Rosetta

What it is

A pipeline that takes a legacy system as it actually exists, the source if there is any, the schema, the logs, and above all the years of data it has produced, and rebuilds the rules it really enforces as something a person can read and a test can run.

Why we built it

Every replacement project stalls in the same place. Nobody knows what the old system does, only what it was meant to do, and the gap between those is two decades of exceptions that are now load-bearing.

What we learned

1. 01

   The data is a better witness than the code. A branch that has not run since 2014 tells you less than a column whose values split into two populations one year.

2. 02

   The most important rules appear in no branch at all. They live in what the data never contains, and absence is much harder to notice than presence.

3. 03

   A recovered rule nobody will confirm is still worth writing down, because the argument it starts is the knowledge transfer.