> 2. Each writes its results to out/accounts_out_*.dat.

> 3. Python scripts convert fixed-width output to CSV and compute SHA-256 checksums.

> 4. If the hashes match — behavior is proven identical.

Step 3 above introduces the possibility that the python scripts alter the output in such a way that the outputs don't actually match prior to the python.

I'm curious why step 3 is not "If the two outputs match — behavior is proven identical."

> This enduring reliance exists not out of nostalgia, but necessity: COBOL’s reliability, stability, and the prohibitive cost and risk of replacing decades of deeply integrated logic make it one of the most mission-critical technologies ever built.

That sentence struck me as odd. Is COBOL any more "reliable" or "stable" than any other language? I'm no COBOL expert, but when I've looked at it and read about how it works, it seems rather verbose and mundane. That's not unexpected; it was developed in a different era with different sensibilities.

Repo: https://github.com/marcoeg/cobol-modernization-playbook

Would love feedback from people who’ve worked on reverse engineering or legacy transformations at scale.