---
title: "PHI403 Lecture 18 — Risky Predictions"
type: source
tags: [philosophy-of-science, prediction, external-validity, overfitting, tendency, popper]
authors: [Anjum, Rani Lill; Rocca, Elena]
year: 2023
venue: "PHI403 Causation in Science, NMBU"
kind: handout
raw_path: "raw/Philosophy of Science/PHI302 18 Risky Predictions.pdf"
created: 2026-04-20
updated: 2026-04-20
key_claims:
  - Any causal claim from a data set goes beyond the data; without a theory, there is no ground for repeatability, generalisability, or external validity.
  - Closed-system / ideal-condition predictions are reliable (lunar eclipses) but don't transfer to open real-world systems.
  - Adding more causal factors to the model trades internal for external validity — overfitting.
  - Tendency predictions ("C disposes toward E") are fallible but useful; they acknowledge the dispositional modality instead of pretending certainty.
---

# PHI403 Lecture 18 — Risky Predictions

On the **epistemology of prediction**. Predictions are always fallible: Hume's problem of induction makes them logically invalid; Popper's falsificationism accepts them only provisionally.

Even with a complete data set, a causal claim goes *beyond* the data — this is what [[concepts/rct-limitations|external validity]] requires but data alone cannot supply. Without a theory linking data points, we have no reason to assume a within-sample pattern transfers to new samples or to the population.

**Closed systems** — labs, models, ideal conditions — allow near-100 % predictions (lunar eclipses, pressure in a gas container). But stepping outside the closed system into real-world complexity, predictions become fallible again.

A central warning: **adding more factors to a model trades internal for external validity** — classic *overfitting*. More realistic-looking models do not automatically produce better predictions.

The lecture's positive proposal: **tendency predictions**. Instead of saying what *will*, *would*, or *must* happen (which fails if any factor is missing), predict what *tends* to happen given current knowledge. Tendencies are not guaranteed, but they are more than accidental — they reflect [[concepts/dispositionalism|dispositions]] that the cause genuinely has.

This lecture is the most directly relevant one for **ML prediction and [[concepts/predictive-process-monitoring|predictive process monitoring]]**: overfitting, external validity, and the framing of predictions as tendency claims all map onto standard ML concerns.

## Connections
Back-link: [[sources/2023-anjum-rocca-phi403-causation-in-science]]. Concepts: [[concepts/dispositionalism]] · [[concepts/rct-limitations]] · [[concepts/probabilistic-causation]] · [[concepts/predictive-process-monitoring]].