Noise Activated Forks (Or How SegWit2X Failed)

It’s become something of a dogma for some in the blockchain scaling debate that users secure the network. By having their nodes signal they won’t accept any blocks which violate any of the rules in their local instance, they force the hand of the miners to do their bidding. Confidence in this newly devised power has only been bolstered by the fact that the majority of nodes were upgraded software the implemented the Blockchain Improvement Proposal 148 (BIP-148), and subsequently the SegWit “User Activated Soft Fork” was performed. Then, despite the fact that SegWit2X Hard Fork was signaled by over 90% of all miners, it was still not activated. Chants of “USA, USA”… I mean “UASF, UASF” were heard in the streets.

This is a nice story. One about the triumph of the common user over the tyrannical miners, and the fending off the “New York Agreement” attack. But it is one that completely ignores the economics of signaling. The reason for SegWit being activated has less to do with saber-rattling, and more to do with a smear campaign against big blockers, and the failure of 2X more to do with a reneged upon compromise and the weakness of “bit 4” to signal commitment. These mechanisms carry none of the clarity proof of work brought to bear on the byzantine generals.

To be honest, I wouldn’t want my fellow investors to be the ones making decisions about the long term health of the blockchain. Their signals are too weak. Bitcoin is an extremely liquid asset. Do you know which type of investors are drawn to liquid assets? It’s not the long-term, vested interest, stuck-with-you-until-the-end-because-they-have-no-other-choice type of investors. What happens when you take their feedback and things go poorly? “So…. my idea is failing, eh? Oh well, guess I better offload my Bitcoin and go find another liquid investment. Hmm, maybe I’ll buy some Ethereum”. You’ll excuse me if I don’t think that memes and the HODL religion will suffice to keep that investor in it for the long haul. Developers suffer from a similar issue, though instead of liquidity, it’s a misaligned interest with the long run health of Bitcoin and their own career advancement.

So then, who are the people with the greatest interest in Bitcoin being successful in the long run? Perhaps somebody with illiquid assets which would greatly diminish in value if Bitcoin’s value fell. Like, I don’t know, merchants and payment processors which have developed systems specific to Bitcoin. The smaller the level of effort for them to keep up with changes to the blockchain, the smaller the incentive for them to find substitutes.

And you know who else has illiquid assets tied up in the success of a particular blockchain? Mining hardware manufacturers. The technology and processes they’ve developed is a steep investment into the future of Bitcoin, and that vested interest would be made even stronger, if say, they developed an optimization which was inapplicable to any other blockchain, like… Hmm… well, I can’t really think of any examples of this, but just imagine if they made an ASIC which could Boost their hash rate specifically for Bitcoin.

Unfortunately, people were deafened to these reputational signals by noise in the form of an analog Sybil attack which preyed on the biases innate to most humans. Anti-market and anti-foreign bias. This noise overwhelmed the signal from miners and merchants about the best way to move forward with improving the network, which made it possible to move in a different direction. User Activated? No. Noise Activated Soft Fork.

ASICBoost was a godsend to the SegWit advocates. “A protocol cheat! A patent awarded! How dare Bitmain use the power of government, arrogating to themselves monopoly power over Bitcoin mining!” ASICBoost provided the excuse for a smear campaign against the single largest party blocking their proposals. It wasn’t necessarily a campaign that needed to be carried out by the interested parties, but rather a seed planted in the fertile soil of the existing biases in Bitcoin’s user base. This became something of a PR problem for the China-based Bitmain, and put them in a poor position when it came time to negotiate a consensus for a fork. They agreed to activate SegWit, and in exchange, the block size would be doubled.

Which brings us to the SegWit2X fork…

The New York agreement was a disastrous tactical error for the Big Blockers. Agreeing to activate SegWit without any form of stake being put up by the opposition to increase the block size was bad, but even worse was the weakness of the ‘bit 4’ signaling chosen to activate the Hard Fork.

The problem with ‘bit 4’ signaling is the multiple ways to interpret it. At it’s very best, it signals not a commitment to contribute hash power to a fork, but rather, the desire for the fork to occur. At worst, it’s a head fake, intended to get other miners to waste hash power on a doomed fork. This creates a collective action problem with everybody waiting for everyone else to “jump first”. This scheme makes for a weak signal.

Of course, the weakness of this signal doesn’t doom the fork to failure, but that weakness makes it possible to overwhelm with noise. Inaction can be the result of not being able to read signals from noise, because if you can’t read the signals, there’s no reason to expect others can. Thus there’s no strong signal to expect the other miners to switch. But imagine for a moment if miners, as part of their signaling, locked up their mining reward which would only be recoverable in a hard-fork? Do you think it would even be possible to cancel the fork with an announcement or because of an off-by-one-error in the code?

In my next article, I’ll go over exactly how a scheme like this could work, and why it would make the next “agreement” a success.

Software Engineer @Go figure it out if you want to