Month: April 2020

The Uses of Distributed Ledger Technology (DLT, or cryptocurrency)

I’ve been following the development of distributed ledger technology (DLT i.e. cryptocurrency) Hedera Hashgraph for a couple years, since hearing an interview with US Airforce Academy mathematics and cryptography professor Leemon Baird on the Hidden Forces podcast. I cannot judge the quality of the mathematics, but he seems to be a credible scholar with lots of publications, including his 2015 paper laying out the mathematics of his elegant patented algorithm. I am not qualified to judge whether his “hashgraph” protocol is better or worse that those of other serious cryptocurrency attempts at creating a fast and secure “consensus algorithm,” like Ethereum, Cardano, Algorand, or Ripple, that make similar claims to Hedera. But the implications of the whole picture below suggest a major role for Hedera in the biggest potential use cases in the sector. Positive mentions of Hedera and Hyperledger in a recent publication on blockchain interoperability by the World Economic Forum supports this case.

A few words of caution are in order. In addition to my own limitations in technical and mathematical understanding, anything I say here is just as speculative as the entire speculative “sector” of cryptocurrency. While Bitcoin and others have offered experimental evidence that you can buy and sell things using cryptocurrency, most activity in cryptocurrency is merely trading one for another. In other words speculation. When you buy a stock, you buy fractional ownership of a profitable concern. A mutual fund is a bundle of such assets, of which you own a small fraction. A bond is a promise by a business or government to pay you a little more than what you initially paid in, regardless of whether business or the budget looks good or bad. The prices of such assets rise and fall as a result of the demand for them, which reflects a communal (market) evaluation of the profitability and safety of the asset. Reality is more complicated of course. But in cryptocurrency people pay good money for some ephemeral digital rights that have no intrinsic value nor legal status as of yet. They have initial value because of a communal (market) evaluation of their future utility and therefore value. But it is a very immature market filled with people who have little idea about how to value a cryptocurrency other than assessments of various kinds of market manipulation, whence cryptocurrencies derive their volatile speculative prices.

As explained above, I don’t think an “investment” of buying a cryptocurrency is wise on the balance. As the more experienced traders in this sector say, “don’t invest what you can’t afford to lose.” It is mere gamble. Better to buy a mutual fund, especially after the recent drop in the stock market due to the ongoing, and tragic, coronavirus pandemic. Better to give a donation to a local charity. Cryptocurrency is an entirely speculative sector, launched in 2009 with the invention of the entirely experimental hobbyist trial called “bitcoin,” which has succeeded far beyond the author’s original intention, and has also shown that its limited experimental algorithm is not able to function at a global scale as a usable currency. But it will likely retain some exchange value as a collector’s item from our current era of economic uncertainty, inequality, and institutional distrust. Hedera’s “coin,” the HBAR, is not designed to appreciate rapidly in value, as it will be continually released over the course of 15 years, and because institutional users will only need a minimal amount of HBAR for most uses as they can use mirror nodes with the Hedera Consensus Service (HCS) which charges dollar-denominated fees (paid in HBAR) that are in the fractions of pennies and won’t automatically rise even if the price of HBAR rises.

I started following cryptocurrency since the mind-boggling New Yorker article on the quirky birth of bitcoin in 2011. I “got it” (Bitcoin derived value from its utility as an alternative payment process) but did not think it would amount to much and could neither afford nor understand how buy or “mine” bitcoin. I suppose I could have made a lot of money. But despite bitcoin’s strange speculative rise, it has not yet found any real-world uses except for a bit of criminal activity and sanctions and tax avoidance. So this essay considers some of the more prominent possible applications of DLT, which are being considered: national digital currencies and interbank transfers.

As explained by a Ben Buchanan on Michael Morell’s Intelligence Matters podcast, the transformative implications of DLT are much overblown, and the more revolutionary technology being used now is Artificial Intelligence, which has huge uses, both scary and useful, and the big tech companies and China are already far along in putting AI to use. (Quantum computing, he says, has not yet matured into a transformative technology). DLT, I expect, will be useful for some marginal improvements to various processes, from money transfer, to supply management, to secure communication, but it is not a transformative technology as the “bitcoin maximalists” and other blockchain ideologues suggest.

I wrote the rest of this up last Sunday, before the post about HCS on Hyperledger, but that only reinforces the argument that follows. There was also a post about Deutsche Telekom moving to Hyperledger and using RAFT, as if this points to a problem for Hedera. It does not. HCS is better than RAFT, and provides the essential “ordering service” Hyperledger has been looking for, that RAFT (and previously Kafka) currently serves. Hyperledger Fabric was and is basically developed by IBM, a Hedera council member. All the evidence is that companies are looking to Hedera to serve as the underlying ordering service for Hyperledger. Then Hyperledger (Fabric) can provide the infrastructure for a wide range of specific enterprise uses of DLT. Finally, take note that the HCS-Hyperledger presentation made quick mention of Quorum being able to work with HCS. Quorum is JPMorgan’s open-source DLT project in development, built on Ethereum. It can easily port to Hedera consensus, and probably will, as Ethereum’s proof-of-stake transition is still stalled.

What follows is an extended comment on Hedera Hashgraph as it relates to two monetary use-cases: central bank digital currency (CBDC) initiatives, and inter-bank transfers. Both of these are still years away from any sort of universal application, but we may see experimental applications within the next 2-3 years. I think the more immediate use cases are in non-monetary management of secured databases, both public distributed ledgers and private or permissioned ledgers that have a need for an external validation control mechanism. Non-monetary applications could include everything from things like Acoer’s coronavirus tracking application to various kinds of inventory and supply chain management to the validation of credential dossiers to secure communications (both human-to-human and machine-to-machine and combinations thereof, e.g. power plants). Small scale experimental versions of all of these kinds of applications are already happening on Hedera Hashgraph. The key to all of these is to demonstrate a particular and competitive service that finds a need in the current marketplace. The surest “bet” on cryptocurrency would not be an investment in a particular token like HBAR, but in a company with a marketable application that uses DLT in some way, and Hedera is a very attractive DLT because of its credible claims to speed and security that few others can rival. Any large scale application of Hashgraph will, presumably, push up the price of HBAR. But for reasons given in previous extended comments, growth in demand for HBAR tokens may, in fact, be lower than expected since, as the price of HBAR goes up, the price-in-HBAR for its services goes down, since they priced in dollars, not HBAR; and because so much work can be done via mirror nodes that periodic authentication of an enterprise DLT on the Hedera mainnet need not be all that frequent. So, the best crypto “bet” is a normal equity investment in a good start-up using DLT for a marketable service. A survey of the current projects on Hashgraph would probably reveal some worthy candidates for investment.

Nonetheless, CBDC and interbank transfers both represent massive use-cases for Hashgraph that would create strong ongoing demand for HBAR if they were implemented on Hashgraph because both would send huge number of transactions or HCS queries through the mainnet. Several countries are looking at CBDC, including most recently France, which put out a call for proposals. South Africa put out a very sophisticated call for proposals last year. Other countries are fairly far along in experimental versions, including China which has announced an imminent public implementation, which will probably be based on technology developed by the online retailer Alibaba, but researchers at the National University of Defense Technology in Changsha have researched Hedera’s technology. A good assessment of CBDC and its necessary functions and trade-offs can be found in a recent report by the Bank of International Settlements (BIS) One thing that is clear is that there is no way Bitcoin can serve any useful role in CBDC, mainly because it cannot offer finality nor erasure of the record of past transactions, both of which will be necessary for CBDC. Bitcoin enthusiasts will still argue for its utility as a separate, autonomous, decentralized currency, and it has certainly been a grand experiment. But that is really all it can ever be. Most who argue for Bitcoin’s utility as alternative global currency base themselves upon completely flawed assumptions derived from a half-baked devotion to Friedrich Hayek. What the last economic crisis and this one have demonstrated, as Adam Tooze showed in “Crashed”, is that the basis of value for a currency is not some simplistic supply-and-demand chart, with higher valuation linked to limited supply. What gives value to a currency is its utility, and this means liquidity. The efforts of the US government and its Federal Bank to ensure liquidity during global economic crisis have only heightened the demand and value of the US dollar. It is possible that at some point, confidence in the dollar will collapse for reasons related to the federal debt, but that is some ways off still, and when it does happen, although there probably will be a spike in the price of cryptocurrencies, the world will simply settle on to another politically-managed sovereign currency. Currency is not a technical or market creation, it is a political creation that requires a political process in its management. Regardless of all of that, the lack of finality and erasure of past transaction records on public databases, will prevent all proof-of-work DLTs from serving as the basis of CBDC.

A secondary reason for Bitcoin’s ineligibility for CBDC is the lack of flexibility for KYC AML etc. These as well as lack of finality also rule out Bitcoin and its like for interbank transfers. So proof-of-stake with finality and asynchronous Byzantine Fault Tolerance (aBFT) are basic requirements, both met by Hashgraph. And then we have the issue of TPS throughput, and again Hashgraph comes out very strong, at least within the limits of its current governance model of a (eventual) council of 39 members each running a permissioned node. Under this governance model, the distribution of stake (and the value of these stakes) are not at issue for security. The presumption is that large public companies and institutions will not risk their reputations to cheat the system, and that in using it themselves, they have an interest in its integrity. The distribution and value of stake only becomes important when non-vetted operators are allowed to run nodes. At that point it will be important that 2/3 of all HBARs are staked/owned to honest operators/nodes (this is a mathematical limit for all consensus algorithms including proof-of-work). The presumption is that, by that point, the value of the stake necessary to run a node will be so high that it would not be in the operator’s interest to cheat the system, as that would risk a catastrophic drop in the price of HBAR, and in any case, at a high valuation it would be nearly impossible for one operator or cabal of operators to get control of 1/3 of all HBAR in circulation. As long as the 2/3 stake on honest nodes condition is met, HBAR and HCS are secure and all transactions are final. According to Hedera’s own testing. A global network of 32 nodes can offer finality for 100,000 TPS with 7 seconds latency. So this is a promising core consensus algorithm for a CBDC.

One issue that arises in the BIS study is doubt about whether any DLT could offer a high enough throughput for what they call a “direct” CBDC in which retail transactions are direct claims on the central bank, rather than on intermediary banks (as is the case with conventional currencies now). Certainly even Hedera’s 100,000 TPS is not sufficient for this. But in the lengthy Hedera whitepaper they explain a vision of how how sharding will work to multiply the capacity of the system, even if they have not demonstrated it.  I do not at all understand how this would work from a programming or mathematical standpoint. But I would assume that any sharding solution would be limited by 1) the inverse relationship between the number of nodes on a shard and the TPS capability on that shard; 2) the network burden that one shard places on another to ensure intercommunication and authentication between the two. That being said, I could imagine a situation whereby each “anchor” shard could have up to 39 nodes, so an initial set of anchor shards could conceivably be something like 39×39 derivative nodes, each authenticating through transactions to other shards. I will assume that this could then expand to 39x39x39. That would equal around 57,000 potential nodes, and thus about 5.7 billion TPS, which may be sufficient for CBDC. Some burden would be placed on system for intercommunication between nodes, but even at 1000 TPS of internal authentication transactions, this would represent a 1% tax on system, and even if all derivative shards authenticate at 1000 TPS to all nodes in an anchor shard, it means anchor shard will bear 39% burden, as trade-off for total capacity of all derivative shards. If this expanded to something like 50,000 shards, that means stake capacity of 50 billion HBAR would allow for 1 million HBAR staked to each shard, which may represent a rationale for the nice round 50 billion number in the first place. … But I am probably just blowing smoke out my non-mathematical ass here, their explanation of inter-shard messaging is actually a much simpler vision….the real point is that Hedera has indicated that sharding is possible and will be able to greatly expand network capacity if current capacity reaches limit.

The next thing to understand is that all CBDC and interbank transfer utilities will be extraordinarily complicated and subject to many changing rules and trade-offs between anonymity and identifiable transactions. The simplistic assumptions of cryptocurrency public/private account keys for transactions on public ledger as analogous to “coins” cannot hold up in court, so to speak. HBAR or any other cryptocurrency will never be anything but a token for network use. If such a token realizes global widespread utility is will settle into an actual market-use value (as opposed to speculative value), and HBAR could achieve that if adopted for these sorts of large-scale uses. But HBAR, or anything like it, will not be the infrastructure for CBDC or interbank transfer. If these use HBAR, it will be for HCS as a consensus mechanism for proprietary or sovereign DLTs.

So, to move to interbank transfers, two years ago the SWIFT cooperative ran a large scale experiment entailing 34 banks to see how a Hyperledger Fabric infrastructure could handle its needs. Hyperledger Fabric is a “network-of-networks” design for enterprise use, contributed by IBM to the Hyperledger open source project, now managed the Linux foundation. IBM is, of course, one of Hedera’s council members. The experiment was a proof-of-concept, and basically worked despite the fact that Hyperledger at that time had no good, efficient, secure “ordering mechanism” (read consensus algorithm). Since then it has tried to use Kafka and then RAFT for its consensus mechanisms, but both are complicated to deploy, and cannot achieve aBFT. But, aside from that, they found that Hyperledger Fabric could provide the needed infrastructure for their needs, and that this would be most useful for small and medium sized banks, since large banks already have their own internal infrastructures for domestic and international settlements. Notably, the experiment required 528 Hyperledger Fabric “channels” and they estimate that a real-world deployment would need over 100,000 channels. So, suffice to say, what Hedera, or any other cryptocurrency, offers does not come close to meeting the infrastructural needs of interbank transfers. Notably, however, Hyperledger Fabric can meet those needs, and Hyperledger Fabric is a modular system, whereby different components can be plugged in and intermixed. What Hedera HCS offers Hyperledger Fabric is an efficient and secure ordering mechanism (consensus algorithm), something which up to now has been sorely missing. This likely explains IBM’s interest in Hedera, and the potential for Hedera HCS to offer a key service for both interbank transfers and CBDC .