Turn failed experiments
into your next idea.
The Failed Experiments Database is a platform for researchers to share unsuccessful experiments and disproven hypotheses, preserving findings often overlooked in scientific literature. Each result is characterized by their field, revealing connections across disciplines and mapping connections from seemingly unrelated failures.
How it works
Log an unsuccessful experiment
Researchers in any discipline submit an unsuccessful experiment, methods that failed, or incomplete ideas.
Find the faulty assumption
Each entry is sorted by its field and type of failure: whether it was built upon a faulty assumption or claimed more than what its data could actually support. AI pinpoints exactly where it went wrong.
Turn it into an idea
Failures in different fields may share unexpected connections. Our database reveals these similarities, helping researchers develop new ideas and hypotheses.
Check before you repeat it
Before starting an experiment, check whether others have tried something similar, what went wrong, and what could be learned from their findings.
What gets captured
Hypothesis
Combining a KRAS inhibitor with a SHP2 inhibitor would delay resistance in G12D-mutant PDAC organoids.
What failed
No delay in resistance vs. monotherapy across 3 organoid lines at 12 weeks.
AI inferred reasoning
Bypass signaling through RTK upregulation, independent of SHP2 pathway — inhibitor combination targeted the wrong resistance mechanism.
Broken assumption
Resistance to KRAS inhibition runs through SHP2.