“ For example, you can’t say in cryptoeconomics, “It’s illegal to bribe people,” because there’s really no simple way to define what a bribe is. If someone really wants to bribe someone else, he can just go and do that outside of the protocol, and the protocol would have no way to tell.”
(Vitalik Buterin, Ethereum, July 2018, on Conversations with Tyler)
Right now it feels as if there is almost no problem in the world that the blockchain and similar distributed-ledger technologies (DLTs) are not going to alleviate. Taking a deeper look into this will be one of the eminent tasks for every policy analyst, including our group of NGO future explorers — and all kudos to Mercy Corps for putting out a really interesting report to get us started, for IDS to summarize some of the related hopes in such a succinct way and for the European Parliament to draw out some policy implications to name just a few interesting reports on this topic.
While I share a lot of excitement about the blockchain and really hope it can evolve into a technology that helps make the world a better place I have to admit that I still do not quite get it. I am neither a tech expert, nor someone who has so far had the chance to dive deeply into the related literature, so here just a few questions that truly puzzle me from an amateurish/naive perspective
1. can an unproductive frenzy of solving progressively more difficult cryptographic puzzles be part of the core architecture for our trusted-transactions future?
Pictured here is a server farm solely dedicated to mining bitcoins, the incentive system to keep the much hyped blockchain going. And other distributed ledger systems built around tokens and proof-of-work verification go down the same route as I understand it, although some of them are a bit more computationally efficient. Can such a frenzy that already in current blockchain infancy engages thousands of purpose-built computers, gobbles up the low-cost energy supplies of local communities, fuels large-scale hacking attacks to hijack processing power for mining, and that is expected to consume by late 2018 per year more energy than the entire country of Austria be the building block for new financial systems and so many other things? Can a system that enables the most potent of these contributors to disproportionately influence both the value and future development of the fundamental technical architecture of DLTs truly underwrite hopes for a radical democratization of many societal arrangements? Take bitcoin for example. As of mid-2018 two Chinese miners alone own more than 51 percent of mining power and thus the entire system hinges on the trust of the bitcoin community that these two do not team up and decide to game the fundamental proof of work mechanism and thus the entire system. The promise (spectre) of enormous economic and, yes of course also political power that comes with a potential ability to shape this architecture and its future build-out is already fueling fierce economic and geo-political competition for influence that somewhat reminds one of the Internet wars about root server control, DNS governance etc. that did not exactly lead us down towards a democratised Internet infrastructure. So this time is different?
2. how can a distributed ledger system that publishes the entire history and current state of the ledger across the entire participating system be computationally efficient and in the longer-run sustainable?
Even if only some of the envisioned blockchain applications are posting fast growth in transaction volumes and even if only some of these front-runners are sharing one and the same infrastructure, wouldn’t that ledger grow out of bounds in no time and start consuming large chunks of the global computational capacity, and thus quite a bit of the world’s energy to just keep going? Or would we see a big fragmentation into lots of different distributed ledger infrastructures, thus raising another round of questions about critical mass of participants, trust in individual architectures, protections against hijacking by dominant players…
3. is digital representation always tightly link-able to the real thing?
Supply chain integrity and transparency is just one example of the many DLT applications that aim at validating the identity and flows of physical objects. Yet, wouldn’t such systems stand and fall with the ability to actually police the unbreakable connection between physical widgets and their digital identity? This may be possible for some items whose physical properties are sufficiently distinctive to be converted into unique digital identities, such as diamonds. But I do not quite get how this would be feasible for more standardized things and how it would be economically feasible in the first place for a myriad of high volume, yet low value items, such as chicken. And how is this supposed to work out in a context where collusive practices might occur and several of the transaction partners along the value chain have a strong incentive to collectively manipulate origin/identity or physical qualities of the objects involved? The verification and trust architecture that ascertains the integrity of the physical-digital identity link would thus require a much more sophisticated and costly apparatus that goes way beyond a validated electronic trail as currently provided by DLTs. Some automated time-stamping, geo-tagging and provenience-tracking techniques might offer validation features for some product attributes that are more difficult (but far from impossible) to tamper with, yet many other entry points for fraud at the information-physical product nexus are much more costly to protect against. At the end of the day it remains very unclear to me how trust in the integrity of all actors along the supply chain (from raw materials to individual units of standard commodities) could be supplanted by DLTs other than offering consumers a rather symbolic technology-branded information trail.
4. garbage in — greatness out?
Now there is a lot of excitement about DLTs in land management, and I can see that it could prevent corrupt officials from fraudulently changing specific entries to move ownership of particular plots in return for bribes or for other purposes. This is a significant problem, give that for example more than 60% of people surveyed in 34 countries believe that it is very likely that rich people could get away with paying bribes to falsely register property that they do not own (Afrobarometer R7 2016/2018). So fantastic to have new ways of stamping out that type of fraud. Yet the bigger issue seems to be (including for many countries for which DLT is promoted in land management) that the existing land registries are often woefully incomplete and inconsistent, overlapping with customary law and historical usage practices in complex ways. Putting a DLT on top would thus either simply codify current irregularities. Even worse and at least equally likely, it could also amplify unjust conditions. This could be the case when DLTing of land claims runs into the same problems as many conventional land regularization projects. The more savvy and powerful tend to readily avail themselves of these opportunities and formalize their entitlements, while the more marginalized without ready access to and knowledge about the bureaucracies and entitlements involved loose out despite holding potentially equally legitimate or even superior claims. And finally, would DLT really solve the false input problem mentioned at the outset? Given that such systems would most likely still have to rely on some degree of intermediation and delegated authority for data input and validation this is not a given at all. It might simply move the bribery/manipulation problem into new institutional settings.
5. the fate of transparency — from creative accounting to creative DLTing?
I am not an accountant, but am always in awe as to how complex and open to creative interpretation even simple accounting for resource flows across and between organisations can be. Things get really messy and entail lots of judgement calls, when a bit of time-shifting, valuation issues, and simple derivatives enter the fray. Alas I fail to understand how this might be different in a fully evolved DLT world, particularly when algorithms that machine-learn and that design and execute on forward contracting and n-th grade derivatives might soon be at work to take such alchemy and inter-dependencies to unimaginable levels of complexities, inconsistent levels of temporal precision, incompatible if-then execution orders and thus quite possibly maximum human inscrutability. I would thus suspect that the radical transparency of simple transactions will soon cede to an overwhelming obscurity qua complexity and non-linearity that is close to impossible to untie, retrace and represent in intelligible terms. If we fail to understand the dynamics and implications of high-frequency trading in what eventually is a very bounded and simple market exchange, how are we going to grapple with such dynamics in a rather unbounded DLT-enabled contracting and resource flow space? Add to this the likely migration of crucial trading operations from carefully regulated public exchanges with stringent transparency and price disclosure requirements into private blockchain clubs run by groups of financial institutions or value chain leaders and the prospects for a DLT-ed age of enlightenment do not exactly look splendid.
6. how smart can smart contracts be?
Here I am on particularly thin ice since I do not claim to understand this at all. I am thoroughly puzzled by the hype around smart contracts. I get that it is very useful to immutably commit to a specific future action triggered by a specific future event without a trusted intermediary introducing inefficiencies and consuming rents in the middle. But isn’t this just confined to the — (pardon) dumbest segment of contracts out there — contracts for which all related states, acts and contingencies can be perfectly specified and monitored? This condition of what Williamson in his seminal work on transaction costs has called “contract utopia” might underpin some of the most simplified economic models in circulation but it is something that is rarely (ever?) approximated by reality. In the real world contracts are pretty much always incomplete. Disputes around contracts almost always revolve around different interpretations of complex realities, competing claims that need to be carefully re-aligned when mutual trust among transaction partners is high and adjudicated at court when this is not the case. I am neither a lawyer nor a coder, but cannot see how this ambiguity could be possibly coded away and nicely solved through so called smart contracts.
[update: Frank Pasquale has made a similar argument in early 2018 here, albeit much more systematically and eloquently]
It is simple to write and encode a contract that triggers payment when a cargo shipment of jeans is received at port. But what about if the cargo vessel is infested with moths that ate half of the pants along the way? What if the jeans are badly sown together or the shipment sat on the dock and incurred extra costs for four more days since the receiving cargo trucks arrived too late? If everything goes according to plan a smart contract does not seem to add much value. If things go off-rail a smart contract is unlikely to sort out the situation. From the very outside it looks like the DLT eco-system is far away from resolving such messy reality issues, even in its own architecture. And then there are still significant challenges for building secure interfaces and validation mechanisms for rather simple, unambiguous real-world information inputs, such as the yes/no information whether a flight was cancelled that could trigger an automatic pay-out for a smart insurance contract. Even the most simple variant of this so called Oracle problem is far from being resolved satisfactorily with some proposed solutions creating another bloated DLT system just for the purpose of generating and validating such real world inputs thus further amplifying the DLT related efficiency and resource issues, while other solution aim to accept external single-source inputs that thus sacrifice the alleged core advantages of distributed and thus hard-to-manipulate information processing with all its built in checks and balances.
Plus, it looks like there are additional issues when it comes to smart contracts in the financial sphere. Aren’t many business models or even simple future transactions such as rent payments premised on funds that at the very moment of the promise being made are not yet available and cannot be ring-fenced and held in escrow to generate air-tight trust that the transaction will be executed as anticipated? Instead, the real trust part of a financial promise is not that a future execution will indeed be done but that this execution will be possible in the first place. The real trust attaches to the anticipation that at the time of execution the promising party will have made the resources available to be able to fulfil the promise (or transferred the commitment to other parties that are in a position to serve the contract). Isn’t this the very principle that underpins financial leverage and simple future financial promises alike? The trust involved is the trust that the counter-party will competently and in good faith deal with future uncertainties to be able to serve the contract or enter with good-will into negotiations for re-adjustment, if this becomes necessary. The trust that the actual transaction will be executed and thus the type of trust that smart contracts seem to supplant is rather marginal. Perhaps this is relevant when dealing with actual fraudsters but less so in every-day business relations with counter-parties that inevitably take risks and face uncertainties about future events. Of course there are nuances to this and smart contracts can lock-in the superiority of some claims vs. others in the case of bankruptcy, or generate a chain of contractual claims so that rent payments can be directly taken out of monthly salary payments by employers, but overall it looks to me this skeptical conjecture holds. And (a 2022 update) this is further backed up by a first wave of empirical analysis that shows how smart contracts related to initial coin offerings are often embedded in and ultimately governed by broader architectures of code that sometimes comes with secret backdoors that offer higher-level options for manipulation to the code provider, unbeknownst to the smart contract counterparty.
Now given that I have not had time to dig deeper into the workings of the blockchain these questions might be somewhat simplistic. Some of them may be easy to address and put to rest by experts and I am sure some of the remaining challenges are actively worked upon already. Yet so far I have not found any satisfactory replies and my skeptical puzzlement persists (and as a quick update CGD just put out an interesting report that seems to confirm some of my concerns raised here). So some efficiency gains perhaps on the horizon, provided they can be more efficiently computed? But beyond that little real hope to fix broken systems, little hope to force integrity where integrity is not already an incentive for most players involved and very little hope to provide transparency when complex algorithms come into play. So still some way to go for the blockchain to live up to its hype?
Yes, but as a 2018 addendum to this text, let’s end on a positive note.
Having by now sat through some blockchain conferences and workshops I begin to believe that the performative power of DLTs might indeed be real and offer some hope for positive impact even if this comes in a bit of a warped way. There are many very pedestrian data standardisation and integration projects out there that would make a huge difference for all kinds of governance, market and broader development issues, yet that remain fallow or have gotten stuck due to insufficient budgets and/or a lack of enthusiasm, and buy-in by key players. Think for example supply chain integrity that could be greatly helped by stakeholders adopting common data standards and integrated systems. Putting the label blockchain on such initiatives, irrespective of how much DLT actually ends up in the package could help kick-start or re-energize such data flow harmonisation efforts. It offers the opportunity to unlock fresh funding. And it serves up stronger incentives for some essential but previously reluctant data holders — public and private — to jump right in and put on symbolic display a DLT-certified agility and tech-savvy that supervisors, political principals and the broader public might just come to expect of any organisation in the era of happy blockchain mania. No. I am not endorsing cheating. The good-faith imagined affordances of DLT may just provide sufficient impetus for making this happen. Or more tactically why not ride the rhetorical wave and opt for a selective embrace of specific blockchain features and principles for particular contexts and particular information systems challenges? That sounds just reasonable. And it would not be the first technology that created real self-fulfilling impact through its hype, rather than its core functional substance at a given point in time. So let’s see where this goes. Apparently we will find out more pretty soon at least when it comes to DLTs in the area of governance where according to a recent mapping more than 80% of projects seek to demonstrate first impacts by 2019. As of late 2018 the picture does not look too great, with many blockchain-in-development projects still lacking stats on impact and even basic transparency about how operations have fared so far.