Why do New things break Old things?
For one of my clients a certain feature was broken that had been working fine for months. My initial reaction was that the front-end changed. Upon testing I tracked it down to a session variable no longer being set during login. It turns out in the conversion from username/password to SSO a number of session variables were no longer being set.
For another situation a database job that had been running for some time on the first of the month apparently stopped working. Upon investigation, the job had been completely removed from the system. This was done when another client’s data was moved onto the database server.
In both cases New things, that are good ideas, were implemented in such a slipshod manner that existing, functional aspects were trashed.
Thinking about this some it seems like this is a very common occurrence: the developer of the New thing has the priority of getting the New thing done. Old things take a distant second in anyones mind - developer, pm, client - anyone.
I know you can use a regression test suite to verify you didn’t break something, but this is an afterthought. It is not in the developer’s frame of mind to determine what else might be affected by the New thing. From my own experience if you happen to think of side effects you get looked at with squinted eye, not a team player, naysayer, problem child.
Further, fixing the broken Old thing after the fact is also looked at as a good thing. “Yes, we fixed the Old thing. It turns out the New thing broke it.” Everyone is happy that the Old thing got fixed. No one asks why this wasn’t discovered beforehand.
You can see this on software developer web sites. Everyone can develop something New for you. Or even more interesting they can debug your code or evaluate your system performance. No one is advertising they will maintain your current system. System maintenance is not valued.
Maybe this is just a trickle-down of the quarterly hockey stick affect working into how things get coded. Imagine that, developers modeling their behavior after sales people.