The War and Peace list

What was done

WRR had used the opening of a Hebrew War and Peace as a control: the first 78,064 letters, the length of Genesis. In their paper it came out flat. Dror Bar-Natan and Brendan McKay asked what would happen if they kept the same text, the same dates, the same 32 rabbis and the same test, and changed only the list of names.

Rabbis are known by many names, and WRR's list used fewer than half of them. Bar-Natan and McKay built a new list from the same pool. They kept 83 names, removed 20 and added 29. Many changes are one valid spelling for another. Oppenheim, for example, can be written two ways. Each change came with a reason, and each was defended as no more doubtful than a choice in the original list. Later they revised it after Witztum's criticism, moving closer to WRR's list.

The result did not fall out by accident, and they never said so. They chose the changes to raise the score in War and Peace. That was the experiment.

The result

In P₄, a rank of 57 in 100 million is stronger than WRR's rank in Genesis. The same list does nothing much in Genesis. Menachem Cohen of Bar-Ilan University, a Bible scholar, compared the two lists and said there was no essential difference between them.

What it shows

It answers one question, and the authors state which one. Was there enough freedom in the experiment that a small number could be reached by using that freedom? Their answer is yes. Before this, defenders of the codes argued that fixing the rules on the first list left the second list no room. This demonstration shows the room was there.

What it does not show

It does not show that WRR did the same. A list tuned on purpose is one thing. WRR's account is that their list was built by a scholar with no such aim. MBBK treat this demonstration as an answer to their first question only, and they turn to the study of variations and the replications for the second. The War and Peace list opens a possibility. Other evidence has to decide whether that possibility is what happened.

It also does not show that the Torah and Tolstoy are the same to every test. Witztum makes this point. He replied that the new list breaks the rules behind the original, and in 2022 he wrote that the list gets 7 in 10 million on the permutation test but only 12 in 10,000 when the words of each verse are shuffled. MBBK reply that the rules were written down a decade after the lists, and that WRR's own list breaks them. Even so, 12 in 10,000 is low for a text nobody considers special.

What this means for your searches

You will never tune a list as hard as this by accident, but the logic scales down. Each spelling you try and each name you drop is a small step of the same kind. The Trial counter exists for this reason. So does Lock and search.

We do not host War and Peace, since this site searches the Tanakh and nothing else. The list and method are public, and the figures on this page come from the authors. MBBK's own summary is that the flexibility alone is enough to reach a similar result in a secular text. You should read their reasoning and Witztum's, and judge. Our critique page lays out both.