For the starters in this field and to the topic - test automation (or simply automation) is an idea of some computer program "checking" behavior of another computer program (called application under test) and some sense "replicating" what a sufficiently disengaged or brain dead tester would do in the name of testing the application. In order to drive home the idea - early automation tools such as Winrunner introduced the idea of "record and playback". Wow - what a way to simplify a really complex and difficult work of testing a software application.
Thanks to IT consultants and managers - whenever the problem of "speed" showed up in meetings - automation was proposed as potential solution. This has grown to such an insanity that today for almost all problems of software projects - automation is common solution. But, very soon people in IT realized that doing automation requires Money - additional money than you pay to a tester.
Some clever fellow in the consulting company they shouted 'Return on Investment" - from that fateful day - life of tester or someone who supports tester through automation - has never been the same. Since automation requires funding extra to what is spent on testing - execs obviously want their pond of flesh in return.
This leads to popular equation. Without automation if you need 5 testers to do testing for a project/release, with 50% of automation - you would need 50% less people. That is how automation pays back itself. Since software requires repeated testing when changes happen - automation once done can be repeatedly used without paying for human tester. That is how conventional and most popular thinking on automation goes.
While it is not very difficult to reason why and when automation cannot reduce the need of number of testers - very proposition of "automatic" testing and removal of need of some dumb tester staring at screens of automation test run is simply irresistible.
I fought many losing or lost battles in explaining my stakeholders as why automation should not be thought about as means to reduce number of testers or cost of testing. Everytime I lost, I was made to understand - it is just execs made explicit choice not to reason but to continue to insist that if automation can not reduce manpower required for testing - it is useless or at least not worth investing.
I am thinking of refusing to do automation if right expectations are not set with stakeholders - will it work? Will I be given first right of refusal not do automation if right level of awareness exists?
But then - if you are a business leader, IT manager (not someone with deep understanding or appreciation for testing and automation) - you would believe what a consultant or tool vendor would say.
As I close another pessimistic post on automation - I realize - it is tough to be in automation where everyone has opinion (strong one) and I have to force my agenda through.... tough...
But I have not given up --- trying to bring sanity in the mad world of test automation.