'm guessing with the generous warning they will be better prepared if it ever happens again. My understanding is that they don't get unwound, they are retransmitted, so no user intervention needed.

30 blocks is probably not enough for a Gordian knot. What do you do when you people claim very important transactions that they don't want unwound? How do you identify who is who they claim to be as a sender in order to pick and choose which transactions to retain and which to unwind?