Area Statistician Hails Thrice-Scheduled NASA Fortran Talk As First Properly Replicated Study Of Career

ANN ARBOR, MI—Calling it "the closest thing to a replication protocol I have ever seen printed on a booth schedule," reproducibility researcher Delphine Okafor-Marsh, 47, confirmed Sunday that she had cleared her entire Tuesday morning to attend every listed session of a single NASA talk.

NASA's Supercomputing 2026 tech demo schedule, published November 6, lists every presentation three times in consecutive half-hour slots, including Tom Clune's "Type-safe Templates in Fortran 2028" at 10:00, 10:30 and 11:00 AM on Tuesday, November 17.

Okafor-Marsh has filed a 14-page preregistration with her employer, the Center for Conference Reproducibility. Its primary outcome is whether the 10:30 slot reproduces the 10:00 slot. Its secondary outcome is whether she can still feel her legs by 11:00. The schedule does not say whether the three sessions are identical or different, a gap Okafor-Marsh described as "a double-blind design, provided I don't read anything else."

Her methods section specifies a different seat for each session to control for sightline effects, arrival at exactly 9:58 each time, and a standardized 240 ml of coffee per slot. Colleague Prof. Henrik Asante, 58, warned that three consecutive sessions by the same presenter could not be considered independent samples. "Henrik has never been offered n=3 at a conference," Okafor-Marsh said. "He doesn't know what it does to a person."

After noticing that the talk concerns a Fortran 2028 standard at a 2026 conference, she reclassified the project as a prospective cohort study and booked a follow-up for 2028 to check whether the future holds up under scrutiny. "Most predictions in my field don't come with a version number," she said.

Okafor-Marsh conceded that her favorite part of the study so far has been the schedule itself, which NASA describes as a 2-minute read. She called this "a rare instance of an institution reporting its own effect size up front."

At press time, Okafor-Marsh had read the schedule 41 times, logged a mean read time of 1 minute 52 seconds with a standard deviation of four seconds, and declared it the most reproducible document in her field.