Scaling Bitcoin

Your mother taught you not to discuss religion or politics in polite conversation.  Bitcoin scaling is far more contentious.  The debate over how to enable the Bitcoin network to handle more transaction volume has divided the community into factions that hurl accusations of treason and conspiracy at one another and even launch DDOS attacks at “enemy” full nodes.  If you’ve been following the debate for the last two years, you’re probably exhausted or bored by it by now, quite understandably.  But it’s a debate that just won’t die because of its important practical implications for an $18 billion asset and the risk of a contentious hard fork.  If you’re a crypto expert, you’ll probably find my next post more interesting.  If you’re moderately familiar with bitcoin, skip to ” Technological and Game Theory Concerns.”  For those unfamiliar, start with “Background” and I’ll provide a little history and explain some of the jargon.

Background

Bitcoin blocks are created about every 10 minutes. Miners all over the world compete to solve a complex math problem, and whoever solves it first gets the right to mine a particular block.  Mining a block enables the miner to collect the “block reward” of newly created bitcoins, as well as to collect any fees associated with the transactions they include in the block.  Miners generally want to include as many transactions as possible to maximize the fees they collect.  Any transactions not included in a particular block can be included by other miners in a later block.

Back in 2010, bitcoin’s creator Satoshi Nakomoto (a pseudonym for an unknown developer) put a cap of 1 MB on bitcoin blocks.  That means only about 3500 transactions can be included in each block, about 6 transactions per second.  At the time Satoshi introduced it, there were hardly any transactions at all, so the limit had no practical effects on the network; it was just intended to prevent a malicious actor from flooding the network with spam.  Over the last year however, bitcoin volume has been growing strongly and we’re now consistently at the 1 MB limit.  When people submit more than 3500 transactions, how do miners decide which transactions to include?  They pick the transactions with the highest fees.  Users can specify the fee they want to pay for their bitcoin transaction.  At times of peak network demand, the fee today exceeds $0.75 per transaction.  While that might sound small compared to a wire-transfer, it’s substantially more than many credit card transaction fees, and it makes paying for coffee with bitcoin uneconomic.  And since we’re at the upper bound of network capacity, every small additional increase in demand for transactions produces a large increase in fees; everyone has to compete to get their transaction included in the blockchain.

Why not raise the 1 MB cap?  One of Bitcoin’s core value propositions is that it is decentralized.  “Decentralized” can refer to a bunch of different things.  Here, I’m referring to the backbone of the network – full node operators and miners.  A full node is just a computer running software that propagates and validates bitcoin transactions – a $25 raspberry pi can do this.  Miners add blocks to the blockchain and secure the blockchain from doublespends (spending the same bitcoin twice.)  As long as there is one full node and one miner, the Bitcoin network is up and running.  By having full nodes and miners located all over the world, it makes the network resilient from both natural and man-made destruction.  If all of the nodes were in the US, the US government could more easily shut down those nodes and kill bitcoin.  So the bitcoin community greatly values having as many full nodes and miners as possible, located in as many distinct political jurisdictions as possible.

Technological and Game Theory Concerns

Currently, there are about 7000 full nodes globally.  Some of these nodes are run by bitcoin businesses that need to be able to validate transactions, but many node operators are just bitcoin hobbyists.  These hobbyists are willing to invest the time and money to keep a node running with no economic incentive.  The biggest cost associated with running a node is bandwidth: 1 MB every 10 minutes adds up.  For node operators in developed economies, this might be a small burden.  But handling that bandwidth in rural India or rural Peru is already prohibitive for some.  Raising the block size would directly increase the cost of running a full node.  I have yet to see a good estimate of how many nodes would fall off the network with an increase to, say, 2 or 4 MB, but it’s non-zero.  Many in the Bitcoin community are reluctant to raise the block size because it would likely result in a loss of full nodes in the most scarce areas: rural emerging and frontier market locales.  The vast majority of full nodes are in Western Europe, Japan, the US and China, so losing even a few of the full nodes in other countries would represent a meaningful loss in decentralization.  Additionally, a miner who solves a block receives a slight head start on the next block; other miners have to wait for the solved block to propagate to them.  Much larger blocks might take longer to propagate to some miners, which could encourage mining consolidation, a further loss of decentralization.

Further complicating the situation are second order effects.  The current cap on bitcoin transactions reduces its usage.  This may reduce the number of bitcoin businesses operating and the number of merchants that accept bitcoin (and are thus incentivized to run full nodes), so it may be that reducing the fee to transact in bitcoin would indirectly increase decentralization by encouraging adoption and thus the creation of full nodes.

Additionally, some economic uses may be priced out of utilizing the bitcoin network, which may encourage the growth of Bitcoin competitors.  For example, a pharmacy may not be happy with $0.75 transaction fees and might choose to utilize another cryptocurrency like Monero that costs less than 1/10 as much to transact.  These marginal use cases may give competing cryptocurrencies a foot in the door that helps them achieve network effects and eventually makes them generally competitive with Bitcoin.

This is mostly a timing issue.  All cryptocurrencies face similar scaling issues, but other cryptocurrencies have much less demand on their networks currently.  Eventually, all successful cryptocurrencies will have to utilize new scaling technologies or scale off-chain, but those solutions aren’t yet ready for widespread adoption.  So this is really a debate over how to handle the 6-18 month period where Bitcoin transaction demand exceeds the network cap, but other solutions are not yet available.

Solutions

The Core team (a group that includes a plurality of Bitcoin developers) is trying to implement a slight increase in block size via a more complex change called “Segregated Witness” or SegWit for short. SegWit can be implemented via “soft fork” which will not produce two competing chains. SegWit would eventually be the equivalent of an increase in block size to 1.7 MB.  The SegWit soft fork also includes a number of technical improvements, including an end to “transaction malleability”, and an improvement in the efficiency of sighash operations.

The simplest immediate solution is to “hard fork”, to change the block size to 2 MB or 4 MB, but this may result in some reduction in decentralization.  Additionally, there is a concern that a hard fork could create confusion for consumers and investors by producing two competing blockchains.  Lastly, even for those who believe that the pernicious effects of bigger blocks on decentralization are unlikely to be felt at 2 MB or 4 MB, scaling via hard forks to bigger blocks is an inherently unsustainable approach, since we hope that demand for transaction volume will increase far faster than bandwidth costs fall.

Attempts to produce such a hard fork over the last 2 years have appeared in various software and marketing implementations and failed to gain traction.  The latest version is “Bitcoin Unlimited” and includes some additional important changes to the network.  Bitcoin Unlimited allows for miners to decide block size dynamically, which is viewed as particularly dangerous by some analysts because of the possible feedback loop in which the largest miners might enlarge the block size, encouraging miner centralization, thus giving themselves more power to push for additional block size increases.

Lastly, there’s a “hybrid” approach: adopt SegWit and hard fork to 2 MB.  Such a hybrid solution would provide Bitcoin with some room to breathe while longer term solutions like the Lightning Network can be rolled out, but it still stokes the fears related to hard forks in general.

The Politics

Some of Bitcoin’s earliest supporters were self-described cyberpunks and anarchists.  Many hoped (at least in Bitcoin’s first couple years) that Bitcoin would eventually replace all fiat currency.  Those who hope for Bitcoin to replace the US dollar recoil at the rising transaction fees that make such a replacement all but impossible.

Most of the Bitcoin community accepts that Bitcoin can’t compete with fiat for transactions until sustainable scaling solutions are in place (off-chain scaling, or other technological innovations).  The rising fees are viewed as an acceptable nuisance.  Some Bitcoin investors think of Bitcoin as though it was IBM – a big established company, with much to lose; they are risk averse and view hard forks as inherently risky.  In contrast, other Bitcoin investors view Bitcoin like Uber – a young upstart with little to lose and lots to gain.  Technologies generally have to “move fast and break things” to avoid obsolescence, and some members of the bitcoin community fear that Bitcoin will be left behind by newer cryptocurrencies if it fails to aggressively upgrade its protocol.  Other investors think that Bitcoin is unlikely to ever be able to compete with younger competitors on features or technology, and must compete instead on the strength of its stability and network effects.

There’s also a rift between people who believe that bitcoin decision making should be viewed amorally as a process of game theory including the inevitable politics, while others believe that decisions should be driven by consensus based on the technical merit of the proposal.  The former group accuse the latter of centralizing control of the Bitcoin network in the hands of a few developers.  The latter accuse the former of pushing their self-interest over the superior technical solutions that would benefit the network as a whole.

At first analysis, everyone in the Bitcoin community wants the same thing – the price of bitcoins to appreciate.  This is because most of miners’ income comes from the “block reward’ (a fixed number of bitcoins awarded to miners for solving blocks, currently 12.5 BTC per block) not transaction fees (currently averaging 1.6 BTC per block).  But in a couple years, transaction fees may be greater than the block reward.  Miners eventually want lots of on-chain transactions and high fees.  But miners understand that if fees get too high, it will incentivize the creation of off-chain scaling solutions.  Additionally, miners are competing against one another, and the largest miners may view Bitcoin Unlimited’s dynamic scaling as a way for them to gain an advantage over smaller miners.  There may also be perverse economic incentives for some Core supporters.  Some of the most prominent Core developers are employees of the private for-profit company Blockstream.  Blockstream profits by providing services related to off-chain transactions, so they may benefit from constraining bitcoin’s on-chain bandwidth.  I have no special insight into the motives of the players – I’m summarizing the political issues as perceived by major groups in the Bitcoin community in this paragraph, and nothing here should be interpreted as impugning the character of miners, Core developers, or anyone else.

The “Crisis”

Some of the biggest miners are threatening to hard fork the bitcoin network by supporting Bitcoin Unlimited.  Many Core supporters have said they will view any such hard fork as an illegitimate competing cryptocurrency, and even as an “attack on Bitcoin.”  Some Core supporters have said they will attack the Bitcoin Unlimited network via “zero day attacks.”  One miner who supports Bitcoin Unlimited threatened to spend $100m to attack the Core branch in the event there’s a fork and things get ugly.  The fear is that the two resulting blockchains might destroy one another, or at least produce general confusion and uncertainty that permanently impairs bitcoins as an investment.  Additionally, most miners are currently signaling that they will not implement SegWit, which means that even a modest on-chain scaling improvement may be far away or even politically unattainable.

The Future

I think it’s unlikely (<35%) that bitcoin suffers a contentious hard fork of any consequence in the next year.  Miners have too much to lose and too little to gain from such a hard fork.  In the event miners do HF the network in support of Bitcoin Unlimited, I think it’s likely to be a relatively minor event that doesn’t impair (economically or operationally) the Core branch for more than a few months.  I think it's about 50% that SegWit is implemented via soft fork in the next 12 months.  In the meantime, we are likely to see fees continue to rise substantially.  If SegWit is not implemented, I view that as a detrimental outcome, but one that does not fundamentally change the Bitcoin value proposition nor the investment opportunity.

ConclusionBitcoins (BTC) as an Investment

 There’s an almost daily announcement of a new exciting cryptocurrency project.  Some are competitors to Bitcoin, but most are not.  Some of these projects offer interoperability between blockchains, decentralized file storage, or trustless casinos.  Many of these projects are exciting and a few are great investment opportunities, but I think over the next 12 months, Bitcoin will remain one of the top investments in the cryptocurrency world.  I have about 37% of my cryptocurrency portfolio in BTC.

There’s a huge amount of capital poised to enter the cryptocurrency markets over the next year, and much of that will flow to Bitcoin due to capital market availability and liquidity – Bitcoin is the only cryptocurrency with exchange traded funds (GBTC in the US, XBT in Stockholm) and a liquid futures market.  While rising fees make Bitcoin unusable for small transactions, I think this is relatively unimportant in the short-term.   In the next 12-24 months, the vast majority of cryptocurrency demand is likely to come from speculative investment, not consumer usage.  And if someone wants to buy or transfer $5,000 worth of BTC for long-term investment, whether the fee is $0.05 or $5 is unimportant.

I’ll provide a lot more detail for how and why I expect capital to flow into Bitcoin in an upcoming post.