Traceby EphemerisX

Trace the record. Prove the chain.

Each challenge gives you a question about a real regulatory or safety case and a docket of evidence. Build the reasoning chain that answers it. The judge checks every step against the record.

Evidence is paraphrased from the public record of each case and labeled with its source. Your email identifies your account and is never displayed.

Current challenges

Each one is a real case from the public record of an NTSB investigation or an FDA drug review.

  1. NTSB investigation · 2018 Tesla Model X crash, US-101 at SR-85, Mountain View, CA · HSR-19/01, HAR-20/01

    Mountain View: the barrier that was already broken

    The Model X struck the gore-point barrier at about 71 mph and the driver died. Eleven days earlier, another car hit the same barrier faster and that driver walked away. Why was the second crash the fatal one?

  2. FDA drug approval · Aducanumab (Biogen/Eisai) for Alzheimer's disease · BLA 761178, 2019–2021

    Aduhelm: from failed trials to approval

    Both Phase 3 trials were stopped for futility, and the advisory committee later found the evidence insufficient. How did aducanumab get approved anyway?

  3. FDA drug approval · Duchenne cardiomyopathy cell therapy · 2025–2026

    Deramiocel: the vote that didn't end it

    An FDA advisory committee voted 9–3 that the evidence did not show effectiveness. Three weeks later the agency agreed to keep reviewing and moved its decision to November. What kept the application alive?

  4. NTSB investigation · Regional jet and Army helicopter, Washington DC · Jan 2025 – Feb 2026

    DCA midair collision: why the Board blamed the airspace

    The Board's probable cause names the FAA's airspace design and safety-management failures, not only the crews. Why did the Board reach past the crews to the airspace itself?