Quick takeaways from Griffes’ Doing good while clueless
By Jim Buhler @ 2025-08-07T09:59 (+5)
Written in August 2024. I’ve only made some minor style edits before posting here.
Doing good while clueless is a 2018 EA Forum post written by Milan Griffes where he recommends some concrete interventions we should believe are good despite some cluelessness concerns, thanks to the assumption that there are “five attributes” (namely intent, coordination, wisdom, capability, and predictive power) that predictably positively influence the long-term future.
Griffes’ “steering capacity” argument
Griffes first describes the problem of cluelessness with an interstellar travel analogy:
Consider a spacecraft, journeying out into space. The occupants of the craft are searching for a star system to settle. Promising destination systems are all very far away, and the voyagers don’t have a complete map of how to get to any of them. Indeed, they know very little about the space they will travel through.
He then suggests: (I removed a footnote)
"Steering capacity" can be broken down into the following five attributes:
- The voyagers must have a clear idea of what they are looking for. (Intent)
- The voyagers must be able to reach agreement about where to go. (Coordination)
- The voyagers must be discerning enough to identify promising systems as promising, when they encounter them. Similarly, they must be discerning enough to accurately identify threats & obstacles. (Wisdom)
- Their craft must be powerful enough to reach the destinations they choose. (Capability)
- Because the voyagers travel through unmapped territory, they must be able to see far enough ahead to avoid obstacles they encounter. (Predictive power)
Considering these five attributes, he goes on to conclude that we should believe the following concrete interventions are good (i.e., that we’re not clueless about their sign):
- Better understanding what matters
- Improving governance
- Improving foresight
- Reducing existential risk
- Capacity building.
Important limitations of this work (in my opinion)
In his space journey analogy, Griffes doesn’t explain why we are conveniently assuming the map is merely incomplete rather than nonexistent and hardly possible to create (beyond a certain point at least), i.e., why we are assuming we have any single clue where we might be going.
To the extent that the analogy doesn’t break, the problem that has been raised by cluelessness arguments (see Lenman 2000; Greaves 2016; Yim 2019; Thorstad & Mogensen 2020; Mogensen 2021; Roussos 2021; Williamson 2022; Tarsney 2023; Schwitzgebel 2024; Thorstad 2024; 2024, §6.4; Lundgren & Kudlek 2024; Tarsney et al. 2024, §3) is that it is hard to justify believing we can know or learn anything beyond what’s beyond a certain point on the map. It follows that we should question whether the five attributes Griffes identifies are even good attributes to begin with. If we can’t predict where our ship might be going, how is, e.g., making sure the people on board are “well-intentioned” of any help? How can we rationally believe it’s more plausible than the exact opposite, where having “well-intentioned” people on board actually makes things worse due to some sign-flipping crucial considerations we are neglecting (or totally unaware of)[1]? While providing an interesting analogy, Griffes implicitly answers these questions without providing justifications, which leaves the key problems raised in the cluelessness literature unaddressed.