Erin and I met yesterday on the topic of reducing test case bloat. While on her team Automated cases have not yet grown to the level that they are causing bloat, I'm worried about it as a potential problem. Automated cases are no different than manual cases in the fact that they take resources and time to maintain. It may be free to run them each time once it is set up, but fixing them when they break and going through false error reports is a huge time sink and that adds up to real money a company is spending.

For that reason, I believe we should set up a better process for reducing test case bloat for automated cases. What strategy does your team use? Are you using the addition method like we are? Just keep adding more and more cases, never remove any? How long have you done it for? Is it still working, or is it causing madness and wasted time/effort?

I think the idea that we should always add more automated cases is based in false assumption and we should only bother to run and maintain the most important automtaed cases. What would you use to pick the optimum set of automated cases to run?