Sean King

My photo
Knoxville, Tennessee, United States

Wednesday, December 31, 2014

Thoughts On the Significance of the FEDS Bitcoin Paper, And a Challenge to Bitcoin's Detractors

An October 7, 2014 working paper prepared by staff of the Washington D.C. Federal Reserve as part of the its Finance and Economics Discussion Series (FEDS) reaches some important conclusions regarding the current status of the Bitcoin network.  The full paper is available here.

The primary purpose of this essay is to ponder the potential future significance of what I believe to be the paper's two most important findings:

1) That, as of "the beginning of 2014", there were an estimated 100,000 unique daily users of Bitcoin (p. 17), and

2) That the number of unique daily users doubled every eight months over the three year period ended the beginning of 2014 (also p. 17).

A second purpose of this essay is to challenge Bitcoin's detractors to explain why the trend identified by the FEDS working paper is unlikely to persist.  

Implication of the FEDS Findings

Extrapolating the two numbered statistics above into the future has potentially extraordinary implications. (Set aside, for a moment, the question of whether it is reasonable to extrapolate from these statistics.  I will return to this question in a moment).

First, if the growth trend identified in the FEDS paper has continued unabated since early 2014, then the current number of unique daily users of Bitcoin stands at over 200,000 (since more than another 8 months has passed since then). Furthermore, we are only 14 more doublings away (400k, 800k, 1.6M, 3.2M, 6.4M, 12.8M, 25.6M, 51.2M, 102.4M, 204.8M, 409.6M, 819.2M, 1.6B, 3.2B) from nearly half the world's population using Bitcoin on a daily basis.  For fun, I will call this date "Worldwide Domination Day", or WDD. Extrapolating from the FEDS' calculated trend rate (doubling every 8 months), WDD is about 9.5 years away and will happen in early 2024.   

Even if the FEDS’s estimate of the number of UDUs is off by a full order of magnitude (e.g., because its sorting algorithm fails to sufficiently account for multiple transactions by the same user in a day)--that is, if there were only 10k unique daily users at the beginning of 2014 rather than 100k--the time necessary for Bitcoin's worldwide domination only extends an additional 1.5 years from the present, or until late 2025, if the growth trend is nonetheless correct. Again, that's if the FEDS calculated trend is accurate and it holds for all future time periods. There are, of course, many reasons why the trend rate could decelerate or accelerate appreciably, so extrapolating from the FEDS’s report is far from certain.

Turkey Fallacy?

Projecting the present trend into the future is a dangerous endeavor.  Nassim Taleb has famously dubbed this all-too-common human habit the “Turkey Fallacy”.  A turkey, you see, may come to believe based on present experience that humans are benevolent and will continue to be so in the future. After all, humans reliably provide the turkey with food and care each day such that it grows healthier and plumper over time.  But, of course, the turkey is wrong, and only discovers the terrible truth on the day of its slaughter.

Are my projections above regarding the date for Bitcoin’s WDD simply another example of the Turkey Fallacy?  Maybe.  Perhaps that’s even likely.  But, as we shall see, the world of technology is different from the world of turkeys in important ways.  

To be precise, there are "rules of thumb" (none of which are infallible) often employed by the tech community to forecast future technological trends with reasonably accuracy.  Applying those rules of thumb to Bitcoin, as I do below, requires us to at least consider the possibility that Bitcoin’s Worldwide Domination Day is coming, and sooner than we think.

Rules of Thumb     

The first such rule of thumb is the "Zero to One" principle, articulated recently by tech legend Peter Thiel. Among other things, it stands for the proposition that getting a venture from zero to one is much more difficult than getting it from one to two or two to four.  Said another way, creating a revolutionary new technology or product, and furthering its adoption, is infinitely more difficult than making marginal improvements on existing ones.  For instance, creating the airplane and getting people to use it was much, much more difficult than simply creating a faster train.  

One implication of the Zero to One principle is this:  A revolutionary tech business that is successfully gaining users at a fairly consistent exponential rate has likely already overcome its most difficult challenge--getting from zero to one. By “revolutionary”, I mean one that satisfies the 10X Principle (as discussed below).

Thus, with tens of thousands of daily users and growing fast, Bitcoin is almost certainly past it's “zero to one” phase.  This means that its most difficult days are likely behind it.  

The second rule of thumb is Ray Kurzweil's "law of accelerating returns", or LAR for short.  To simplify, the LAR states (among other things) that, once something becomes an information technology, once it's digitized, its usefulness or its “returns” (measured in terms of price-performance, availability, rates of adoption, rates of expansion, or whatever) grows exponentially, and often at a very predictable rate.  From basic calculations based upon this premise, Kurzweil was able to predict in 1990 that a computer would, in 1998 (precisely), be capable of defeating even the world’s top chess players, something most every other technological expert in 1990 thought was many decades away (at a minimum).  Just a few month earlier than predicted, in 1997 IBM’s Deep Blue famously defeated Russian Chess Grandmaster Gary Kasparov.  For numerous additional examples of the LAR in action, see the many charts at the link in the beginning of this paragraph.   

So, how does the law of accelerating returns apply to Bitcoin?  Well, although it may be hard for newbies to grasp, Bitcoin represents nothing short of the digitization of trust. Bitcoin technology has turned human trust into an algorithm, an information science, allowing it for the first time to be immutably programmed into software. As a result, the Bitcoin technology eliminates, or at least significantly reduces, the need for certain trusted third-party intermediaries--brokers, exchanges, registrars, escrow agents, clearing houses, banks, etc.-- by permitting these intermediaries’s business models to be hard-coded into inexpensive software that no central party controls or practically can control, not even governments.

It's no surprise then that we see clear evidence of the LAR's applicability to Bitcoin.  Not only is Bitcoin's exponential growth evident in statistics offered in the FEDS paper, but in numerous other places as well.  Consider, for example, the number of daily transactions on the block chain (excluding common addresses).  Or, the number of unique bitcoin addresses used.  Or the average number of transactions per block on the block chain.  Or the hash rate (security) of the decentralized bitcoin network.  Or, finally, the number of users of blockchain.info's wallet service. Note that each of the above-linked charts are logarithmic and begin with Bitcoin's inception in 2009.  Since Bitcoin represents the initial digitization of something very, very important--that is, trust--it would be shocking, and seemingly a violation of the LAR--if we were not witnessing exponential growth patterns such as these. 

There is one other very important implication of the LAR:  Once a revolutionary technology starts down the exponential path, exponential growth generally continues indefinitely, only slowing after achieving saturation, upon hitting a hard limit of physics (as Moore’s Law seems poised to do), upon experiencing a catastrophic technical failure, or upon encountering a competing technology that is much, much better (such that the public abandons the former in favor of the better alternative). Baring a catastrophic failure of the Bitcoin technology itself, its difficult to see how Bitcoin's adoption rate slows except for one of these three other reasons.  

Furthermore, observers have noted that certain types of information technologies grow doubly exponentially, meaning that the time it takes for their “returns” (usefulness, adoption, etc.) to double decreases at an ever-increasing rate.  For example, if Bitcoin is such a technology, then we can expect its user base (a proxy for its usefulness) to double next year in less than 8 months, and a little faster the year after that, and so on.  

“Network effects”, which is our third rule of thumb, often explains the doubly exponential growth pattern of many such technologies. The "network effect" explains that certain goods or services, or especially certain information technologies, are only useful provided that multiple people use them, and that they become exponentially more useful as each new adopter is added.  For instance, a hammer is useful to me whether or not other people know of and use the technology, so hammers are not subject to network effects.  This is not so with, say, telephones.  A telephone is useless to me if I am the only one who owns one.  It's an exponentially more useful if at least one other person owns one and uses it, and its exponentially more useful still if a thousand people own and use them.  Further, the more useful the telephone network becomes, the more new adopters are incentivized to begin using it, increasingly its value exponentially more each time, thereby incentivizing even more adoption at an even faster rate. This phenomenon explains the doubly exponentially growth pattern so common with these technologies.  

Network effects are particularly profound with languages and currencies, for obvious reasons.  A language is useless if I’m the only one who speaks it, and a currency is similarly useless if nobody else accepts it.  Recognizing that Bitcoin is both a language (a communications protocol that for the first time permits the programming of nearly full-proof trust into a decentralized network) and, arguably, a currency (medium of exchange), we would expect Bitcoin to exhibit network effect behaviors.  So far, it seemingly has.  

Importantly, network effect technologies, especially open source ones like the Internet and Bitcoin, tend to exhibit the same indomitable characteristics as other information technologies subject to the law of accelerating returns.  Actually, even more so.  That is, after achieving a certain critical mass, network effect technologies tend to continue with their doubly exponential growth in adoption until they either hit a hard limit of physics, experience a catastrophic technical failure, run up against something much, much better, or saturate (dominate) their market.  

Given that there seems to be no hard limits of physics hindering Bitcoin’s growth and adoption anytime soon, and that there is little reason to date to expect catastrophic failure of the technology, the rules of thumb discussed so far suggest that Bitcoin will either come to dominate its market (which, given its open source nature and nearly infinite number of potential applications, is really, really broad), or it will ultimately perish at the hands of a far superior competitor.  

With that in mind, how much “better” must such a competitor be in order to stop Bitcoin's growth trajectory?  This brings us to our fourth and final rule of thumb, which is the "10X" Principle.  The 10X Principle is the idea, commonly accepted in technology circles, that any tech with radical potential must be at least ten times better than what came before if it is  to gain traction and overcome the public's inertial preference for the status quo.  Incremental improvement, or even being five to eight times better, isn't usually sufficient enough catalyst for the public to change its habits en masse, at least not in any reasonable amount of time.    

Bitcoin's unique attributes make it far more than 10 times better than anything that has come before, at least for certain use cases.  In fact, Bitcoin technology makes possible a great many things that were previously not possible at all, such as:

   sending micropayments and tips from person to person over the Internet with no intermediary

   engaging in permissionless, cross border peer-to-peer financial transactions at low to no cost (including but not limited to remittances)

   storing and transporting wealth at little to no cost and without fear of liens, seizure, or theft

   provably fair online gaming

   provably sound financial institutions (with proof of reserves visible on the block chain)

   "trustless" financial transactions of an infinite variety

    widely accepted supranational medium of exchange

And those are just a few of Bitcoin's use cases that have no prior parallels in finance.  There are many others, almost too many to count, that I've not listed.  

And there are yet other things presently doable with our financial system's current technology that Bitcoin will do ten times better/cheaper/faster in the future (such as clearing checks or securities transactions in 10 minutes rather than three days, and at virtually no cost, to name just one).  And lastly, Bitcoin has a nearly infinite number of non-financial use cases that were impossible with prior technologies (like incontrovertible proof of existence of digital documents, provably fair voting, decentralized identity, an uncensorable and incorruptible historical database, smart contracts, smart property, etc.).

But my opinion that Bitcoin is 10 times better isn't what matters.  Its the numbers that matter.  That Bitcoin has over the last five years experienced nearly uninterrupted exponential growth in measures of use and adoption is the most powerful evidence that Bitcoin meets the 10X threshold, at least for certain important purposes.  

But, the 10X principle has one other important implication for Bitcoin:  Any would-be usurper of Bitcoin's place as King of the Programmable Trust must be a full order of magnitude better than Bitcoin to interrupt its trajectory.  Mere incremental improvement is likely not enough.  This is, perhaps, why none of the hundreds of other altcoins launched since Bitcoin have experienced anything approaching Bitcoin's success, at least not over any appreciable length of time, despite that some of them represented clear, though only incremental, improvements upon original.  

Some argue that certain contenders for the cryptocurrency throne that are still in development will ultimately represent a 10X improvement over Bitcoin, at a least for certain use cases.  That's possible.  But, given the hugely ambitious (and therefore significantly more fragile) nature of these coming alternatives, significant doubt remains as to whether they can live up to the hype. 

And, even if they do, it will be at least a few more years (lets say at least three) before any of them are truly ready for prime time.  By that time, and if current trend holds, Bitcoin’s user base will have had at about four more doublings, putting its unique daily users at around 3.2 million.  Since, per Metcalfe’s Law, the value and usefulness of a network is equal to the square of its number of users, the Bitcoin network will be extraordinarily valuable (do that math) and useful by that time, and growing exceedingly quickly.  For an alt coin contender coming out of nowhere and prove itself ten times that useful, a minor miracle may be required. 

Consequently, Bitcoin may remain King of Programmable Trust for quite some time, perhaps decades, and perhaps indefinitely for certain important use cases. With each passing day, WDD looks, at the very least, to be a possibility.   

[Forgive me while I take a moment to note an interesting coincidence (and that’s all it is):  Per the Fed study, Bitcoin had an estimated 100,000 unique daily users as of the beginning of 2014. Squaring this number in accordance with Metalfe’s Law tells us that the “value” of the Bitcoin network at that time was therefore at least $10 billion (100,000^2).  The market cap of all bitcoins in circulation at that time was indeed about $10 billion.]

Conclusion

Many Bitcoin critics have offered up an infinite variety of rationales for why Bitcoin’s growth trend is a fluke that cannot persist.  And, to date, they’ve been completely wrong.  Not only have rumors of Bitcoin’s death been greatly exaggerated (to paraphrase Twain) over the last many years, but it's never even had a close call, at least not in terms of the growth in its adoption measures.  Nothing thus far has managed to even appreciably slow these measures--not the Mt. Gox fiasco, not the Silk Road publicity, not bitcoin's incredibly volatile price (declining nearly two-thirds over the last twelve months), not 2013’s “hard fork” of the block chain, not Charlie Shrem's arrest and prosecution, not the FinCEN guidance, not the IRS guidance, not Warren Buffett's or Paul Krugman's or Jamie Dimon's criticisms of the technology, not Russia's ban, not China's skepticism, not the threat of governmental regulation, and not the rise of an infinite variety of copycats (“altcoins”), just to name a few.

We must be ever cautious of falling for the Turkey Fallacy.  However, at the same time, we must recognize that there are innumerable examples within the tech sector where the Turkey Fallacy seemingly doesn’t apply (for instance, Moore’s law, TCP/IP, Linux adoption, the spread of Internet nodes, mobile phone adoption, and Facebook use, just to name a few).

So, is Bitcoin more like a turkey, soon to see its growth interrupted by slaughter or abandonment?  Or is Bitcoin more like the many examples of information technologies noted above that benefit from the Zero to One Principle, the law of accelerating returns, network effects, and the 10X Principle?  If the latter, then we must consider the possibility that Bitcoin may continue growing at predictable exponential rates indefinitely.  


Regardless, as the above demonstrates, the time has past when those who understand Bitcoin should feel compelled to respond to every ignorant criticism of it. Rather, the burden now lies on Bitcoin's detractors to explain why its consistent five-year exponential growth trend will suddenly end or even appreciably slow.  I ask detractors this:  If none of the great tragedies and threats in Bitcoin’s five-year life have thus far constituted so much as a speed bump on its path to WDD, then what, exactly, do you believe eventually will?


By:  Sean King, JD, CPA, MAcc
General Counsel, Tennessee Bitcoin Alliance