A probability model tests whether the first Easter witnesses could have lied

Nathan Rockwood, a philosopher at Brigham Young University in Utah, has built a probability model to ask whether the first claims that Jesus (YHWShA) was seen alive could have been a coordinated lie. His answer is that an invented story shared by 16 to 23 people would probably have collapsed within 17 to 28 years.

The model adapts a method from David Robert Grimes, who tested large conspiracy theories by asking how likely it is that someone, by choice or slip, would give the game away. Rockwood uses a form suited to small groups. He sets the chance of a confession at one percent per witness each year, because the followers talked about their experiences in public. He states that this is an assumption, not a measured rate. He also shrinks the group over time using Roman lifespan estimates from Robert McIver: 74 percent still alive after 10 years, 32 percent after 30.

The head count comes from early Christian texts. The lower figure of 16 is the 11 apostles, James the brother of YHWShA, Paul and three women at the tomb. Rockwood notes that inventors would more likely have chosen male witnesses. The higher figure of 23 adds two more women, the pair on the Emmaus road, Matthias, Joseph and Stephen. He leans on Paul's list in First Corinthians, which scholars widely see as older tradition, and on Paul's meetings with Peter, James and John in Galatians.

The results are stark. With 23 witnesses, the chance that at least one confession surfaces reaches 95 percent in 17 years. With 16 it takes 28 years. Rockwood runs the same test on David Hume's imagined case of Queen Elizabeth I dying and returning. With 200 witnesses at 0.1 percent a year, exposure is expected in 15 years. With 100 witnesses it takes 30.

He also tests Paul's report of more than 500 witnesses, while granting that its historicity is disputed. At 0.1 percent, exposure comes in 6.5 years. At 0.01 percent, it does not come within 30. Small groups can keep a secret. With only six people present at the raising of Jairus's daughter, no failure is expected within 25 years.

The limits are plain. The 95 percent figure is the chance of disclosure under stated assumptions. It is not the chance that YHWShA rose. The model treats confessions as independent, and the witness totals are reconstructed from Christian writings, not verified by outside records. Rockwood adds a second step: reported conversations, meals and touch make honest error unlikely. "And it wasn't deception," he writes.

The paper turns the early lists of witnesses in Paul and the Gospels into a measurable question about sincerity and duration. That is a different question from what the witnesses actually saw.

Compiled from the reports below. Read the originals.