THE HORIZON
SECTION 20
ISSUE 001
Hypothesis Markets for Science
Projection: AlphaEvolve allocates search through automated scores; PaperBench decomposes uncertain research replication into evaluable work. Scientists and agents might similarly allocate experimental budgets across competing hypotheses using forecasts and evidence updates. The market remains speculative until calibrated allocations outperform conventional review on prospective information gain without gaming the evaluator.
Why this idea is here
What the evidence establishes.
Google DeepMind documents an evolutionary coding agent that proposes programs and scores them with automated evaluators; PaperBench evaluates end-to-end AI research replication and reports large remaining headroom on the tested agents. These are source-backed premises for this projection; they do not by themselves prove broad adoption or the eventual outcome.
Source ledger
Read the sources.
- S01AlphaEvolve
official research release / published 2025-05-14 / retrieved 2026-07-09
- S02PaperBench
official benchmark release / published 2025-04-02 / retrieved 2026-07-09