I will preface this post by saying to Pierre Rochard, Nic Carter, and @Bitstein that they should withhold judgment on it until reading through to the very end.
I was perusing Twitter on my way home from Washington, D.C. this evening when I noticed that someone was wrong on the Internet.
Jake is a litigator with the very well-respected disputes firm Kobre & Kim, which (in my several interactions with Kobre & Kim lawyers over the years) also knows the cryptocurrency business extremely well. So let me preface this blog post by saying that I throw absolutely no shade at Jake, whose professional opinion I greatly respect, and that on this question reasonable people can disagree.
Jake’s view is informed by a pervasive meme in the cryptocurrency world – chiefly that Bitcoin, despite being expensive in the extreme to operate and secure, with such expense taking the form of expenditure on electricity to “mine” Bitcoins through proof-of-work, is in the final analysis more efficient than “fiat-based” systems.
In the view of some, fiat money requires a massive military and bureaucratic apparatus to secure and administer:
A recent Medium post by Dan Held attempted to quantify “[proof of work]’s costs relative to existing governance systems” by way of a “rough comparison to the existing financial, military, and political systems.”
Or, as put by Jake,
I think this reasoning is fallacious.
This is not really an ideological point but rather a point concerning the analytical methods we should use when asking questions about the efficiency, security and censorship tradeoffs of distributed cryptosystems and what the appropriate meatspace or centralized comparators are for the purposes of that analysis.
Bitcoin’s best comparator, in my view, is existing bank infrastructure. And not all existing bank infrastructure, but rather existing bank infrastructure which mediates either (a) equivalent notional value to Bitcoin or (b) equivalent transactional throughput to Bitcoin.
In terms of notional value, a good jumping-off point would be the Federal Reserve System. In 2016, the Fed possessed a balance sheet of $5 trillion and managed it on a budget of $4 billion. Most of those expenses were not directly related to its balance sheet but rather were labor costs associated with the Fed’s bank supervision function. The costs of electricity, as far as I can tell, were negligible.
Bitcoin, by contrast is worth $100 billion, depending on the time of day, yet costs $3.5/4 billion of electricity alone to mine/secure in a given year.
Comparing these two systems, as they currently stand, is like comparing apples and oranges. Bitcoin is, on many measures, a loser in efficiency and performance terms. Bitcoin cannot, as yet, be the world’s money; it processes no more than 7 transactions per second. It maybe does as much transactional throughput as one might attribute to a single Bank of America branch in a large city.
Dan’s response to this point was to point to a chart in his blog post:
On this basis, we are told, Bitcoin is more efficient than gold mining/recycling, paper currency, banking as a whole and governments as a whole because Bitcoin costs less than these other systems.
I find such an argument disingenuous, if for no other reason than the fact that Bitcoin does not perform the functions of gold, paper currency, the banking system, or governments. While it sort of operates on a similar theme (setting down rules that govern human conduct combined with a store of value), suggesting that it is comparable as a consequence of occupying a similar conceptual space would be like if I, after deciding that I wanted to compete with Elon Musk in the orbital lift business, began experimenting with moving a watermelon with a Radio Flyer wagon in my garage and, after carefully towing the wagon back and forth across the concrete floor a few times, plugged some figures into Excel and exclaimed, “UREKA! This uses less energy than a SpaceX Falcon 9!”
That statement would, of course, be a true statement. It would not, however, be an especially helpful one if the end-goal of a nascent transportation entrepreneur was to try to devise a more efficient low-orbit heavy lift system. The new data point would be of no use because where we compare a wagon to a rocket, although these are each means of transportation, they are nonetheless very much NOT ALIKE. When the not-alikeness of two things is of a substantial enough degree, comparison does not provide clarity but rather it obfuscates. Where we try to avoid comparing apples and oranges, we must also be careful to avoid comparing apples and Tyrannosaurs.
No offense to Dan, but the chart in his blog post is just plain silly. The chart compares apples with Tyrannosaurs.
Gold, for example, is not used uniformly as a medium of exchange; indeed, not once in my life have I used gold to buy or sell anything, and the only gold I own is a crucifix that I occasionally wear around my neck. Being metal (and an excellent conductor), it is used for a range of processes in manufacturing ranging from jewellery (British spelling) to electronics.
Governments, similarly, are not primarily concerned with providing me with an online P2P payment platform. To the contrary, every government I have ever had the pleasure of dealing with has dealt with enforcing complex bodies of law and policy, including harmonization of transportation and communication standards, social welfare, public and consumer safety, spaceflight, and providing physical security from the nation’s enemies – no small task, and in any case one which does not consume electricity as much as it consumes labor.
When it comes to banks, Bitcoin does not replace those functions, either. Banks provide savings services on commercially very beneficial terms. Sure, I don’t get a lot of interest on a bank account, but my deposit is guaranteed up to a certain amount by the state, I am not liable for fraud, and there are other services – including insurance and loans – which banks offer which Bitcoin does not provide.
The closest comparison is “paper currency and minting.” Here, the Bitcoiners may have a point, but to the extent that they do it requires substantial qualification. Bitcoin in terms of its function may be best compared to the Fed – in a wacky, automatic combination of the Fed’s seignorage function and Fedwire – or the UK’s Faster Payments.
The Federal Reserve System’s opex was, in very broad strokes, dealt with above. As I mentioned, it is likely that Fedwire is a fraction of the operating costs of the Federal Reserve System as a whole. Even if we’re giving BTC the benefit of the doubt, the Federal Reserve System is still, dollar per dollar, about 1,000x as efficient as BTC, costing roughly $1 of expenditure for every $1,000 of notional ($4 billion in operating costs vs $4.5 trillion in assets), rather than $1 of mining for every $25 of notional ($4 billion / $100 billion market cap).
The UK’s Faster Payments system, operated by Vocalink, illustrates the other end of that equation; except, rather than assets, Vocalink more or less blows away any other payment services provider on the planet in terms of performance. It would not be much of an exaggeration to say that nearly every payment in the UK runs through Vocalink. Its revenue per year is £195 million. Even assuming £0 profit, that system runs an entire country’s retail banking transactions – and it does so today at 1/20th the annual cost of mining the Bitcoin network, which (maxing out at 7 TPS) might be able to compete with a single branch of a high street bank in London’s West End.
What to do when presented with Bitcoin, then? Well, I think we can give Bitcoin some credit for being sui generis (that’s lawyer-speak for “unique, with emphasis”). This recognition includes acknowledging Bitcoin’s potential as an alternative to centrally issued, or “fiat,” monies such as those used by the United States.
What we shouldn’t do is get sloppy when we think about Bitcoin. That includes not comparing the cost of operating Bitcoin to bloated strawmen such as, e.g., the total cost of world government, which, while more expensive than Bitcoin, also provide far more by way of services than Bitcoin does (of which the issuance of money is merely one). On a similar vein, we shouldn’t overstate Bitcoin’s abilities. We should ensure that when we compare it we compare it to systems with similar throughput, such as a prepaid web app like PayPal, rather than projecting hopes and dreams into the analysis.
Above all, it also means that we really shouldn’t make the amateurish mistake of making claims for Bitcoin that are unfalsifiable. This includes thinking along the lines that because “fiat money has no worth beyond the military power required to ‘back’ it” this means that the cost of fiat money is equal to the cost of the military apparatus. One could just as easily argue that it is the military apparatus which makes the fiat worth buying, and the continuing value of the currency is a dividend of power, rather than a reflection of the expenditures incurred to obtain power. This is not something which, as yet, we can empirically test.
Finally, it’s just generally a good idea to keep expectations low, that way people will never be disappointed in what they receive.
With that in mind, I say gladly: today, Bitcoin is as efficient as a lame hippopotamus with a hangover. And that’s totally cool. Projects like Lightning promise to increase its performance. Brilliant people the world over are working on scalability proposals to help make Bitcoin more useful to everyday people. I know many of the people working on those proposals and respect them and their work. I look forward to seeing what they build.
With time and effort, Bitcoin might one day become as efficient as a creature far sleeker than aforementioned hung-over hippo, such as, e.g., a Class 4 Battle Marmot equipped with 2,000 megajoule ion thrusters. But today is not that day and continued reliance on proof of work, which consumes more energy as Bitcoin’s price rises, will not help to bring that day about.
Bitcoin with POW is already vastly more expensive than existing systems which, being proper points for comparison with Bitcoin, perform substantially the same function as Bitcoin. If efficiency is the objective, Bitcoin’s basic operating assumptions – including whether to use POW to determine consensus on each block – will need to be revised. If efficiency is not the objective but, say, censorship resistance is, then let us continue as we have before. But let’s not delude ourselves about the costs.
A good point: