Combining Segwit2x and BIP141, is BIP91 the best option to solve the scalability problem?

Last week, Bitmain shocked the Bitcoin community with a press release detailing their plans to hard fork in the event that BIP148 (UASF) gains traction. The day after the press release, Segwit2x merged a pull request to make client software compatible with BIP91. After that, a group of miners announced that they would support Segwit2x starting June 19.

In this post, I will describe the series of events that have occurred over the past 5 days and explain the subsequent developments. I hope to shed some light on how BIP91 works and how Segregated Witness is activated on the network, and give you an idea of ​​what is needed to avoid a soft fork on August 1st.

Bitmain’s press release

I’ve read the press release, but I just saw what Charlie Shrem said on Twitter that I missed one thing about Segwit2x:

Bitmain said that the New York Consensus takes precedence over all hard fork solutions. Bitmain prefers to activate Segwit2x (i.e., the New York Consensus) before BIP148. One of the main reasons is that Segwit2x requires segregated witness, so if segregated witness is locked in the network, BIP148 can do nothing.

So you might naturally ask, how does Segwit2x get activated? In fact, this question is trickier than it seems.

Segwit2x Activation Mechanism

The New York Consensus states:

We agree to immediately support the following parallel Bitcoin protocol upgrades, which will be deployed simultaneously and build upon the original SegWit 2Mb proposal.

1. The segregated witness activation threshold is set to 80%, and bit 4 is used as the signaling method;

2. A 2 MB hard fork will be implemented every six months;

The press release does not discuss the technical details of Segwit activation, except that "the Segwit activation threshold is set at 80%, and bit 4 is used as the signaling method." You should know that the Segwit scheme, BIP141, has an activation threshold of 95% and a signaling method of bit 1. There is a significant difference between the two, so the community is wondering, are the two compatible? Does supporting Segwit2x mean opposing BIP141?

What if Segwit2x successfully activates Segregated Witness, but nodes running BIP141 believe that Segregated Witness is not activated?

To solve this kind of problem, James Hilliard proposed a clever mechanism in the Bitcoin developer mailing list (interestingly, he proposed this mechanism one day before the New York Consensus was released). It is this clever mechanism (BIP91) that opens a new chapter in our story.

Combination of two SegWit activation mechanisms

BIP91 cleverly changes the 80% threshold to 95%. The specific method is as follows:

Both bit 1 and bit 4 signaling are acceptable.
If 269 out of 336 blocks send either bit 1 or bit 4, BIP91 will be locked.
In another 336 blocks, BIP91 will be activated.

After BIP91 comes into effect, blocks that do not send bit 1 (i.e. blocks that support BIP141) will be rejected by the network.

The last point should sound familiar. BIP148 does the same thing. After "forced support", the entire chain will support BIP141 (Segregated Witness) 100%, and it will be locked and activated. The difference between BIP91 and BIP148 is that BIP91 first requires 80% of miners to agree. Last Thursday, BIP91 was merged into the Segwit2x codebase, which aims to ensure that Segregated Witness has 80% miner support at the same time as activation.

Miners Roundtable Resolution

The Chinese Miners Alliance issued a press release last week announcing that they will run the Segwit2x software. Since the software is not yet available, they will not send bit 1 or bit 4, but will write the "NYA" (abbreviation for New York Agreement) mark in the Coinbase string. These miners represent 80% of the network's computing power, and once the Segwit2x software is ready, they should send bit 4 as a signal.

What it means to you

In order to avoid the soft fork on August 1, the following conditions must be met:

1. Segwit2x must be launched and in the hands of miners
2. At least 269 out of 336 blocks send bit 1 or bit 4 signals (about 80%), so that BIP91 can be locked
3. Another 336 blocks must pass before BIP91 can be activated. After that, every block must send a bit 1 signal (Segregated Witness or BIP141) to support

If all of the above conditions are met before August 1st, BIP148 will do nothing and the soft fork will not happen. What is not clear is whether we can complete the second step before August 1st. Will BIP148 supporters think this approach is "good enough" and choose not to soft fork? Maybe, but so far, BIP148 supporters have not stated their position on this approach. If they can make their position clear as soon as possible, I believe it will be beneficial to everyone.

Wait, there's more

This is a positive development if you want to avoid a permanent split (since it looks like Bitcoin will always stay on one chain). But the battle is far from over. The 2MB hard fork is the bigger and more contentious issue, and it needs to be resolved urgently.

According to the New York Consensus, the 2MB hard fork must be activated within 6 months. What software will be used to activate the 2MB hard fork? Bitcoin Core? It is unlikely, because they did not sign the consensus. Segwit2x? Maybe, but how exactly will it be done? Or Bitcoin Unlimited? BUIP0055 has set October 18 as the target activation date, but they are pursuing a block size greater than 2MB.

In addition, if clients other than Bitcoin Core sign the New York Consensus, then their clients can be used, but how will the code base be managed? Will Bitcoin Core developers be involved? What if Segwit2x proposes to merge with Core and Core refuses?

None of these questions have been answered. We are still not out of the woods.

Summarize

What happened in the past week seems to indicate that the activation of Segregated Witness and the network fork on August 1st will not happen. In the short term, this is good news. But in the long term, it is still unknown whether the upgrade of large blocks through hard forks can be carried out smoothly. We can't be too optimistic at the moment, and there are still many battles to fight.