Pre-reading: This post builds on the two previous instalments, which are about what it means to not-know about the relationship between actions and outcomes and what it means to not-know about actions and about outcomes.
The usual caveat: Work in progress; may be only half-baked 🥖.
“Value” is an inherently subjective assessment of how much an outcome is worth to someone. This means that ultimately definitions of value end up being circular: things that have value are valuable, etc.
Though we recognise that some outcomes have intrinsic value (e.g., “ensuring that human freedom is protected” or “ensuring that a human life is saved”) we seem to only be able to make sense of the value of outcomes in terms of proxies that are themselves valuable.
Fred values his salary at BankyBank enough to trade 60 of his waking hours each week to get it (valuable proxy: ⏳). BigOilCo values fracking some oil out of that patch of shale enough to sacrifice local clean groundwater (valuable proxy: 🌎). The most obviously and legibly valuable proxy is 💰: OpenAI was willing to pay $100 million to achieve the outcome of training GPT-4.
When we act with intention (vs acting accidentally or instinctively), it is in the context of a web of assumptions about what actions and outcomes are possible, how those actions are connected to outcomes, and which of those outcomes are desirable — that is to say, which outcomes we value and how much we value them. Much of the time, we assume that the value of these outcomes is a) stable, b) known in advance, and c) agreed among stakeholders.
Recognising the ways in which the value of outcomes could be unstable, incompletely known in advance, and/or heterogeneous among stakeholders certainly messes with existing theories of action — which maybe were unrealistically simplistic.
Below, five sources of not-knowing about value.
Different people often value the same outcome X differently in the present. This can happen in three ways.
How will the value of X be shaped by future events which, by definition, cannot be known for certain in advance? When we consider how outcomes will be valued in the future, two additional sources of not-knowing about value emerge.
Not-knowing about value messes with otherwise clear principles for deciding on action. However, understanding the different sources of not-knowing about value opens up paths for thinking about new ways value can be conceptualised and new constituencies (present and future) whose assessments of value should be considered.1