From Plato's Republic to the question of who an algorithm serves.
The Republic begins as an argument about what justice is and whether it pays to be just. The question has never closed: justice as harmony, as desert, as fairness, as non-domination, as the mandate of heaven withdrawn from a failing ruler. Each tradition ties justice to a picture of the good order, human and cosmic. Today the question arrives in a new form — not only what justice is, but who the systems that increasingly allocate opportunity, risk, and attention are actually serving.
The thinkers this collection draws together. Full profiles live under Thinkers.
Plato
Justice as harmony in the soul and the city.
Aristotle
Distributive and corrective justice; giving each their due.
Mozi
Impartial care as the measure of a just order.
Rawls
Justice as fairness, tested behind a veil of ignorance.
Anchor texts, in translation, with the original where it matters. To be added.
Candidates include Plato's Republic, Nicomachean Ethics V, the Mozi, and Rawls on fairness.
Where the traditions agree, where they part, and why. In progress.
Justice as harmony, desert, or fairness — and which conception a data-driven system tends to encode by default.
Machine learning makes justice concrete and uncomfortable. A model learns from training data, and so inherits the inequities recorded in it; an aligned system encodes somebody's conception of fairness, chosen by whoever shaped it. Beyond present harms of bias and access lies the longer worry of concentration of power. The ancient question — justice for whom, on whose authority — turns out to be the right one to ask of any system that allocates.
Terms used here, defined plainly: Training Data · Alignment · Existential Risk
Annotated bibliography. To be added.
Selected scholarship and translations will be listed here, each with a short note on why it earns its place.