Consensus, Crypto, and Trust
April 13, 2021
1.
Many of the most pressing political questions of our time hinge on the difference between trustful and trustless (social and economic) consensus, and on their relative status as mechanisms for coordinating human action. This is a struggle played out foremost in the architectures of technical and institutional systems.
2.
In an article from 2016 laying out the design philosophy for Ethereum’s proof-of-stake consensus mechanism, Vitalik Buterin argues the long-term security of a blockchain is ultimately provided not by economic incentives, but by social considerations.
Even if an adversary had access to unlimited hashing power, and came out with a 51% attack of any major blockchain that reverted even the last month of history, convincing the community that this chain is legitimate is much harder than just outrunning the main chain’s hashpower. They would need to subvert block explorers, every trusted member in the community, the New York Times, archive.org, and many other sources on the internet; all in all, convincing the world that the new attack chain is the one that came first in the information technology-dense 21st century is about as hard as convincing the world that the US moon landings never happened.
3.
Buterin’s assertion of the primacy of the social layer is heresy to those who locate the blockchain’s emancipatory potential in its capacity to eliminate the social layer. A trustful layer is a corruptable layer, one which can be gamed and distorted. Nick Land’s claim that Bitcoin is the “first intrinsically reliable record” sets its transparent mechanism against the opacity of trust-based social institutions1—the integrity of Bitcoin’s ledger is provided by economic incentives alone, including security against 51% attack. If one node does acquire enough hashpower to outrun the main chain, so the logic goes, it will always be more profitable for them not to attack. According to this line of thought, Bitcoin renders the social layer obsolete, doing away with trust completely.
4.
This dispute makes it possible to consider consensus layers as embedded within one another, forming stacks with different dependencies and affordances. The question becomes: what is the nature of the outermost layer, and how does it relate to its inner layers?
5.
Answer 1: Social Realism (Buterin). The integrity of the economic layer is secured by an outer social layer. The social layer is slow, clunky, and gameable, but ultimately required for long-term integrity. The value of an inner economic layer is that it oils the wheels, allowing transactions to remain safe and lively on smaller timescales. Put together, the benefits of both long-term security and short-term liquidity can be harnessed.
6.
Answer 2: Social Fictionalism (Land). There was never any true social layer, only more tacit and less distributed economic layers.2 Social institutions never really operate on trust, but as Leviathans which monopolise the use of coercive power to skew economic incentives towards cooperative outcomes dictated by the elites that run them. The choice between social consensus and economic consensus is an illusion—there are only different forms of economic consensus. Nakamoto consensus (etc.) should be lauded as the maximally distributed and maximally explicit mechanisms possible. Those still beating the drum for the social layer betray an illiberal impulse to re-centralise and re-mask power.
7.
Answer 3: Social Historicism (Baudrillard). There once was a social layer, but now there is not. The outermost layer—which can never be made fully explicit3—has changed over time, its social structure swapped out for an economic structure. In Baudrillard’s terms, the spread of the commodity-form to the symbolic economy has remodelled the social relation as such, replacing the unconditional pact with the instrumental contract at the deepest level. Since the social layer is required for consensus integrity, integrity is now impossible. Neither will the emancipation promised by Land’s maximally explicit economic consensus ever materialise, because this whole picture depends on a failure to appreciate the necessity of the social layer for long-term cooperation.
8.
Why is the social layer important? What can a trustful consensus achieve that a trustless consensus cannot? As Scott Alexander puts it, purely economic governance mechanisms are susceptible to capture by multipolar traps (prisoner’s dilemma, Malthusian decline, tragedy of the commons, etc).4 To replace human social interactions with cryptographic mechanisms that do not support anything other than economic consensus (by design) is to willingly place ourselves in a Malthusian death spiral. Baudrillard’s point is that this has already happened—it is implicit in the commodification of the symbolic economy, long before crypto enters the picture. Given these initial conditions new forms of cryptographic governance can only accelerate the process. (Scott Alexander seems to agree on this last point.)
9.
Trustless consensus is agent-mediating; trustful consensus is agent-constituting. Where an economic mechanism finds compromise between multiple distinct agents, a social mechanism integrates distinct nodes into a single super-personal agent via the trust operation—i.e. through mutual recognition.5 Social consensus is a property of a recognitive community; economic consensus of a strategic alliance.6 Economic mechanisms converge on Nash equilibria, social mechanisms on Kantian equilibria.
10.
Put differently, an economic consensus can only take shape where future uncertainty is locally quantifiable. Since economic mechanisms only mediate local decisions they cannot integrate global values by definition—when future pay-offs are unquantifiable from local perspectives no consensus can form around them, even if the cost-benefit analysis is stark from the global perspective. This is significant in e.g. questions of existential risk, such as climate change, where local uncertainties are unlikely to become meaningfully quantifiable until it is far too late.
11.
Social consensus solves this problem with a distributed and explicit trust mechanism. Since local decisions aren’t caveated by forecasted value extraction, they are counterfactually robust (even against locally unquantifiable uncertainty). The we-perspective is produced by the suspension of local self-interest.7 (It is hardly a criticism to point out that trust can be exploited—if a trustless mechanism cannot be gamed in the same way it is because it lacks certain capabilities, not because it is more capable.8) Conceived this way social consensus is neither centralised nor tacit. Perhaps Land is right that institutions have often been (perhaps always have been) masks for coercive power, but if so then this just means they were not properly social. The use of coercive force is not an exercise of authority but a failure of authority.
12.
Baudrillard’s local pessimism is globally ambiguous. If the outer layer is mutable, then it can be engineered. Seduced by the radical fungibility of trustless interactions—by a whole world turned to liquid—a slide from the social to the economic strikes a devilish bargain: the collective decision to abandon the labour and obligation of collectivity. If the crytographic revolution marks a major landmark in the engineering of trustlessness, then to move human interactions onto these media is to black-box trustlessness as such—to move it into the outer layer, accelerating the dissolution of the social into the economic. This is both the promise and the peril of technology: to both explicate and black-box, to externalise the norm and then forget it.9 To engage in technical activity is to implicitly engineer the outer layer. What is at stake, ultimately, is the irreversibility of this process. If it is possible to black-box trustlessness, then in principle it should also be possible to black-box trustfulness. On the other hand, if a social consensus to port interactions onto purely economic mechanisms has already been enacted, doesn’t this mean it was the last collective decision that can ever be taken?
Notes
-
See Nick Land, Crypto-Current: Bitcoin and Philosophy, §0.051. ↩
-
And therefore less efficient, for Hayekian reasons. ↩
-
See Wilfrid Sellars, Some Reflections on Language Games. ↩
-
See also In the Place of Conversation. ↩
-
Hayek’s argument that this is necessarily inefficient because it throws away local information only makes sense if a prior and fixed system of preference-bearing agents is presumed. But this is precisely what cannot be presumed, since what is under consideration is the production of the community as a social actor—that a social consensus throws away local information is just to say that a community value overrides at least some individual preferences. But this is a desiderata, not an inefficiency—it is the very property required by coordination against multipolar traps. ↩
-
An analogy can be made with Gödel: 1st-order arithmetic is incomplete, but Euclidean geometry is complete. However, the reason for the discrepancy is simply that 1st-order arithmetic is much more expressive than Euclidean geometry, and it is these extra expressive capabilities that are turned against it by the incompleteness theorem. Similarly, it does not follow from the fact that social mechanisms can be gamed where economic mechanism cannot be that the latter are more capable—the fact that they can’t be gamed may rather point to a lack of capabilities. ↩