Chapter 3 of the book The Blocksize War is published below. The full book is available on Amazon. As a reminder, 50% of any profits from physical book sales will be donated to Médecins Sans Frontières, a charity that provides medical assistance to people affected by conflict, epidemics, disasters, or exclusion from healthcare.
The Blocksize War – Chapter 3 – Scaling I – Montreal
On the weekend of September 12 and 13, 2015, there was a conference called “Scaling Bitcoin 2015 Phase 1” in Montreal, Canada. This was billed as an attempt to help resolve the conflict plaguing the community at the time. It was, at the very least, an opportunity for the major characters on either side of the debate to talk to each other. Most of the discussion thus far appeared to have been conducted via online forums; it was thought that, with face-to-face discussions, people might have a stronger chance of appreciating each other’s point of view. Who could argue against that?
Crucially, those whom many regarded as the major protagonists on each side of the debate would be there, Gavin Andresen (on the large blocker side) and Gregory Maxwell (on the small blocker side). Gregory, a Bitcoin developer, was a staunch and uncompromising supporter of the smaller block side of the argument, perhaps less willing to compromise than anyone else. Gregory was exceptionally intelligent and appeared to me to have an extremely strong understanding of most of the various fields of study Bitcoin touched, from the computer science and cryptography, to the game theory and incentives. He was sometimes referred to as the Bitcoin Wizard. His very first public post on BitcoinTalk, in May 2011, was about Bitcoin transaction fees and mining incentives, describing how fees were necessary to secure the network.[1]
In 2014, Gregory co-founded Blockstream, a company which appeared to consist entirely of small blockers with a business model which depended on Bitcoin fees becoming high, as Blockstream offered potential solutions to this problem. A problem which, according to the large blockers, did not, or more precisely, should not, exist. Blockstream was therefore hated by the large blockers, who considered it to have a financial conflict of interest: an incentive to keep blocks small. In Blockstream’s defense, there is considerable evidence that most of the co-founders and employees of the company supported the small blocker arguments long before the company was even created or conceived. Some of the large blockers appeared to simply have cause and effect the wrong way around. To me, the Blockstream staff appeared to have joined the company because of the pre-existing view they had about scaling, rather than having come to that view after joining Blockstream. On the other hand, confirmation bias and groupthink are real problems, and it may have been a little bit of both. However, contrary to the claims of the conspiracy theorists, nobody at Blockstream was being deliberately malicious; any failure in cognitive reasoning was likely subconscious. The same can be said of Gavin and the large blockers.
Gregory himself was very active in debating this issue on Reddit, and he became one of the most derided figures in the space. Gregory considered the Bitcoin development process as highly complex and scientific, with many challenging technical trade-offs. He did not welcome the involvement of what he considered the uninformed masses in the decision-making process, likening them to “beer cup hat” spectators at a racing event making decisions about how to engineer the racing cars.
Pulling in a car analogy, you have a pit crew that just added hardened pistons, closed loop anti-knock sensing fuel-air mixture control, nitrous, and recently invented and is planning on building the turbo-charger, all while also contributing to maintaining track and painting the car (which happen to be some of their most visible activities; because they’re easy to explain). and while they’re busily debating compression ratios and high octane fuel and the seeming impossibility of getting the car to safely go much faster with the current state of technology you have a guy standing on the sidelines with a beer cup hat, saying “No problem guys: lets remove the breaks!” and the crowd goes wild: Finally someone who cares about speed.[2]
Gregory earned the nickname “One Meg Greg” due to his support for smaller blocks, and was perhaps more detested by the large blockers than anybody else.
As someone fascinated by the blocksize debate, given the players attending, I felt that I simply had to attend the conference in Canada. I thought that perhaps Gavin and Gregory would discuss things in the open and make progress towards resolving the conflict, and this is something I was keen to witness. At the time, I worked in London for an investment management company called Ruffer. I had limited holiday left and couldn’t really take more time off work. However, the conference was over the weekend. I devised a plan to quickly fly in and out, attend the conference and barely miss any work. I took the 6.45pm flight out of London on a Friday, landing in Montreal at 8.50pm local time. Then, I returned home on a night flight, departing Montreal at 10.35pm on the Sunday, arriving in London at 10.10am on the Monday, where I could head straight from the airport to work. I would miss just a couple of hours work on a Friday evening and then arrive a bit late on the Monday: perfect.
I considered explaining myself to my colleagues at the office and asking for more time off. After all, it was an investment company and this trip could be classified as investment research. I therefore would not need such an intense schedule. Ruffer was a macro investment house, known for accurately predicting and profiting from the 2008 global financial crisis. The company appeared to be perpetually worried about the risk of high inflation, driven by monetary easing from central bankers and the prospect of large fiscal deficits. The investment team did seem highly intelligent, curious and open to new ideas. Many people in the company were also highly knowledgeable of monetary history. Indeed, during my first week at the company, one of the senior team had pulled me aside into a room and asked how I would define money. Bitcoin therefore seemed like a potential fit for Ruffer, at least one day. However, a couple of years earlier, in 2012 and 2013, I had talked extensively about Bitcoin to some of my colleagues. Although everyone was always polite and respectful, I thought the overall impression was somewhat negative. It was probably best not to mention this trip, so I chose the path of an intense two-day trip to Canada, rather than the more relaxed schedule of an official business trip.
Little did I know then, but almost exactly five years after this conference, when the blocksize war was nothing but ancient history fading from memory, Ruffer would purchase around US$700 million worth of Bitcoin for its clients (around 2.5 percent of client portfolios). This would prove to be a monumental moment for the ecosystem, especially in the UK. A highly respected and conservative organisation like Ruffer purchasing Bitcoin had a significant impact on how the financial establishment perceived the coin.
I arrived at my hotel in Canada, which was a short walk from the conference venue, at around 2am on Saturday, local time. I was extremely tired, but still managed to listen to an hour-long podcast about the blocksize conflict, a debate between Gavin and Adam Back.[3] Adam was the only person referenced in the main text of the Bitcoin whitepaper for his Hashcash concept in 1997, and he eventually became one of the leading characters on the smaller block side of the debate. Adam was the president of Blockstream. At this time, Adam appeared to take a more moderate stance than Gavin, offering his support for the idea of increasing the blocksize limit to 2 MB, then 4 MB and then 8 MB, with two-year time gaps between the changes (BIP 248). This didn’t seem too far away from what Gavin wanted, except Adam opposed the continued increases to 8,000 MB that Gavin supported. On the podcast, I remember Gavin commenting that Satoshi had said nodes could be operated in data centres processing many transactions. Adam argued back that mining was now more centralised than under Satoshi’s time. This mining centralisation meant that the balance had shifted: users running nodes to validate the rules and keep the network decentralised was now more important than it was before. It is important to realise that, when Satoshi was active in the space, there was no real distinction between validating nodes and mining nodes. They were essentially the same thing. By 2015, things had changed; there were now specialised mining farms.
I arrived at the conference venue at 8am on Saturday and there were a few hundred guests in attendance. The atmosphere was calm and quiet. Most people didn’t seem to know each other and, like me, considered themselves curious observers in this dispute, rather than participants. There did genuinely appear to be a good mix of people on either side of the debate, and I felt the event was helpful and productive. Most of the talks focused on the computer science of scaling Bitcoin, with a particular emphasis on the scientific method and any analysis which contained data and statistics around the technical limits of the network. The main organiser of the event appeared to be Pindar Wong, a former ICANN Board member, with an expertise in internet governance, who had also been behind one of the world’s first ISPs. Much of the focus at the conference was on learning lessons from governance disputes at internet bodies such as the Internet Engineering Task Force (IETF) and potentially applying these lessons to Bitcoin.
There were two talks that stood out the most to me: one by Peter Rizun on blocksize economics, and another from Jeff Garzik on various blocksize limit proposals. Peter’s talk centred on the economic theory behind the blocksize limit. He believed that the fee market death spiral argument did not apply as, without a blocksize limit, one could have a functioning transaction fee market. However, he did state that non-zero inflation was an assumption in his theory, which is fine if one has a short to medium-term outlook. For those focused on the long-term sustainability of the system, this assumption may not have seemed appropriate. Peter regarded the blocksize limit as a production quota, which is an impediment to a free market, which he claimed would determine transaction fees in a more efficient manner. Peter ended his speech by referencing censorship on the Bitcoin Reddit of those who advocated removing the production quota and DDoS attacks on clients and mining pools that supported removing the quota, such as Bitcoin XT and Slushpool respectively. Peter also mentioned nodes supporting the production quota slipping away. He showed pie charts on screen illustrating that two percent of nodes on the Bitcoin network were running Bitcoin XT on August 15, 2015, a figure that increased to 15 percent by August 30, 2015. Peter then predicted this production quota would fail.
Bitcoin will break down dams erected by special interest groups attempting to block the stream of transactions. That’s all I have to say about the transaction fee market.[4]
The above phrase about a “stream” was clearly a reference to Blockstream, and this comment attracted considerable laughter in the audience. There were mutters by some of the more sensitive small blockers that this was trolling, and violating the collaborative spirit of the conference.
The other speech worth mentioning here is Jeff Garzik’s talk entitled “Issues Impacting Block Size Proposals”. Jeff was another early Bitcoin developer, who had proposed removing the blocksize limit a few weeks after it was added in 2010, as was mentioned in Chapter 2. Despite this, Jeff had always come across as somewhat of a moderate on the blocksize issue, often articulating both sides of the argument. He appeared to be trying to position himself as someone capable of bridging the gap between both sides, and he never appeared to support Bitcoin XT. However, he did seem keen on making some sort of decision, sooner rather than later, and did not have the patience of the small blockers. Jeff emphasised in his talk that the 1 MB limit was very much a marketing problem that would deter companies from beginning their Bitcoin programs.
Another problem, is what I call the Fidelity problem. Fidelity is one of many Wall Street companies looking at doing some Bitcoin experiments and they like many others say that if they switch on the switch on their Beta program, they max out Bitcoin’s capacity. So that makes the project a non starter, from the get go, as projects never get started and that growth you hope to measure never appears.[5]
In the afternoon, the conference broke up into smaller groups. I was allocated into a group of five or six people, along with Gavin. Others spoke about lessons they learnt from disputes related to cryptographic protocols, such as contentious decisions on how to select a hash function. They talked about the need for dialogue and patience. The concept of “rough consensus” was discussed, a methodology used by the IETF, which involves judging the “sense of the group”.
Working groups make decisions through a “rough consensus” process. IETF consensus does not require that all participants agree although this is, of course, preferred. In general, the dominant view of the working group shall prevail. (However, it must be noted that “dominance” is not to be determined on the basis of volume or persistence, but rather a more general sense of agreement.) Consensus can be determined by a show of hands, humming, or any other means on which the WG agrees (by rough consensus, of course). Note that 51% of the working group does not qualify as “rough consensus” and 99% is better than rough. It is up to the Chair to determine if rough consensus has been reached.[6]
It was then Gavin’s turn to speak. He essentially said that all this talk about dialogue and patience was great, but at some point a final decision needs to be made and someone, or some process, needs to be in place to make that final decision. The trouble here, he proclaimed, is that nobody knew who or how this final decision would be made. What he said was reasonable, however I sensed he was becoming more frustrated and losing patience. This is no surprise given the immense pressure he must have been under, as one of the people everyone was focused on. I had the utmost respect for Gavin at the time, for his willingness to participate in this discussion process, when the easier thing to do would have been to not bother attending at all, which is the path Mike Hearn had chosen.
This conference was the first time I met Gregory in person. From reading his posts online, I got the impression that he was exceptionally intelligent, a strong personality, a fast thinker and somewhat impatient and intolerant of those with a weak technical understanding of some concepts in computer science or Bitcoin. I was surprised by his persona in the flesh; he seemed calm, curious, polite, thoughtful and open-minded, a very different Gregory than one would have expected.
In the corridors of the conference, during one of the break sessions, I noticed that Gavin and Gregory sat near each other and started talking. This is what many attendees had hoped to see: the main protagonists on each side discussing the issue. As time progressed, the group witnessing the discussion got larger and larger as more wanted to hear what was being said. The conversation seemed quite tangential to the issues and then became slower. Both parties seemed very uncomfortable, especially Gregory. His preferred discussion format was clearly web forums, where the conversation was open to everyone to view. Something as critical as the Bitcoin protocol should not be discussed in this kind of closed format, at least if any decisions were to be made. Therefore, the conversation ended pretty fast and not much of substance was said.
The format and framing of the conference was certainly tilted towards the small blocker vision for how things should evolve. There was emphasis on science and discussion, rather than making any decisions. The scientific merit format was certainly how many small blockers wanted the space to evolve. Large blockers seemed to favour a more business-like approach; they did not regard Bitcoin as some theoretical science project, but a real world, live system, with real users. In general, large blockers were active users of Bitcoin and they wanted to make their usage easier, without being hampered by theoretical computer scientists they perceived as not even using Bitcoin. Large blockers accused the conference organisers of making the issue too complicated and using the event as a stalling tactic, to buy time. They cynically renamed the Scaling Bitcoin conference series as “Stalling Bitcoin”.
[1] https://bitcointalk.org/index.php?topic=7361.msg108052#msg108052
[2] https://www.reddit.com/r/Bitcoin/comments/3hfgpo/an_initiative_to_bring_advanced_privacy_features/cu7mhw8/?context=9
[3] https://www.bitcoin.kn/2015/09/adam-back-gavin-andresen-block-size-increase/
[4] https://diyhpl.us/wiki/transcripts/scalingbitcoin/peter-r/
[5] https://diyhpl.us/wiki/transcripts/scalingbitcoin/issues-impacting-block-size-proposals/