> 1. Both systems run with the same fixed input files (data/accounts.dat, data/txns.dat). > 2. Each writes its results to out/accounts_out_*.dat. > 3. Python

> 1. Both systems run with the same fixed input files (data/accounts.dat, data/txns.dat). > 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."