← Toutes les offres
VG

Moral weight as bits of feedback control

Vasco Grilo🔸

Global
Publié le 17 novembre 2026

Description du poste

This is a crosspost for this post I created with Claude Fable 5.2 and Opus 5.5. SummaryThis post proposes measuring moral weight by how much uncertainty per second a being removes from its world by steering it towards its own goals. • The problem. Rethink Priorities (RP) estimates how much animals matter mostly by scoring traits, such as play or metacognition, as simply present or absent, then multiplying by the probability that the animal is sentient, as described by Fischer et al. (2024) in Weighing Animal Welfare. Illusionism holds that consciousness is a label applied to certain processes, not an extra property that is either there or not, a view defended by Frankish (2022b) and Tomasik (2009). If illusionism is right, there is no yes-or-no fact for this probability to track.• Preferences as control loops. A system has a preference when it runs a feedback loop: it measures something, compares the reading with a goal it stores itself, and acts to close the gap. Thermostats and bacteria do this. Electrons and rocks do not, nor does anything in fundamental physics.• The metric. Moral weight is how many bits per second of uncertainty a system removes from what it controls by measuring it and acting on the reading, in the environments it typically faces. One bit is the information in one yes-or-no answer. Every loop counts, from cellular ones to the brain, but information shared by several loops is counted only once.• Why this metric. It rests on a result from physics, shown by Touchette & Lloyd (2000): each bit a system learns about what it controls can improve its control by at most one bit. It gives the same total whether two systems are counted separately or as one, and whatever units or scales their variables are measured in. It gives electrons, and all fundamental physics, zero. It needs no extra weights for desires about one's own desires, which would be very hard to set in a non-arbitrary way. Finally, it measures every species in the same unit. So, given the