Why the problem is not the block size but the blockchain size, the math and the philosophy

Bitcoin needs to support thousands of transactions per second, maybe millions. No amount of tweaking can make the numbers jive. I argue so in this post. At the same time, some of the fundamental design choices are naïve and unnecessary.  I start with this bold statement because, since I am not Mile Hearn, I do not expect anyone to read on without an initial teaser.

From SN seminal paper, bitcoin intended to have some characteristics that made it into a good currency, money, numeraire, or whatever you want to call it. Initially, it seemed that the design attained those characteristics, and indeed its phenomenal success suggests it did.

However, I think the design has not scaled well and is reaching a point where it cannot continue to grow and maintain the originally desired features.

The design assumed that the pure ‘progress’ (like Moore’s Law) would keep up or surpass the operating requirements of the btc network. Nevertheless, this is being proven unrealistic.  Even if Moore’s law continues, the scaling of btc has been and needs to continue to be explosive.

To be relevant, let’s say Bitcoin needs to account for 1% of the world’s transactions.  Roughly, if Visa processes 10k transactions per second, let’s say overall card transactions are ten times that and other transactions are on the same level, so there are today around two hundred thousand transactions per second.  That would be 17 billion transactions per day, i.e. 2.5 transactions per person per day on the planet.  This is back of the envelope, but probably not far from the right order of magnitude.

With those assumptions, Bitcoin needs a bandwidth of 2000 transactions per second (1% of two hundred thousand) just to be on the radar. And that is just 2015 rate, but it doesn’t take too much reading to see that other innovations (IOT, for instance) will add some zeros to the figure.

A naïve vision of Bitcoin as planetary (or even intergalactic) currency, then, is like the Jetsons’ view of everyday life. Cool in theory, but numerically unviable.

Again, an issue of the blockchain and its size. But there are economic and not just technical questions:

When I was a kid, they sold special “airmail” envelopes, made of lighter paper, because only critical mail was worthy of the expensive air delivery. Nowadays, all mail is airmail, even junk.

For the same reason, critical payments used to be the only handheld ‘real time’ transfers, while run of the mill transactions could take several days. Now payments are converging to real time, regardless of size, and cost is plummeting.  Special high value private networks that made sense before are merging into the internet following the same pattern.

Bitcoin’s current bandwidth, however, doesn’t even match the throughput of high value networks.  There are various proposals to bifurcate the rails, like sidechains, Lightning network, altcoins, etc. It seems as we are reliving a millennium of payments kludge history in just a year. In a world where Faster Payments is implemented in a few countries and private firms are marketing this service to many others however, a system that can only handle high value transfers is a weak competitor.

If I could send a little note to Nakamoto-san in 2008, I would tell (her/him/them/it?): “Think this through a little longer. Bitcoin is not ready.  Your fundamental ideas are sound, the implementation still is not. In six years your heirs will be squabbling over details because the whole thing is bursting at the seams.”

Next issue: Judas, Manhattan and my great grandfather

Tip me in Bitcoin.

If you need to know what Bitcoin is go here

if you need bitcoins, go here

temporary Wrodpress tip address


Fedcoin Helpers

Following up on posts by David Andolfatto et al. on the issue of state sponsored digital currency, I came across the following article.  IBM is looking to support central banks in setting up some form o digital currency.  Whether or not it is the right approach, it seems to me that the arrival of industrial strength suppliers is good news.

Ultimately, even if taxpayers’ money is not put to the best possible use, community/open sourced alternatives will not be crowded out, but rather spurred by any large scale implementation.

Reuters article on IBM move here

temporary Wrodpress tip address



Which Fedcoin?

More FedCoin thoughts.
Actually I eerily coincide with Robert on most points, except for monetary policy.
If anything, the more transactions become frictionless, the less any monetary authority will have room for maneuvering. The minute the interest rate is not of my liking (and I like nothing worse than a negative rate) I would jump into bitcoin (or ECBCoin or SwissCoin) and arbitrage out. I have to give it some more thought to the implications.


temporary Wrodpress tip address



David Andolfatto, Vice President of the St Louis Fed, published a most interesting post yesterday Fedcoin: On the Desirability of a Government Cryptocurrency. Andolfatto’s post is itself in reference to JP Koning’s Fedcoin piece of last October. Back then, I wrote a bit about that on a private email list that is usually devoted to topics relating to blockchain protocol design. I thought the Fedcoin thought experiment was interesting fodder for our monetary intuitions. Still do. So here it goes.

A central bank backed blockchain payments network for a national currency like the USD is a neat idea. It would, first of all, put digital cash on the map for good. And digital cash that trades at parity with an economy’s well-established unit of account is a far more useful medium of exchange than a volatile cryptocurrency like bitcoin. Andolfatto:

And so, here is where the idea of…

View original post 1,401 more words

Yermak Again

David Yermak came out with an article in my Alma Mater’s publication Technology Review.  I had already commented on his NBER “paper” on the topic of Bitcoin.  This new article is somewhat more “diet”, but contains more of the same inaccuracies and intellectual difficulties.  Below is the link to the article and a verbatim of my comment on TR’s site (some minor typos corrected.)


My Comment

Many of the weaknesses have already been addressed by others. I would only like to point out to Mr. Yermack’s statement on deflation.

 It is far from clear that a wider adoption of Bitcoin would imply deflation.  If Bitcoin became the only medium of exchange in a given community (or the world) though, then there would be deflationary pressures, although issues like money velocity would need to be accounted for.

 It seems the deflation issue is raised as a scarecrow (your salary will be cut!)  I would point out that the best parts of our economy have been in permanent deflation for the last half century and that doesn’t seem to have had any detrimental effect or prevented countless of our best minds to try and get in on these industries.  Try explaining consumers they need to pay Intel, Microsoft or Apple as much for unit of their product as in 1984.  Try telling these companies they need to pay their employees the same salary per kilobyte, petaflop or coolness index produced as they did in 1994.

 Maybe Bitcoin will help usher an era when adjustments to relative price are accepted by members of society without the need to recur to the stratagem of inflation.

 I looked up the author, Mr. David Yermack, who is a professor of Finance at NYU, but also the author of a powerpoint presentation called “The Michelle Markup, the First Lady’s impact on stock prices of fashion companies”.   I have no doubt Mr. Yermak must have worked hard for his professorship at NYU and knows a lot more economics and statistics than I do (just like P.Krugman, another Bitcoin basher), but I think he does not look his best here.

 It is notable that this is a summary of an NBER paper from late last year. Unfortunately, you need to shell 5 bucks to buy the article and read it.  And it ain’t worth it.  Despite the formal academic structure of the paper: abstract, notes, references, black ink on white paper, I think it contains nothing if not the writer’s feelings as described in the free abstract. It is full of inaccuracies, crass errors in basic concepts of statistics and lack of common sense.If it weren’t coming from an established academic, I would figure it was written by a student.

 To save TR readers the five bucks (or to entice them to read it for themselves), a few points on the NBER paper:

 – Like any good academic paper, there are references: two of them! They are Satoshi Nakamoto’s original paper and an article on Wired.

 – The impressive looking tables show volatility measures and correlation with other currencies. Those tables are a sham. We all know the usd/BTC rate is highly volatile, but any statistician worth two cents will tell you that the measure of volatility is highly impacted by trend, and btc shows a clear trend: Up. 

 – Also, the correlation tables are badly designed. There is a conceptual error and also a Stat 101 error (carelessness or intention?)

 — The operational Stat 101 error is that the eur/eur rate could not correlate 100% with the usd as the author shows in the table. The correlation must be 0, since eur/eur equals 1 and a constant series is 0-correlated with any variable series. The author is making the point that btc correlates poorly, so he could not show a 0 there.

 — The conceptual error is claiming that btc is poorly correlated. It is indeed (depending on what r-squared you consider poor), but so is the yen and gold. On the other side, eur, gbp and chf are correlated among themselves, since they are all rather connected to the same EU economic environment. This does not prove anything, but the author hopes no one will really look at the tables.

– Lastly, the big nonsense is claiming that btc is a poor risk management tool because it is not correlated. The first thing you learn in finance courses is that good risk management seeks diversification, and diversification is best achieved by owning assets whose performance is uncorrelated!  This statement is modified in the current article, because someone must have pointed out the problem.  It now says that the Bitcoin risk cannot be hedged. This is partly true, but not very relevant to a forward looking article, since the author does not claims that a hedge is impossible (too silly a claim), only that no market to hedge is there yet.