Mapping contradictions in scientific research. What does science actually agree on? Building in public.
Joined May 2026
- Tweets8
- Following4
- Followers3
- Likes4
Pinned Tweet
Six months ago I was doing a literature review and
found two papers making completely opposite claims
about the same phenomenon.
Both were highly cited. Both seemed rigorous.
I spent 4 hours figuring out why they disagreed.
I'm building the tool I wished existed.
Waitlist open!
Causalith retweeted
Working on the verdict page in @Causalith now.
Enter a claim →
we retrieve relevant papers →
surface supporting + contradictory findings →
and trace the claim back to actual source evidence.
Trying to make research verification less of a black box.
#buildinpublic #indiehackers
Causalith retweeted
This week in @Causalith :
Upload papers → extract claims → preserve source evidence → inspect every decision.
Each candidate sentence now gets a status: extracted, background, method detail, not-a-claim, or retry.
Less black box. More audit trail.
#buildinpublic #indiehackers
VEB score went from 53% to 71% this week
The bug: we were using TF-IDF similarity to match
scientific claims. 'Previous best results' and
'existing best results' scored 0.25 similarity.
Switched to semantic embeddings. Score jumped 8x.
Building @causalith — waitlist in bio.