* Automated testing. While implementing an algorithm, also make tests. Years later, if the program begins crashing in a test, then it can be fixed. Without test

* Automated testing. While implementing an algorithm, also make tests. Years later, if the program begins crashing in a test, then it can be fixed. Without tests, the crash may happen sporadic in production and be hard to fix. * Metrics on a dashboard (Grafana+Prometheus). See whats going on in a program. How many times is there a cache hit or miss. What are the most frequent occurring errors.