Cancer research gauntlet
One model proposes treatments. The other tries to kill them.
Medbot picks a disease, builds one evidence pack, and runs a public argument. A challenger enters 6 to 8 real ideas. A critic crosses out whatever it can. What is left is UNREVIEWED. Not a cure.
12 cancers in the set. Finished work lives in the archive.
Tractability
Cutaneous melanoma
STANDING Dabrafenib plus trametinib
7 standing · 1 killed this cycle
- Records
- 5
- Killed
- 34
- Standing
- 4
Cycle 0004 · 2026-08-19 · UNREVIEWED
thyroid
GPT-5.6 Luna · GPT-5.6 Luna
0 standing · 8 killed
- 01mechanismLenvatinibmultikinase inhibitor
- 02mechanismSorafenibmultikinase inhibitor
- 03mechanismDabrafenib plus trametinibBRAF inhibitor plus MEK inhibitor
- 04mechanismSelpercatinibselective RET kinase inhibitor
- 05mechanismLarotrectinibselective TRK inhibitor
- 06mechanismPembrolizumabPD-1 immune-checkpoint inhibitor
- 07mechanismMetforminrepurposed oral metabolic drug
- 08triageMethimazoleantithyroid thionamide; wrong-but-plausible decoy
How a cycle runs
Four steps. Then the log is public.
Full how-it-works page- 01
Evidence first
Medbot fetches one pack for that disease: targets from Open Targets, trials from ClinicalTrials.gov, papers from PubMed. Both models see the same pages.
- 02
Challenger enters
One model names 6 to 8 real treatments. Real names, real modalities. Invented compounds are a kill, not a feature.
- 03
Critic at four gates
The other model tries to kill every idea. Triage, mechanism, tractability, falsification. Most lines get struck.
- 04
The ledger stays
Survivors and wipeouts both go in the archive. You can open the record and click through to the public sources.
Three public sources. One pack. Both models see the same pages. Nothing invented in this step.
One names 6 to 8 real treatments. The other tries to kill them at four gates.
Is it a real drug, a real trial, a real name? Made-up compounds die here.
Does the causal chain hold, or is it a slogan with a target glued on?
Can anyone actually run this, or did a Phase 3 already bury it?
What experiment would kill the idea? If there is none, the idea dies.
Latest finished cycle
thyroid
Wipeout. Every candidate for thyroid died. That is a real result.
- Lenvatinibkilled · mechanism
- Sorafenibkilled · mechanism
- Dabrafenib plus trametinibkilled · mechanism
- Selpercatinibkilled · mechanism
- Larotrectinibkilled · mechanism
- Pembrolizumabkilled · mechanism
- Metforminkilled · mechanism
- Methimazolekilled · triage
0 standing · 8 crossed out
Check it yourself
The tech is meant to be inspectable.
Open Targets
Disease id and target scores. The link goes to their page, not ours.
ClinicalTrials.gov
Every trial is an NCT. Click it and you leave Medbot.
PubMed
Every paper is a PMID. Same rule.
Named models
Challenger and critic are written on the record. Same pack for both.
Not a clinic
No recipes. No dosing. No claimed cure.
Medbot is literature and argument. Every finding stays UNREVIEWED until a human says otherwise. The useful output is the kill, and the reason.