Automation has gained traction in recent years in the testing community because the idea of automating tests has been sold off as a solution to many of the common problems of traditional testing. And the advertisements worked, as many software businesses now rely heavily on automated checks to monitor and check known system bugs, even performance and security tests. It’s not surprising anymore that automation has been accepted by the general public as one of the cornerstones of effective testing today, especially since releases to production for most businesses has now become more frequent and fast-paced. I use such tools at work myself, because they complement the exploratory testing that I do.
James Bach and Michael Bolton, two of the finest software testers known to many in the industry, however reminds us, in their whitepaper titled ‘A Context-Driven Approach to Automation in Testing‘, to be vigilant about the use of these tools, as well as the practice of the terms ‘automation‘ and (especially) ‘test automation‘ when speaking about the testing work that we do, because they can imply dangerous connotations about what testing really is. They remind us that testing is so much more than automation, much more than using tools, and that to be a remarkable software tester we need to keep thinking and improving our craft, including the ways we perform and explain testing to others.
Some takeaway quotes from the whitepaper:
- If you need good testing, then good tool support will be part of the picture, and that means you must learn how and why we can go wrong with tools.
- The word ‘automation’ is misleading. We cannot automate users. We automate some actions they perform, but users do so much more than that. Output checking is interesting and can be automated, but testers and tools do so much more than that. Although certain user and tester actions can be simulated, users and testers themselves cannot be replicated in software. Failure to understand this simple truth will trivialize testing, and will allow many bugs to escape our notice.
- To test is to seek the true status of a product, which in complex products is hidden from casual view. Tester’s do this to discover trouble. A tester, working with limited resources must sniff out the trouble before it’s too late. This requires careful attention to subtle clues in the behavior of a product within a rapid and ongoing learning process. Testers engage in sensemaking, critical thinking, and experimentation, none of which can be done by mechanical means.
- Much of what informs a human tester’s behavior is tacit knowledge.
- Everyone knows programming cannot be automated. Although many early programming languages were called ‘autocodes’ and early computers were called ‘autocoders,’ that way of speaking peaked around 1965. The term ‘compiler’ became far more popular. In other words, when software started coding, they changed the name of that activity to compiling, assembling, or interpreting. That way the programmer is someone who always sits on top of all the technology and no manager is saying ‘when can we automate all this programming?’
- The common terms ‘manual testers’ and ‘automated testers’ to distinguish testers are misleading, because all competent testers use tools.
- Some testers also make tools – writing code and creating utilities and instruments that aid in testing. We suggest calling such technical testers ‘toolsmiths.’
- We emphasize experimentation because good tests are literally experiments in the scientific sense of the word. What scientists mean by experiment is precisely what we mean by test. Testing is necessarily a process of incremental, speculative, self-directed search.
- Although routine output-checking is part of our work, we continually re-focus in non-routine, novel observations. Our attitude must be one of seeking to find trouble, not verifying the absence of trouble – otherwise we will test in shallow ways and blind ourselves to the true nature of the product.
- Read the situation around you, discover the factors that matter, generate options, weigh your options and build a defensible case for choosing a particular option over all others. Then put that option to practice and take responsibility for what happens next. Learn and get better.
- Testing is necessarily a human process. Only humans can learn. Only humans can determine value. Value is a social judgment, and different people value things differently.
- Call them test tools, not ‘test automation.’