Effective Altruism's Soft Capture Scenario

Why Organizations Must Be Social Sovereigns

This is the third entry in the EA Global DC series. It is based off of my experience at that conference and a variety of other ideas relating to Effective Altruism. This article is addressed more directly to EAs than the others, but it is also useful in the general case for any organization or movement that wishes to avoid political or social capture. 

Today, let's consider a hypothetical scenario. Competing for political power is as close to a zero-sum game as it gets, so it is only inevitable that EA becomes an enemy of some ineffective altruist organization, otherwise known as a normal NGO. These organizations may be ineffective at doing good, but they are reasonably effective at wielding media and political influence. When NGOs compete, a typical strategy is to conduct opposition research on opposing groups to discover spurious but politically impactful scandals about their personal lives. Unless all prominent EAs truly live like monks, they are bound to discover something. Let's say some EA figure is recorded using the n-word non-pejoratively and this is leaked to the New York Times. What is the reaction?

One positive case is that the good epistemics of EA win out. Everyone evaluates the cost-benefit and decides that saving humanity from extinction is more important than saying the n-word. They work together to rally around the figure, create a united front and reject the New York Times. EA remains a single movement. 

Another case is that EA splits into factions. Some organizations denounce the figure while others defend and necessarily reject those conclusions. With these scandals likely being Scissors in nature, this disagreement is unlikely to be mild. This has occured numerous times in the past, particularly within progressive organizations directly under the influence of left-wing media. While I believe the average EA is significantly more intelligent than the average member of those groups, they are not without brand loyalty and social desirability bias. Evidence for left-wing bias include existing inefficiencies relating to the Republican party, demographics (young, coastal, college-educated), and anecdotal encounters at EADC. Consequently, I find the latter scenario to be overwhelmingly likely (~90% probability).

Throughout the EA global conference, the failure mode of *capture* was mentioned frequently. EAs do not want to become a client group of the Democratic party. After all, their existenced is premised on the idea that the current NGO sector, which are mostly client groups of the Democratic party, is ineffective. Most people were generally focused on a naive idea of what capture looks like where EA devolves into explicit obsession over Democratic politics. I'm uncertain if EA will even defend itself against this version; however, the second scenario described in section one is a subtler but almost as destructive version of capture. In this scenario, its organization response indicates that it invests the marginal resource into punishing people who say the n-word rather than saving humanity. 

The lesson here is that for EAs to succeed, its epistemics must be socially sovereign. They cannot be subject to social desirability bias to organizations that are antithetical to its existence. The judgement of individuals within EA must operate by the same rules as the judgement of issues within EA. Achieving this practically is a difficult organizational problem. Tribal loyalty to the New York Times or some similar prestige-oriented organization is fundamental to evolved human nature. Becoming a socially soverign movement means elevating tribal loyalty towards EA above all other tribal loyalties. This presents a difficult tradeoff to navigate, since rational analysis leads us to the conclusion that we must short circuit rational analysis during rapid response PR. Ideally, EA's reaction is similar to the first scenario in part 1, where everyone makes the rational a cost-benefit analysis and no split occurs, but for reasons already described this is unlikely. A more realistic scenario is the following:

The New York Times drafts an attack and reaches out to the victim for comment. The victim contacts an EA organization (for example, CEA) which has already established a rapid response PR team for this scenario. That team contacts a wide range of EA organizations to give advance warning and collectively repudiate the New York Times piece as soon as possible. A schelling point is established through institutional authority and the social pressures now point towards the rational choice, not against it. 

While the initial coordination and response operates fundamentally by appeal to authority, it is meta-rational within the practical constraints of PR. I would argue that it would not count as a violation of rationalist norms for this reason. There is also the possibility that the scope of this response team overreaches to non-spurious attacks, or that it itself is captured and fails to respond properly to spurious attacks. In short, becoming a social sovereign is a necessary but not sufficient condition to avoid capture. 

Tomorrow we answer a related question: are (EA) demographics destiny?