The so-called "natural" sort makes sense for version numbers and enumeration (without zero-padding) but I'm more often dealing with file names with a datetime (like in the article), a hexadecimal hash, or just randomized string of characters that includes numbers. In those cases "natural" sort makes it harder to find the file you're looking for.
Even when files are enumerated it's pretty rare to have more than 9 parts and no zero-padding, whereas there are almost always multiple consecutive digits in the use cases for which "natural" sort is not a good fit for. It just feels like a bad default, at least for a programmer's workload.
AFAICT, natural sort shouldn't ever make datetimes harder to find, unless they are formatted inconsistently, as in the author's case. Suppose one camera wrote dates as 20250928 and another as 2025-09-28. ASCIIbetical sort would do nothing to help here.
Natural sort can even improve things over ASCII sort, for instance if someone is stuck with a format like "28/9/2025" or "September 2 2025"
Even when files are enumerated it's pretty rare to have more than 9 parts and no zero-padding, whereas there are almost always multiple consecutive digits in the use cases for which "natural" sort is not a good fit for. It just feels like a bad default, at least for a programmer's workload.