This is a proposed doctrine for political prioritization. It is the set of explicit and implicit principles that guide the Candidate Scoring System; here I collect them in compact form that can be cited and used much more broadly. I figured it might be helpful because EAs (like everyone) seem pretty confused about how to approach politics in a rigorous manner, and because it would be good to have a document which can show other people what they're often doing wrong on the object level when they incorrectly set political priorities. It can be scaled down to the level of individual thinking about politics, or scaled up to the level of major organizational projects.
It draws upon my familiarity with economics, political science, moral philosophy, decision theory, philosophy of science, cause prioritization, the EA community, political social media, and military doctrine, and it is also shaped by my experience producing the CSS reports and viewing people's feedback. It could probably be adapted to prioritization in other contexts if you want.
Foundations and definitions
Political prioritization is the process of selecting, ranking and scoring political objectives and matching the appropriate response to them, considering the ability of the audience to influence political processes.
An objective is an outcome of political processes that may be be achieved by the efforts of the audience participating in these processes. Political processes can be electoral, executive, legislative, judicial, or sociopolitical. Political processes can be influenced by votes, activism, or policymaking.
The purpose of political prioritization is to enable the audience to take the actions that best meet the goals of the Effective Altruist philosophy. The emphasis is on identifying actions which will have the greatest positive impact on political systems.
Principles Guiding the Political Prioritization Process
Provide useful conclusions to the audience
The sole purpose of political prioritization is to enable the audience to direct their actions towards valuable goals. Conclusions should follow five criteria in order to be useful.
Conclusions should be accurate - this is self-explanatory. Conclusions should be decision-relevant - they should have concrete implications for the choices of actions taken by the audience. Conclusions should be complete - they should not require the audience to engage in further evaluation efforts in order to select the right actions. Conclusions should be accessible - they should not require the audience to draw upon any particular education or knowledge base in order to select the right actions. Finally, conclusions should be unambiguous - they should not suggest conflicting actions for those with plausible differences in point of view.
The prioritization process must be structured to produce conclusions that can satisfy these criteria. Those conclusions must then be written and presented in a manner that does satisfy the criteria. Of course, the prioritization process should not ignore genuine uncertainty in order to produce a definite conclusion. If the prioritization process cannot produce conclusions which meet these goals to an acceptable degree, then it is better to recommend no action.
Follow a coherent conception of EA goals
The prioritization process must be driven by a coherent conception of ethical goals. This involves axiology, the judgment of what is valuable or disvaluable and by how much. It also involves normative decision theory, the judgment of how we should respond to values and probabilities.
Complete theories with specific judgments for everything are rarely necessary and often difficult to justify. However, incorrect assumptions do not automatically render conclusions invalid - they merely make them progressively less accurate as the true assumptions differ from the assumptions used in the prioritization process. Therefore, it is acceptable for the prioritization process to be driven by a complete specific normative theory as long as the leaders of the prioritization process are sensitive to how things might change with plausible variations in assumptions.
The prioritization process can pursue a variety of goals. However, certain principles are universal or near-universal in the Effective Altruist movement. First, improving well-being is generally the biggest priority. Second, the well-being of a wide range of patients (typically including foreigners, animals and future people) must be considered and aggregated without scope neglect or parochialism. Third, serious attention must be paid to unlikely outcomes, and the decision theory should be closer to expected-value maximization than people's naive instincts are.
Defer to expertise when it is relevant and available
The prioritization process should have a theoretical framework of epistemic modesty, using consensus among experts as the highest standard of evidence on empirical matters. Participants in the prioritization process must be able to judge expertise at the individual level and be able to evaluate the collective rationality of groups of experts.
Participants should have a nuanced, eclectic view of expertise; they should give varying amounts of weight to people and groups with varying degrees of expertise rather than drawing explicit thresholds. They should also have a flexible view of expertise, where expertise only exists relative to particular topics. Whenever expert surveys are lacking or incomplete, the prioritization process should look at meta-reviews, studies, blogs/commentary/journalism or bare arguments.
Participants in the prioritization process must be able to integrate these different types of evidence into an overall picture to judge an issue. They should be aware of the implicit value judgments and gaps between external sources and the social good targeted by the prioritization process. Rarely does any study or expert survey directly target global well-being, rarely do external sources understand the dynamics of global well-being as well as does a typical researcher in the EA community, and rarely are they guided by completely right ethical goals. Understanding the true implications and meaning of a source or argument is just as important as judging its soundness.
Further reading:
Lewis (2017). In Defence of Epistemic Modesty.
Yudkowsky (2017). Inadequate Equilibria: Why and How Civilizations Get Stuck.
Avoid sweeping worldviews
Judgments of political objectives should be directly grounded in the ethical goals as conceived by the leader(s) of the prioritization process, and analysis should focus on answering the actual questions at stake rather than satisfying some grander narrative about society. The prioritization process should not be driven by the sweeping assumption or rejection of entire political ideologies and scientific methodologies (insofar as expert consensus on these things is lacking). Instead, it should extract the lightest arguments and ideas that may be useful for answering the question at hand, and treat them as independent pieces of evidence just like any other.
Further reading:
Acemoglu and Robinson (2015). The Rise and Decline of General Laws of Capitalism.
Make the prioritization process robustly inclusive
To maximize accuracy, the prioritization process must be open to input from a wide range of ideas, sources and individuals. They should be included based on their potential to help answer the questions at hand, never outright excluded for dogmatic or social reasons.
Participants in the prioritization process should be cognizant of psychological and social blind spots which may lead them to ignore or misrepresent certain points of view, and should deliberately seek out evidence and arguments from people and sources which they may have marginalized. This is a continuous and cyclical process, driven by constant re-evaluation of one's current social and ideological position. It should be grounded in the cognition of the individual and the characteristics of his/her local environment. It should not be confused with the adoption of a fixed set of rules about which groups are unfairly marginalized in society at a broader level.
Elements of the Political Prioritization Process
Leadership
Leadership of the political prioritization process starts with selecting its basic goals, estimating the available time and resources, and then implementing an appropriate project structure. Contributors are gathered to provide relevant sources and arguments, directing them to focus on things with high value-of-information. Project goals are stated clearly enough to facilitate effective modeling. An overall picture is maintained of the tradeoffs and uncertainties surrounding the project, judging when it is appropriate to release, revise or close it.
Contribution
Contribution to the political prioritization process is the identification and writing of specific evidence and arguments to meet the priorities as set by leadership. Contribution involves judgment on specific issues to provide the information relevant to modeling.
Modeling
Modeling is the explicit integration of judgments on specific issues into an overall picture with useful conclusions meeting the goals of the project. Modeling can make use of quantitative or qualitative tools.
Solicitation
Solicitation is the process of gathering evidence, arguments and goals from the intended audience and from other outsiders in order to contribute to the prioritization process. Solicitation should sometimes - but not always - be packaged into a two-way communicative engagement process that shows others the nature of the project, how their input is (or is not) being included, and how their input could (or could not) change the conclusions of the project.
Dissemination
Dissemination is the sharing of the project products with the intended audience. Dissemination should always be followed by two-way communicative engagement that answers the questions and concerns of the audience and uses their input to inform future directions.