For: Witztum-Rips-Rosenberg 1994 (Statistical Science 9(3):429–438), Doron Witztum's subsequent work, Harold Gans's follow-up; the significance levels reported. Against: McKay, Bar-Natan, Bar-Hillel, Kalai 1999 ("Solving the Bible Code Puzzle," Statistical Science 14(2)): randomization critique; Bar-Hillel's linguistic/statistical objections; the broader skeptical replication tradition. A Hebrew University committee under Robert Aumann re-ran controlled experiments in 2004 and failed to confirm the codes.
Open questions: whether ELS findings are statistically significant given the search-space; whether the original protocol was sufficiently constrained; whether Hebrew's consonantal structure inflates apparent pattern density.
The edition question sharpens the critique further. WRR's 1994 result was run on the copyrighted Koren (1962) text; the public-domain Leningrad Codex (the base of nearly every open digital Tanakh) differs from Koren in roughly 100 places across the Torah, mostly plene-versus-defective spelling (a vav or yod present or absent). A single added or dropped letter shifts every subsequent equidistant position and can make a long-skip pattern appear or vanish. Even the Talmud concedes the uncertainty (b. Kiddushin 30a: scribes are "no longer expert in defective and plene spellings").
Founder editorial note on how to hold the material: interesting enough to surface, uncertain enough not to rest doctrinal weight on.