Possibility and Legitimacy of a Rollback

[qwk]

First things first: I am neither in favor of nor advocating for any rollback whatsoever.
I am playing "devil's advocate" a little, and pondering the potential feasibility and/or legitimacy.

In light of a recent hack of a Bitcoin exchange, the option of a rollback has been brought up, and quickly discarded.
The obvious reason in that case was the (comparably) low value of the stolen coins and the timeframe between the hack and a potential rollback, which made it clear that it wouldn't be worth the hassle. But it got me thinking.


A rollback in the form of a double spend against a hacker's transaction(s) is possible, if expensive. [1]
Such a rollback would simply create a new "longest chain" which would replace the old one.

In Bitcoin, the longest chain is the "true" chain.

Therefore, based on the assumption that the defining property of Bitcoin is the fact that the blockchain (or better: the network consensus) is the unerring arbiter of the "truth" about "who owns what", a rollback is not only feasible, but also legitimate.

A "true" Bitcoiner can not in good conscience vote for a "none-aggression pact" opposing such a rollback.
If the economic reality of the Blockchain (i.e. the incentives for the miners) make a rollback a reality, he would have to accept it.


Now, I've already discussed this at length in my local forum, and the obvious reaction of everybody was "no way, that's a sin, blasphemer!", which is more than understandable and was also my first thought.
But I've come to the conclusion that at least for me, the implications of a "non-aggression pact" against rollbacks are probably worse for Bitcoin than the (very) low possibility of a rollback in the case of a real, major hack.


I also assume that "behind closed doors" there might already be agreements in place between large exchanges and large miners to perform such a rollback in the case of a really substantial hack. But of course, that's just an unbiased assumption.


The real point is:
- a rollback is possible and not completely unlikely, given a large enough hack
- it is also "in line" with the consensus rules of the Bitcoin network
- the longest chain is the true chain
I.e., a rollback is a "feature" of Bitcoin.

Discuss!

[qwk]

[1] Cost estimate for a rollback

Let's assume a scenario where a rollback could be considered much more likely to occur:
1. Exchange "Octopus" is being hacked, 100.000 BTC are stolen.
2. Octopus' admins notice the hack "right away" after a measly 20 confirmations on the blockchain.
3. Octopus' CEO gets on the phone and calls his CEO friends at some of the major Mining Pools.

Now it's obvious that Octopus could create a double spend of the transaction(s) that stole the coins and submit those transactions to the miners.
The miners could decide to specifically mine a new longest chain starting with the double spend, thereby undoing the theft of the coins.
Obviously, Octopus would have to pay the miners for this rollback.

The direct cost of such a double-spend-rollback can be estimated as
- block rewards of all rollbacked blocks
- fees of all roll-backed transactions

Also, to help acceptance of the rollback, it would probably be wise to give all block rewards that were in the "original chain" to their respective "owners", as well as any transaction fees. That way, if I were a miner who received a few coins in the original chain, I would also receive the same amount in the new chain, making it less likely for me to oppose the rollback. [2]

All in all, I estimate the ground cost for a rollback at somewhere on the order of 30 BTC per confirmation.
Of course, the rollbackers would also expect to get their share, and probably they wouldn't accept a pittance.
So, as a lower bound for the real cost of a rollback, I estimate something on the order of at the very least a hundred BTC per confirmation.

Now that doesn't include any potential "side costs" for the rollback, e.g. the implications it would have an the resulting effect on the exchange rate of Bitcoin. That's much harder to estimate, but let's just say that there is probably some point where the miners (or rather: a majority of them) would consider it more profitable to commit themselves to the rollback than to stay true to the original chain.
Maybe a thousand BTC per confirmation will do the trick.

[2] Gentlemen's Agreement

In order to lower the resistance of other participants of Bitcoin against such a rollback, it would probably be wise to agree on some ground rules:
1. revert as few transactions as possible
2. re-instantiate all other transactions of the rollbacked chain
3. grant all block rewards & fees in the new chain to their respective recipient addresses in the rollbacked chain
Of course, this cannot be enforced, but would probably be self-enforcing, since it would most likely be cheaper to avoid resistance rather than fight it.

[pawanjain]

It wouldn't be wise to perform a rollback since it would mean that a majority of miners combined would have the power to dominate the blockchain. It would mean that the blockchain can be centralized just as banks are.

Even if Octopus' CEO tell his CEO friends to rollback it wouldn't be guaranteed that they will do just as he said.
There is a possibility that his CEO friends would also get the huge fees from Octopus' CEO and perform a rollback filling their own wallets instead of reverting that particular transaction for which the rollback had to be done.

The longest chain is the true chain. So even if a rollback is done, it will have to override all the transactions from the original chain that were performed after the rollback was done. It will take a good amount of time doing that since the original chain will still be confirming new blocks. All of these new blocks should then be added to the new chain as well because if not then it would be a loss for those people whose transactions are getting confirmed on the original chain. Things will just get more and more complicated and this will in turn increase the cost of performing a roll back.

So in my opinion, a roll back is not a wise thing to do as it will make things more complicated while increasing the cost of the roll back and more importantly, leaving the blockchain to be controlled by the miners.

[qwk]

It wouldn't be wise to perform a rollback since it would mean that a majority of miners combined would have the power to dominate the blockchain. It would mean that the blockchain can be centralized just as banks are.

The point is:
1. they already have that power
2. centralization is not a result of this

The reason why this is not centralization is quite simple: it's costly.
The basic premise of the Bitcoin blockchain

6. Incentive
[...]
The incentive may help encourage nodes to stay honest. If a greedy attacker is able to assemble more CPU power than all the honest nodes, he would have to choose between using it to defraud people by stealing back his payments, or using it to generate new coins. He ought to find it more profitable to play by the rules, such rules that favour him with more new coins than everyone else combined, than to undermine the system and the validity of his own wealth.

still holds true.

The security of a transaction is not determined by the fact that it cannot be rolled back, but rather by the fact that it is costly to do so.

Even if Octopus' CEO tell his CEO friends to rollback it wouldn't be guaranteed that they will do just as he said.
There is a possibility that his CEO friends would also get the huge fees from Octopus' CEO and perform a rollback filling their own wallets instead of reverting that particular transaction for which the rollback had to be done.

That can be mitigated by Octopus' CEO.
He could simply "pay" his miner friends by creating a double spend and some follow-up transactions with huge fees.
I.e. he first creates a double spend of the stolen coins to address A.
2. Tx A -> B 100 BTC fee
3. Tx B -> C 100 BTC fee
etc. pp.

The longest chain is the true chain. So even if a rollback is done, it will have to override all the transactions from the original chain that were performed after the rollback was done. It will take a good amount of time doing that since the original chain will still be confirming new blocks.

Yup, that adds to the cost, making it highly unlikely to be able to pull off a rollback, if you don't have overwhelming support.
I.e. a simple 51% majority most likely won't do the trick, you'd rather go for something on the order of at least 75%, preferably more.

And yes, in the case of a "long rollback", i.e. a rollback after more than just a few dozen confirmations, the Bitcoin network would effectively be stalled for quite some time.

[qwk]

If such a rollback would occur,
1. It proves beyond a doubt, Control of bitcoin is Centralized to a group willing to rewrite the chain for their own personal reasons.

No. Absolutely not.
They will have to bear the cost (and then some), i.e. they have no meaningful "control" that isn't already there.

The loss of faith in BTC as a non centralized, autonomous payment platform would cause a massive selloff.

Possible, and also likely.
OTOH, assume 100.000 BTC stolen from an exchange.
What would that do to the market?
The question is: which one is more costly for the miners?
1. huge selloff of 100.000 stolen BTC, and then panic sells
2. huge selloff because of "lost trust", but compensation from Octopus' CEO

2. Users that sent out payments, before the rollback that had been accepted by vendors, only need to leave their wallets offline during the rollback and restore a wallet.dat before the rollback , and if they don't manually resend the coins , they just performed a double spend with the help of the centralized miners, which would be a major catastrophic event in bitcoin history, as their could be 100s of double spends.

No.
1. Miners would (should) most likely include all transactions other than the hacked coins in the new chain
2. If such a rollback were to be performed, it wouldn't go unnoticed, i.e. people would probably "wait it out" before accepting your payment.
In the end, double spends as side effects are highly unlikely.

Problem is the core devs in another bad decision on their part , removed the alert code in the bitcoin wallets that could warn users of problems, by doing so , they have left the bitcoin community deaf dumb and blind until they receive news from a news source or other communication, so fast responses are a thing of the past.

The first thing you would most likely notice would be very long confirmation times.
If the rollback were to happen with e.g. 90% of the hashing power, your transactions would confirm for the first time after more than one and a half hours.
Also, any "major" players who accept Bitcoin payments would likely be monitoring the "media" closely, and simply shut down service for a while (or bear some risk, or negotiate deals with miners, etc. pp.).
I doubt that it would really lead to a substantial amount of confusion because of a lack of communication, but of course, the possibility exists.

3. By proving bitcoin has centralized control, the world governments may then pass laws requiring the miners to censor certain addresses submitted to them by the government bodies or face criminal penalties.

Again, no.
1. centralization doesn't follow.
2. whichever government wants to control the miners can already do so.
They're right there, and they can't just move their farms.

The Mythology that bitcoin is not centralized is it's only protection from the corrupt world governments.

That "myth" is in no way in danger.
But most likely the mis-understanding of that myth is.

Proof otherwise, means miners will no longer be able to deny government requests for censorship.

They already can't.

[Tipstar]

Such rollback had been done in the history of Bitcoin and is a useful tool in drastic circumstances.
But consensus of simple majority of miners could not be considered as the agreement of the bitcoin community.
There should be some sort of voting mechanism in between miner and holders to reach a consensus as these two groups are the ones most affected.

[sheenshane]

Well, I can tell that this always happens in each and every industry where from the first days, the thing has a very small value and to the later part it will have a great value. Because who would know? Who would know that one thing will have a chance to have a great value for the upcoming years?

Like today, it is really hard to tell if the bitcoin would still go high. There is no accurate speculation and predictions because something might happen that would crash the whole system. The lesson that I learned from what I see is to become realistic but still HODL.

[TimeBits]

Why not just have a withdraw queue on your exchange? Max btc withdraw withing 1 minute = .5 btc, Hacker might get 1-100 btc then before reports come in.
Why not just have a max account withdraw queue on your exchange? Only 2-3 withdraws at a time to be processed manually and a way notice they are from the same i.p. or when mass accounts are trying to withdraw at the same time.

Why not, just not give another man your bitcoins to begin with, it defeats the whole purpose of being your own bank.

A rollback is like a kid spilling milk and expecting to go back in time to clean up the milk and making everyone else go back in time to, rather than grabbing a cloth and cleaning it up. We all cannot go back in time because of ones mistake.

Roll backs should be out of the question unless it was like that time they forked the coin and made 8.2 billion bitcoins or something.

[pooya87]

bitcoin like everything else has certain characteristics that defines it. for example a car has 4 wheels and if you cut one, it can still run but it turns into garbage. one of the characteristics of bitcoin is immutability of its blockchain. it has nothing to do with ethics anymore, if you rollback you are removing that characteristic and bitcoin is not longer bitcoin. i will go as far as saying it will turn into garbage and every user and miner and node owner knows this well. so nobody would want such thing.

The real point is:
- a rollback is possible and not completely unlikely, given a large enough hack

i disagree with this and i say this is wrong because you are "assuming" that there exists a (singular) miner that "controls" more than 70% of the mining power and that is simply wrong. there are pools that many (multiple) miners are sharing the work but the pool is not in control of their hashrate, there are big farms that have high hashing power but they are nowhere near enough percentage to even make a difference let alone to a 51% attack and a successful rollback.

remember that a 51% attack (or calling it a pretty word like rollback or getting stolen money back) is going to crash bitcoin and the miner's ASICs is going to probably turn into bricks. in other words the "payment" that the "Octopus" pays is probably the last time they make a decent amount of money with bitcoin.

[Beerwizzard]

Those stolen bitcoins are still gonna get back to the market and the entire community shouldn't worry about that.
People gave to Binance their money. Binance is an official and totally legal exchange so if they lost someone else's money then they have to give it back. It is only their problem and I bet that top crypto exchange can afford itself to repay this money. It is only a problem of Binance, not the entire cummunity.
The other thing is that mining community is not that smart and they don't really care about the others. It mainly consists of greedy (and pretty often retarded) asians and for obvious reason, I don't trust them.