My job requires me to write programs that automate the testing of certain features that have been implemented or certain bugs that have been fixed in the network chips the company makes. These tests are designed to detect problems as they arise, and bugs if they ever start to occur again. Sometimes a test only needs to test one small thing, and it’s easy to only that one feature and get the job done. So it’s easy to write really quick test, get the jobs done, add the results to your weekly reports, seem like a really productive employee and eventually get a promotion. In order to really verify that the network is working properly, however, the test requires a lot more work. It should test all the edge cases looking for obscure bugs, the style should be good, and it should have cleanup code so the network doesn’t fail if the test stops half way through. Creating the most comprehensive test possible does not always have the highest outcome for the programmer, but it is the right thing to do, in a deontological sense, to ensure that the most obscure bugs aren’t found by customers when they need a working system most.