[HWR_14]

You wouldn't have to reverse all the transactions. You could trivially create a fork (which has to be longer, and therefore have more transactions available) that includes every transaction but one from the blockchain. Well, that is you can create that fork as trivially as you can create any other fork.

[leifg]

Sure maybe. But that only really works if few (if not all) entities have control over the consensus mechanism.

On a regular PoW blockchain you will have to recalculate all the hashes according to the difficulty which will up to the miners.

But even if you could, it's an absolute technical nightmare.

To build an analogy that somehow fits. If you have git repo and you find out that a particular commit that you want to undo, what do you do?

- Rebase all changes to an earlier commit, remove the faulty commit and recalculate all commit hashes that follow it.

or?

- Create a new commit that reverts the old commit.

In reality you opt for option 2 99.99% of the time. The only reason you would ever want to remove a commit from history is if you accidentally exposed information to an audience that is not supposed to see it.

[HWR_14]

When you first responded to chippiewillie you talked about how forking would produce a reversal of all the transactions. That's not true, but it is what you identified as a "much deeper problem"

[leifg]

My apologies, I used “reverse” and “invalidate” synonymously.

Nevertheless on a public blockchain all transactions would be invalidated and that indeed is a problem.

Because everyone who received coins would have to wait again for n confirmations in order to be sure they got their money. In theory nobody should be able to add a double spend transaction to the pool but I wouldn’t bet on it.

That’s what I mean with technical nightmare.

You would have to make sure to properly identify all transactions. Possibly take down the system, exclude a single transaction. Make sure that the miner who will find the next block will include the right transactions. Make sure of that for the following block. I don’t see that happening with a large coordination effort, meaning: centralization.

And when you come to that conclusion you should probably take a step back and rethink “why are we doing all of thatch blockchain stuff when we need to rely on a central authority?”

[HWR_14]

> when you come to that conclusion you should probably take a step back and rethink “why are we doing all of thatch blockchain stuff when we need to rely on a central authority?”

I think blockchain is going to eventually die for that exact chain of reasoning.