One currency to rule them all

JP Morgan recently came up with a digital currency called JPM coin. This currency or some such will make crypto  currencies useless. This blog discusses the reasons for that conclusion.
Addendum: Facebook's libra is similar to JPM except it is based on a basket of currencies not just the US dollar.

Brief History of Currency. 

At the dawn of human history there was no private property. The tribe honcho decided who gets what. Next came division of labour and job specialisation. People had to exchange goods because they could not or would not produce everything themselves. This barter system gave way to one commodity being used as currency. Cattle, sheep, salt , vegetables and other commodities were used as currency.

Later the ruler issued a token such as  gold or silver coins to represent the commodity. These coins were then replaced with cheaper tokens like paper and metal coins that represents a fixed amount of say gold or or silver. The government issuing the tokens began to issue more tokens than they had gold or silver to back them up with. Quantitative easing is the modern name for this ancient practice.

The US dollar used to be backed by gold for a long time. Bretton Woods convention established the US dollar as a reserve currency pegged to the price of gold. This came to an end in 1973 when the Bretton Woods Agreement was  formally dissolved.

Now a US dollar represents nothing other than the US dollar. It is not underpinned by any commodity. 

Bitcoin

Satoshi Nakamoto, who is rumoured to be a group of individuals, sprung a new digital currency called Bitcoin in 2008. This currency has only a digital token. Admittedly, most currencies are by and large digital currencies with only a small portion of it in the form of currency notes and coins. The difference is that there is no central authority controlling the flow; only a set rules that govern the creation and transfer of bitcoins. Bitcoin is a distributed ledger. All transactions are public, although the physical identity of an account holder may not be known. After more than ten years the actual identity of Satoshi Nakamoto is not public knowledge. 

After the initial creation of coins new Bitcoins are minted by a process known as mining. This limits the rate at which new coins can be minted.  Since then many crypto-currencies have sprung up. Almost all of them are based on the same model with some distinct attributes. The fact that crypto-currencies are initially created out of nothing makes it look like a Ponzi scheme and has many traditional economists worried.

The original paper by Satoshi Nakamoto is well worth reading. A modicum of understanding of  public key cryptography is enough to appreciate some of the issues that the paper addresses and the solutions it suggests.

JPM Coin

J P Morgan Inc. (JPM) after stating as much has now come up with their digital own currency with a crucial difference in that one JPM coin is backed by one dollar. Why bother with an intermediate coin? Why not just use the US$ as the reserve currency?

Within the US, JPM coin makes no sense. Outside the US there are a number of benefits. If there are n countries trading then there are n(n-1)/2 exchange rates. But with JPM coins there are just n exchange rates. Thus if two banks dealing with currencies X and Y have sufficient JPM coins to their credit then the transfer between the two can be done outside the purview of existing international transfer systems like SWIFT. In other words it will be as simple as transferring money from one account to another within the same bank.

For a start JPM coin will be restricted to institutional investors of JPM. But JPM plans to extend it to individual users. For example one could buy JPM coins in England; and then assuming JPM operates in New Zealand, have those JPM coins sold for NZ$ and the money transferred to a New Zealand account; all in a matter of minutes.

In addition to speed of transfer the other advantage would be stability. The US$ changes slowly. Twenty years back one Australian dollar bought US$ 0.66 now it buys US$ 0.70 or thereabouts. Hence keeping the JPM coins without converting to local currency will not incur much loss.

Crypto-currency is a Passing Fad

 Most crypto-currencies will die due to a number of reasons. Their currency supply cannot expand fast nor can they contract at all. JPM coin can expand or contract depending on economic situation. Secondly JPM coins are more stable. This will reduce speculation and the coins will be used mostly for exchange. People outside the US may use it to hedge against inflation in their own countries. Finally the transactions can be done extremely rapidly in the order of hundreds per second, a far cry from what VISA does now but still much better the crypto currencies like Bitcoin.


I would speculate that JPM coin or its equivalent will replace both centralised  crypto currencies like Ripple's XRP or decentralised blockchains like Bitcoin for the most part. Only a  few anarchists who do not believe in the state or government having a monopoly over currency will find any need for crypto currency that is not linked to something external like the US$ or the price of gold. The rest of us are better off dealing with local currency and using some standard like JPM for exchange between currencies. 

Machine Learning is Learning a Machine more than a Learning Machine

What is Machine Learning

Before we ask what Machine Learning is, we would do well to clarify what we mean by a machine. In normal usage we refer to any device that performs a job as machine whether it is a kitchen appliance or a lathe. In computing, a machine is not something physical. Rather it is refers to a type of computer program. Just like a physical device, a machine in this case is a program that can be configured to perform a task subject to some rules. Let  start with a definition:

Definition [Mitchell]: A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P if its performance at tasks in T, as measured by P, improves with experience. 

Informally, a machine learns when its performance improves with more data. Consider an example like weather forecasting. We would like to create a machine that could predict the maximum temperature to-morrow. To do that we could create a model that computes likely values given today's temperature, humidity level and other such measurements. Assuming that it is workable model, the learning consists of generating a function f(temperature,humidity,...) that computes the predicted temperature. One way of modelling this function would be to compute a linear combination of predictors:

β₀ + β₁ *temperature + β₂*humidity.

In this case learning consists of computing β₀ ,  β₁ and β₂ based on past records such that predicted value matches the actual results closely. One would expect that the prediction will improve if the volume of historical data is increased. This improvement is referred to as learning.  What has just been described is known as linear regression, a technique known for more than a century. There are many models that can be used. It is possible to combine many such models to provide a composite machine.

 

Notice that the machine cannot pick the predictors at random, although given a large number of predictors the machine learning model can select the best predictors. The choice of predictors is still mostly human. However a day may come when there are a large number of machine learning systems and a meta learning program can identify from its vast repository of learned machines the most suitable factors that can be used to model a new learning task. Although we are not there yet, programs like IBM's Watson give a good idea of what the future portends.

What Machine Learning is Not

Machine learning involves uncertainity. When a problem is well defined it is no more a case of Machine Learning; it is a search problem or an optimisation problem. 'Where is the closest Starbucks?' is a search problem. 'Will there be a Starbucks within a 5 mile radius one year from now?' is a machine learning problem.

Applications of Machine Learning

Based on the past growth it is safe to assume that there are still a large number of applications of machine learning that people have yet to discover. To discover the promising new ones it would be good to have an idea as to what the current applications are. As Steve Jobs once said "Good artists copy, great artists steal." Here are a few:

• Computer Vision
• Text analysis to authenticate literary works
• Spam detection
• Speech recognition
• Speech synthesis
• Robotics
• Prediction: Which candidate is more likely to complete a Ph.D. Thesis?
• Which toddler is likely to have learning difficulties?
• Which employee will stick around for five years?
• Recommender Systems for movies, books, fashion accessories
• Fault diagnosis of machines
• Medical Diagnosis
• Performance Analysis and Tuning of Computer Systems
• Predicting house Prices
• Economic Forecast

What I would like to see:

• Predicting marriage success
• Predicting potential terrorists
• Predicting potential mass murders
• Analysing cargo movements to detect contraband trafficking.

and on a more techical level:

• Generating Unit Tests
• Software debugging using crash dumps and code analysis
• Analysing network protocols to predict failure and congestion scenarios and
• Analysing multi-threaded applications to locate deadloacks and starvation

Reference:
[1] Tom M. Mitchell, "Machine Learning," McGraw-Hill, 1997.