Blockchain beware: Old Systems Title and non-Markov processes

“Possession is 90% of the Law” is the famous, unattributed saying in English Common law. In the context of proving ownership of an asset, being in possession of that asset makes a pretty strong presumption that you own it. The problem with assets that are electronically generated, registered and traded is that they do not physically exist. Possession is very difficult to demonstrate in cyberspace … so either (i) out goes 90% of the Law, or (ii) the world reverts back to pen, paper and ink and slows up everything.

A “Blockchain” is an electronic method for recording and verifying title to an asset. The Blockchain technology is touted as revolutionary, enabling rapid electronic settlement of transactions such as purchases or sales of securities which hitherto take several days or weeks to complete. Blockchain algorithms keep records of ownership in a ‘distributed ledger’ from the date the security was first created. A distributed ledger is a broad and diversely replicated database in cyberspace which records all current and former owners of the security. In the event of transfer, each copy of the ledger must be checked and verified to prove the claim to title of the current owner, together with verification of the historic path of ownership. Exact verification and cross-validation with each copy of the record proves title. Fraud is eliminated since it is impossible for a rogue to reverse the clock and insert false holdings in all copies of the ledger proving title history.

So is Blockchain revolutionary and does it offer the ability to cut settlement times down to nano-seconds from several days or weeks? The answer is NO.

Revolutionary it is not. Our legal ancestors worried about proof of title just as much as we do. Historically, therefore, proof of land title in particular required meticulous documentation before any money changed hands. ‘Old System Land Title’, as it became known, required the original of every document from the initial grant by the Crown to subsequent transfers to the chain of owners to be verified and physically given to the current owners for safe keeping. Any gap in the time-line from initial grant to current ownership was prima-facie evidence of fraud. Thus, a document that had faded or disintegrated from several hundred years of storage in damp surroundings could effectively stymie and attempt at transfer. ‘Old System Land Title’ is clearly the forefather of the Blockchain – they both have the common requirement of continuous historical record. But whereas transfers under the Old System require continuity with original documents, Blockchain requires continuity with multiple copies of the same exact document. The unique nature of the original documents is substituted by multiple copies diversely distributed around cyberspace.

Fast it may begin, but slow it shall become. Blockchain verifications require inspection of every transaction at every point in time of every transfer of every security in order to prove title. In mathematics, this is known as a ‘path-dependent’ or non-Markov process. Markov processes have the admirable property that the ‘past and future are independent given the present’ – or put more simply, the current value tells us everything we need to know and we can ignore the historical path. The Markov property dramatically cuts down the computational time that is required to calculate anything related to this process. Blockchains are NOT Markov processes since they explicitly require the time path to be accounted for, and not just once, but many times as a function of the number of distributed copies. Computationally, this will become cumbersome since the number of ‘nodes’ in the Blockchain that need to be checked grows exponentially with the time path itself.

This means that the Blockchain may start out to be very quick, but as the security transfers ownership the computer time it takes to check the distributed ledger at each and every transaction node will begin to become more intensive, eventually slowing the settlement process down. This presents a number of problems. For instance, two identical shares of stock may take wildly different times to settle depending on how many times each has been traded. A buy and hold investor’s stock that is sold after 5 years should be quickly settled, while a share of stock that somehow found its way into the High Frequency Trading world and bounced around a thousand times will be much, much slower to verify and settle. Similarly, longer dated securities and equities in particular, will progressively take more and more time computationally to verify, than, say a short-dated option which expires after a week.

The solution to the computational problem is either to simply add more computational power, or to jack-knife the verification process for securities which are slowing and reset the initial starting point (very much like what the modern Torrens Title system does to replace the Old System). The jack-knife solution, however, weakens security at this node which will attract the cyber-criminals.

Anyone who has experimented with Bitcoin knows only too well how long it takes computationally to establish a ‘wallet’ and to buy or sell bitcoin. If Blockchain becomes the settlement standard for every security transaction, the whole system may one day grind to a halt.