Bitcoin as a novel market institution — Nic Carter (talk at Baltic Honeybadger 2018)
Today I am going to be talking about bitcoin as an economic system, not as a software or cryptographic system.
This talk has two parts. In the first, I’m going to take a retrospective of the past 10 years of bitcoin as a functioning economic system. In the second part, I will be looking at bitcoin as it is today and as it is likely to be in the future. In particular, I will be fairly critical of the way that we measure bitcoin, especially of the amount of wealth stored in bitcoin and also of the value of the economic throughput of bitcoin. We have popular measures for measuring these things, but they are inferior in a variety of ways.
Let’s start with the throughput of bitcoin. It’s very common in the press to measure bitcoin in terms of tps (transactions per second) or transactions per day. But this kind of neglects the essential point which is that bitcoin transactions are extremely large from an economic perspective. They are nothing like Visa transactions, which they are commonly compared to. I like to refer to them as containerships, not parcels. The average bitcoin transaction is about $3k and your median bitcoin transaction is about $200. I think this is often overlooked, especially in popular narratives. We should try to re-orient the narrative here a little bit - it’s containerships, not parcels.
We also have a lot of uncertainty about the actual economic throughput of bitcoin. There’s currently no accepted best measure here. Most people tend to refer to the blockchain.info number, but it’s fairly conservative and also a black box and we don’t know precisely how they drive that algorithm. If you look at the raw output, without adjusting for change addresses, you’ll typically get a daily figure of a few billions. I run coinmetrics.io and we have a few estimates which are transparent and not black boxes. We reduce the raw output from multiple billions down to about $2b/day in terms of economic throughput, and then blockchain.info is the more conservative metric. So in the charts that follow I’ll be using the blockchain.info and the coinmetrics numbers for transaction volume. The lesson here is that there are many ways to slice the economic throughput for bitcoin and depending on your inputs you get very different velocity figures.
Compared to Visa, over the last decade Bitcoin has moved up seven orders of magnitude in terms of its daily throughput. Visa has also been increasing in the last decade. I included the exponential extrapolation here. This is a log chart. The lesson from this (and this is using the conservative figures for bitcoin throughput) is that bitcoin is a single order of magnitude away from Visa’s transaction volumes. Even though they are two very distinct systems with lots of dissimilarities, the throughput’s volume is only one order of magnitude away from Visa’s volume.
In this chart, I’ve aggregated a great number of different economic systems including payment systems, settlement systems, OTC gold (which I think is probably the most interesting), and mobile settlement systems and obviously at the top you have SWIFT and Fedwire, which are the ultimate global settlement systems. And that’s what bitcoin is really competing with, at the end of the day.
Here we have two measures of bitcoin throughput that have been gradually increasing and traversing the boundaries, traversing Western Union, traversing the global settlement and so on. I think the most interesting one is the physical gold settlement markets. Saifedean (Ammous) likes to compare bitcoin to gold; bitcoin is essentially physical settlement (although it is not physical) and I think it makes a lot of sense to compare it to the OTC gold settlement markets. Depending on how you slice it, we recently surpassed the OTC gold settlement markets. I would consider that a significant victory, although obviously there are quite a few more targets to hit there.
So that was my first measure of throughput. Let’s focus on the aggregate wealth stored in bitcoin.
This is kind of an issue because people like to use market cap. Today’s bitcoin market cap is about $115 billion. The hidden assumption there is that bitcoin’s supply of 17m currently is all active, but we all know that’s not the case. We have Satoshi’s coins, which are assumed to be lost, and then we have coins that are provably lost, and some of those are counted in the market cap today. This chart says there are a couple millions of coins that haven’t moved since 2009–2010. It’s possible they are in deep cold storage, but it’s also likely they are permanently lost. I’m not sure it makes a huge amount of sense to count these permanently lost coins in market cap. This is my critique of our measures of wealth. The main point of this chart is that the measure of market cap is extremely inferior, especially in the case of coins that are non-circulating.
This is one potential alternative that I like to propose, which I call “accumulated security spend”.
It’s essentially aggregating miner revenue over the entire history of bitcoin with the idea that miners have fiat-denominated costs, so they are essentially forced to sell their bitcoins when they mine them. So since they are commited sellers, there has to be someone on the other end of that trade. This measure aggregates that over time and it gets us a very rough esitmate of the total wealth inflow into bitcoin. Added up over bitcoin’s ten years history, what you get is about $10 billion which is the floor of an estimate for the amount of wealth inflow into bitcoin. Right now, the market cap is $115 billion but based on this extremely rough heuristic, the inflow is somewhere between $10 billion and the marketcap.
This is an alternative to market cap that I would like to propose today, a novel methodology that I developed with Antoine Le Calvez, called “realized cap”.
The idea here is that you’re aggregating UTXOs based on the price at which they last moved. If there’s an UTXO that last moved in 2009–2010, it would be priced at the price back then, which was extremely minimal. What you get is a realized cap of about $88 billion, as opposed to a market cap of $115 billion. This tells you that there’s a significant amount of coins which are lost or should not be priced because they are not active. The criticism of realized cap is that we have long-term holders that have been holding for 7–8 years and they do intend to sell some day and they should be counted. But this is just an interesting, rough heuristic which is my counterpoint to marketcap.
I am going to apply this to some other coins to really drive my point home.
I don’t want to dwell on competitors for too long, but Bitcoin Cash is an interesting case study. The market cap of Bitcoin Cash exceeded $60 billion at one point. But the “realized cap”, if you weighted by the UTXOs that were actually active and moving, was a small fraction of that. The reason for this is that many people never sold or moved their bcash that they received in the airdrop of August 2017. Those coins were never activated at all. The market cap is an inferior measure here, and I would argue shouldn’t be used here.
Another case study was Bitcoin Private which was an exotic merger of Zclassic and the bitcoin UTXO set. And almost no one claimed their Bitcoin Private. The market cap exceeded $2 billion at one point but the realized cap was a tiny, tiny fraction of that, which reflects that fact that almost no one claimed it. Using marketcap as a measure is essentially totally illegitimate.
This concludes the first part which is my critique of market cap, which I suggest you abandon as a measure. You don’t have to use realized cap, but I think we should definitely be more critical about the kinds of metrics we use to measure economic wealth in these systems.
The second part is about the bitcoin network itself.
Today, if you look at where transactions are going on bitcoin, and the throughput, it’s mostly going between large exchanges. It’s settling flows between large exchanges, especially between many fiat on-ramps. Bitcoin is the global clearinghouse for that system. I think many of us would like it to be a functioning payment system, but there’s only a few million dollars worth of payments flowing per day for actual merchant services. But the vast majority of volume is between exchanges. Now a lot of people would say that means bitcoin is just for speculation. I would say bitcoin is a really efficient clearinghouse for value, and it doesn’t matter what it’s used for, all that matters is that it’s being used.
How do I know that bitcoin is mostly transmitted between exchanges?
We can look at batching and see where coins are being sent and how. You can look at batching figures on p2sh.info. It’s typically exchanges that batch multiple inputs together into a single transaction. You can see that 30–40% of all on-chain transaction volume is batched. You can also tag a bunch of exchange addresses and see that about 18% of all bitcoins is held within those exchanges’ hot wallets. So the exchanges typically do dominate.
I also spot-checked the batching figures. If you look at the last week of transactions, there are many kind of exotic “many-inputs/many-outputs” transactions which are the hallmarks of exchanges. So the exchanges do really dominate, for better or worse, and I would argue that we probably need to accept that. A lot of people say “not your keys — not your bitcoin”; it’s definitely true but it’s hard to convince everyone else of that fact.
This is my model of how bitcoin is likely to look like in the future.
I don’t endorse intermediation in bitcoin and I strongly support all measures to give users autonomy and ownership over their keys, especially using non-custodial methods. But the reality of the situation is that bitcoin is highly intermediated today. Most people only touch it through Xapo or through Coinbase or through any of the numerous exchanges. It’s our responsibility to police those intermedaries. We have banks, fiat off-ramps, custodians, lightning custodians, merchant solutions, and obviously there is still direct access and many still do transact with the base layer, but as transaction fees get more expensive, it’s likely that it continues to become more and more intermediated. This is a challenge that we all have to take on as the stewards of bitcoin.
I would like to briefly talk about a couple of measures to determine how well those large intermedaries are taking care of this commons which is the bitcoin block space.
This is the average transaction size in bytes.
It’s creeping up, which is interesting. In my ideal world, we’d get better and better compressing that. But it’s creeping up and some people say this is because of batching and compressing many payments into single transactions. This doesn’t necessarily seem to be the case right now. We haven’t had too much success with batching.
I would like to propose a measure to really determine how efficient bitcoin transactions are, which is the “economic density”.
This is the total economic throughput divided by the bytes that are being throughput on bitcoin. So what this tells you is that a single byte on bitcoin today represents about $15 in transfer activity. That figure might not mean a lot to everyone, but that tells us that we have natural constraints on the system, because we can only transmit so many bytes per units of time. Ideally we want this value be as large as possible, so with limited throughput (so we can keep running full nodes really cheap) we can transmit a huge amount of economic activity. We have to get better and better at this efficiency if we want the system to work long-term.
Returning to that “bitcoin is an intermediated chain” idea.
The criticism of that is: are we not meant to be dis-intermediating economic activity? Obviously we are, but we just have natural constraints. We have the problem that the UX of using bitcoin at the base layer is in many cases worse than using it through an intermediary. Many people much prefer using it through intermediaries, whether that is custodial exchanges of some sort, or payment processors, rather than using BTCPay server or something.
My argument here is that Hal Finney was actually right about this “bitcoin banking” idea, it’s just that today we call banks “exchanges”. There are huge amounts of bitcoin held on exchanges and they’re essentially operating as full reserve custodial banks. Right now the exchanges aren’t doing fractional reserve (I hope), but people do treat them as banks. People treat Coinbase as a bank. In many cases, this is a better UX for users even though I would encourage people to hold their own bitcoin.
The question here is: how do we cope with this reality? I would prefer that we don’t plug fingers in our ears and ignore the reality of the intermediated chain today. My argument is that we should be trying to reason with these intermediaries and ensure that they are behaving appropriately.
Again, I probably agree with Saifedean on this, that we can retain the “hard money” and the scarcity properties that make bitcoin so great, even if there are intermediaries that are serving the chain for us. And we can probably still retain trust minimization because not all intermedaries are the same.
I would say it is on us as bitcoiners to encourage large operating companies that use the bitcoin chain in some manner to behave appropriately. We’ve actually seen a lot of this in December 2017, people were really hounding exchanges to use SegWit, to batch, and to some degree that worked pretty well, and many exchanges adopted it. But many still don’t use SegWit, and it is completely and utterly irresponsible of them to not include the latest efficiencies, if they have the operational capacity to do so. In many cases it just comes down to laziness or not caring too much about the on-chain impact.
I would argue that they are the stewards of the chain, and if we don’t police them, we’ll have a “tragedy of the commons” situation developing. If you look at Ethereum, there has been absolutely no stewardship or policing of that chain. The outcome is that it is just packed full with spam and being abused. And now you have Ethereum devs asking users to stop running dApps and so on. That’s not an ideal situation. It’s better to get ahead of that and just encourage the large industrial users who have large chain impacts to behave appropriately and right now that means SegWit and batching. But more generally it means respecting the p2p network governance process which was demonstrated with the UASF last year, and not supporting hostile forks. If these intermedaries are supporting hostile forks to bitcoin, then they are overtly hostile to bitcoin and we should make them aware of that and treat them accordingly.
I would argue that bitcoin activism with regards to intermedaries is tremendously important. We have been successful so far but there are many operating businesses that rely on bitcoin but are totally deaf to bitcoiners and it’s up to us to actually make our voices heard.
As far as allocators go, it’s important to support businesses that retain bitcoin’s trust-minimization properties, so HodlHodl is a really great example for that, as it is a functional, operating business but also a p2p exchange which doesn’t involve full custody or intermediation.
Then, I would say lastly, it’s very very important to maintain the focus on the verification costs, so that if people want to exit from that intermediated world and run their own nodes and become fully sovereign, they should still have the ability to do that. I’m very pleased to see a lot of people talking about maintaining an own full node as cheaply as possible.
So the takeaways from this talk are:
I critiqued market cap and I urge you all to be very critical, especially of the supply figures which are quoted on websites like Coinmarketcap, which are total nonsense in my opinion, and very uncritical in terms of listening to the data claims of altcoin promoters. We should be much more critical and we should actually look at the chain to determine what the real float of these things is and what the real supply is, because you have multi-billion market caps for many altcoins, and effective (or realized) caps in the low millions. It’s completely unacceptable to just listen to folks who have a huge incentive to get that glamorous spot on Coinmarketcap. What we should be doing instead is inspecting the chain, or at least adjusting the supply accordingly.
With bitcoin, the problem is not as extreme, but you probably should be making an about 15% adjustment for the lost coins, and one way to do that is with the “realized cap” methodology, which I publish an explainer about soon.
I think our preferred measure of bitcoin’s success should not be transaction count, because that’s capped in a variety of ways, but economic throughput and “economic density”. We should be trying to make bitcoin transactions as efficient and as large as possible. We’ve had a lot of success doing that so far and I’m very optimistic about that. I think we definitely should be trying to maximize the throughput of bitcoin, which we’ve had great success with so far, and I’m certain that in the next five years or so, we’ll be surpassing Visa and that will be an extremely exciting day.
Let me return back to this slide here:
I think it’s also important to be realistic about the nature of bitcoin’s economy today. To the extent that we have intermediation, that’s ok, and not all intermediaries are hostile to bitcoin. Many of them are attempting to be a good steward of the chain but unfortunately, some of them are hostile, or at least careless. So at that point it’s on bitcoiners to try and force them to behave appropriately.
But I would argue that we cannot hide from intermediaries. There are many operational businesses that are being built on bitcoin, and many many millions of users are using those instead of running their own full node, or using Electrum server, or custodying their coins themselves. To the degree that there is intermediation, I think that’s ok, we should be aware of that and encourage them to be as responsible as possible.
Nic Carter (Twitter)