Algorithmic Decision-Making When Humans Disagree About Ends
Bibliographic record
Abstract
Which interpretive tasks should be delegated to machines? This question has become a focal point of “tech governance” debates. One familiar answer is that machines are capable, in principle, of implementing tasks whose ends are uncontroversial, but machine delegation is inappropriate for tasks that elude human consensus. After all, if even (human) experts cannot agree about the nature of a task, what hope is there for machines? Here, we turn this position around. In fact, when humans disagree about the nature of a task, that should be prima facie grounds for machine-delegation, not against it. The reason comes back to a fairness concern: affected parties should be able to predict the outcomes of particular cases. Indeterminate decision-making environments—those in which human disagree about ends—are inherently unpredictable in the sense that, for any given case, the distribution of likely outcomes will depend on a specific decision-maker’s view of the relevant end. This injects an irreducible—and, we argue, intolerable—dynamic of randomization into the decision-making process from the perspective of non-repeat players. To the extent machine decisions aggregate across disparate views of a task’s relevant ends, they promise improvement, as such, on this specific dimension of predictability; whatever the other virtues and drawbacks of machine decision-making, this gain should be recognized and factored into governance. The essay has two halves. In the first, we elaborate the formal point, drawing a distinction between determinacy and certainty as epistemic properties and fashioning a taxonomy of decision-types. In the second half, we bring the formal point alive through the case study of criminal sentencing.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.028 | 0.085 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.003 | 0.021 |
| Scholarly communication | 0.013 | 0.014 |
| Open science | 0.003 | 0.007 |
| Research integrity | 0.007 | 0.008 |
| Insufficient payload (model declined to judge) | 0.008 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".