Topic: A variant on the Internet failure question. - page 2. (Read 339 times)

In the scenario of two chains being separate for a period of months or even years, and with the default mempool limit of 300 MB, then the vast majority of transactions in the losing chain will not be returned to the mempool, but rather will be dropped altogether and therefore trivial to double spend. Only a minority of the highest paying transactions will be returned to the mempool and be available for miners. All the transactions which are based on these now-dropped transaction which the winning nodes have no knowledge of would be invalid as far as the winning nodes are concerned.

If a small proportion of the hash power, say less than 20-30% splits off, then over an hour they would almost certainly be the losing chain and simply rejoin the winning chain. Bear in mind, of course, that they are only going to be finding 1-2 blocks an hour on average with this much of the hashrate, so it would only be overturning maybe 1 confirmation rather than 6. If they forked, then I suspect over the coming days or weeks the users, nodes, and miners on that losing chain would gradually move back over to the winning chain and the losing chain would disappear. If somewhere around 40-50% of the hash power splits, then it becomes more complex. But again, over an hour and with half the hashrate each, each chain is only going to be finding ~3 blocks on average. One chain might not even find a single block if they are unlucky.

If something like that happened, Russia + China would ban Bitcoin and everything related to it, and Bitcoin on this network would either be abandoned or forked to be more like Monero - more private and mined on home computers, because big mining farms would be impossible under ban.

Theoretically, it could be possible to make an algorithm to merge both chains into a new one, minus the inputs that were spent on both chains (but if this percentage is huge, that would be a big problem). But I doubt that existing holders would be happy about it, so such fork would most likely be doomed to fail.

Indeed, if such a split is to happen for - let's say - more than 1h, a hard fork is a must. Reconciliation is no longer possible, else the result would be much worse than after a 51% attack.
So on the scenario from OP, I think that by far the only realistic solution is that the chains cannot be merged anymore. We will have 2 types of Bitcoin then and I don't even want to think what would this mean for the price.

A sudden segregation of WWW will result in the blockchains being split with the connections to external addresses being restricted and the chain being able to continue within its own bubble only. If a state is willing to do this, they can easily either ban Bitcoin completely or basically negate any benefits Bitcoin provide, due to the effective sybil being done on the citizens.

Whether they will piggyback on Bitcoin or make their own remains to be seen. Without completely removing encryption, Bitcoin can still function within the bubble itself but it won't be able to transmit any information outside of it, provided that there isn't any efficient system to maintain consensus with the other systems.

If this possibility really happens and Russia/China create "their own internet",I don't believe that the Russians or Chinese will allow blockchains or cryptocurrencies.Maybe Russia and China will impose their own CBDCs and ban all the cryptocurrencies.Countries that have authoritarian regimes will never like and adopt the idea of a decentralized peer-to-peer payment system,a system they cannot control.
The theory about splitting the blockchain into two sounds very interesting,but I don't think that this will happen in reality(unless we are talking about soft forks/hard forks).

In a reorg, the unique transactions in the shorter chain are returned to the mempool and re-confirmed in subsequent blocks.

Even if a UTXO is only spent on one chain, as would be the case in two completely separate internets as Jet Cash has described, then there will still be a huge number of transactions and their descendants which would immediately become invalid if the chain with the lower work was abandoned.

If a fork to keep the chains separate did not occur, then the value of bitcoin on one or both sides would fall dramatically I would think. No one is going to want to accept a currency whose transactions could be reversed suddenly and without warning even when several hundred blocks deep.

The primary problems that reconciling the split cause are the invalidation of block rewards and the possibility of double spending.

• Block rewards in the shorter chain would be invalidated. That could create a huge problem if the split is longer than 100 blocks and the miners spend their block rewards.
• If a UTXO is spent on both chains but differently, then one of the transactions (and all of its subsequent UTXOs) will become invalid when the chains are reconciled.

For that reason, I believe that the users of one or both of the chains would probably do a hard fork in order to avoid this situation.

Thanks for that explanation. I hadn't associated hash rate with the chain length.

I agree. In the highly unlikely scenario that there would be two separate internets for even a period of a few months, then I suspect one or both chains may deliberately create a fork which would keep the two chains separate in the event that the barrier between them was removed.

Because of difficulty changes.

Let's say the internet does split in to two. After a couple of readjustment periods, the difficulty on both chains will have readjusted to the decrease in hash rate, since a proportion of the hash rate has just disappeared from each chain and is mining a different chain. Let's say Chain A is left with 70% of the global hash rate, and Chain B is left with 30% of the global hash rate. The difficulty on Chain A will adjust to around 70% of what it was, while the difficulty on Chain B will adjust to around 30% of what it was. Both chains will continue to mine blocks every 10 minutes on average, but these blocks are not equal in terms of the amount of work done. It requires significantly more work on Chain A to find a valid block than it does on Chain B.

There will of course be random variance between the two chains, so they will not find blocks at exactly the same rate, and as we know, hash rate can come and go which also affects the rate of block creation. If these two chains were to merge and Chain B happened to be longer, it would not necessarily have more proof of work since it has done less work overall than Chain A has, even with more blocks, since these blocks were much easier to find.

That is what I would suspect, but if it went on for a year of so (say), I can't see the losing chain allowing a merger, especially if the newly created coins have been traded. I can see  this happening if autocratic governments like China want to control Internet use.

Just out of interest - how could the shortest chain have the greatest proof of work?

If, during the split, neither side makes any changes to the protocol which would invalidate blocks from the other side, then whenever the two chains were reconnected the chain with the most proof of work (note - not necessarily the longest chain) would be the winner. All the nodes on the losing chain would switch over to the winning chain, and all the blocks the losing chain had mined during the split would be invalidated.

If one of the sides did make any changes which would prevent this from happening, then there would indeed be two blockchains going forward, and everyone would end up with the same number of coins they originally had on the other chain at the time of the split.

I don't believe that the Internet will suffer a world-wide failure for any length of time. What could be a possibility id that it splits in two. Perhaps Russia and China will control one with restricted access, and the five eyes will operate another. If there is no possibility of communication between the two, and miners continue to operate on both nets ( but not the same miners). Then this could create two blockchains. How would "they" handle a reconciliation? Would existing owners end up with coins on both chains, as I can't see a way of merging the chains?