Temporal Leakage in Search-Engine Date-Filtered Web Retrieval: A Retrospective Forecasting Case Study

TL;DR

This study reveals significant temporal leakage in search-engine date filters, leading to unreliable retrospective evaluations of search-augmented forecasters, and suggests more robust safeguards are needed.

Contribution

The paper identifies and characterizes mechanisms of temporal leakage in search-engine date filters and demonstrates their impact on forecasting accuracy.

Findings

At least 71% of questions on Google contain post-cutoff leakage.

Forecasting with leaky documents inflates accuracy metrics.

Leakage mechanisms include updated articles and unreliable metadata.

Abstract

Search-engine date filters are widely used to enforce pre-cutoff retrieval in retrospective evaluations of search-augmented forecasters. We show this approach is unreliable across two major search engines: auditing Google Search's before: filter and DuckDuckGo's date-range filter, we find that at least one retrieved page contains major post-cutoff leakage for 71% of questions on Google and 81% on DuckDuckGo, and the answer is directly revealed for 41% and 55%, respectively. Using gpt-oss-120b to forecast with these leaky documents, we demonstrate inflated prediction accuracy (Brier score 0.10 vs. 0.24 with leak-free documents). We characterize recurring leakage mechanisms, including updated articles, related-content modules, unreliable metadata, and absence-based signals, and argue that date-restricted search on these engines is insufficient for credible retrospective evaluation. We recommend stronger retrieval safeguards or evaluation on frozen, time-stamped web snapshots.