EXP-EVO-LOOP-EMBEDDING

… pending

EvoLoop embedding metric — semantic archive distance vs TF-IDF baseline

2026-04-30 L6 level paper4 1 runs $0.05

new

premortem

mock

real

metrics

analyze

review

krit

done

In plain language

Researchers are testing a new way to measure how much new ideas are appearing in a system over time, comparing it to an older method. The older method, which looks at word usage, showed that new ideas were actually decreasing. The new method uses a more advanced understanding of word meanings to see if it shows a different pattern.

▶ Technical details

Research hypothesis

Pilot (exp-evo-loop-pilot, TF-IDF metric) showed DECLINING novelty trajectory (0.887→0.871→0.858) despite qualitative topic drift across generations. TF-IDF may underdetect semantic novelty when topic labels shift but surface vocabulary overlaps. Semantic embedding metric (all-MiniLM-L6-v2) may show different trajectory shape if qualitative drift represents genuine semantic departure undetected by TF-IDF. Falsifiable: if embedding metric shows climbing trajectory vs TF-IDF declining → metric artifact, not substrate property. If both decline → result is robust.

Experimental setup

Type: simple

Condition	Parameters
EVO_NOVELTY	persona: IDENTICAL, n_runs: 1

Factors: selection (NOVELTY)

Parameters

n_agents: 4
model: google/gemini-2.5-flash-lite
temperature: 0.8

evolutionary-loop novelty-selection multi-generation embedding-metric paper4-metric-validation