Saturday, March 16, 2019

Guilty Pleasure


In Episode 10 of The Guilty Tester podcast Dave Duke took questions from the Twitterverse. The first one was this:
Think of a piece of work you’ve done recently, what did you actually do to test it and looking back, what surprised you?

In his answer, Dave told a story about a recent series of bugs, his involvement in sharing knowledge about them, and how that prompted an unexpected fix. It's a nice story, but a different reading of the original question interests me too: how do we  test our own work? As he was asking for more questions, I suggested it to him, and he tweeted back:
That's an interesting question. Back at you though. Do you test your own work differently to how you test others? Do you practice what you preach? Is it realistic to expect people to test their own work the same as they test others? Is there an advantage though.

Juicy! I didn't want to spend loads of time trying to craft an answer into a sequence of 240-character soundbites but I liked the idea of thinking it through, so I gave myself an hour to write a response. This is it.

I like how you loaded your questions. Take "differently to" and "the same as." When I hear those terms, I want to ask you which dimensions you have in mind. On a very strict interpretation, an identity relationship, I'd say that testing my work and testing others' work is not the same. But then I'd also suggest that two testers given identical applications to test in identical environments would not test the same either. In fact, my instinct is that the same tester given the same application in the same environment on two different occasions would be unlikely to test the same.

But, without asking for your clarification, I can answer the question this way: I do test the same way, because I always test the same way. I spent time working out what testing is for me (the pursuit of relevant incongruity) and so, at a strategic level, if I'm testing I'm doing the same thing.  However I also never test the same way, because I strive to take the context into account when I'm testing. At a tactical level, the context for two pieces of work is rarely truly identical in all important dimensions, not least because I am in the context and I am not reliably identical over time.

Just semantics? Leave it aside then. What about my work and others' work? As a test manager, my work doesn't overlap much with the people whose work I test. Whose work do I test? I'd say it includes developers, business development staff, testers, technical authors, technical support staff, and others. What overlaps with what I do? Some examples would be when I code (tools, test automation), or test (features, ideas), or write (usually raw doc, process doc). These would be places where I might have an opportunity to test my work in the same way as I'd test others'.

What might stop me from doing that? Here's a few thoughts: I would expect that I already understand the motivation for the work, and for the approach I took with it, so I might not spend time exploring it. Likewise, I might not look again at the problem I'm trying to solve with this work and wonder whether other approaches could be better; in particular I won't know of approaches that the implementer didn't know about. I might not want to criticise the approach taken, because I know it'll mean more work for me. I will also likely be subject to the same biases I had when building the solution.

Given that, what kinds of approaches do I use to test my own work? I use writing a great deal to provide a way to externalise thoughts in a way that detaches them from me. I then find that I can challenge them in a way that's less easy when they reside in my head. It doesn't have to be prose (although I find that helps me to nail down what I think), a mind map or a table or a sketch can all help. I also seek review; I might ask for review of ideas and proposals before implementing something, or I might run some kind of retrospective afterwards, or both.

Over time I have found that working incrementally, and finding ways to cheaply check that I'm on track at each step, helps me to produce the kind of work I want to produce and interleaves the implementation and the testing of it. I like to monitor and review myself. I am a believer in open notebook testing, so I like to commit my code to a public repository, I post my test notes to our company wiki. I may not ask for review of these things all the time, but they are available for anyone who wants to look and I am open to suggestions on them at any time. I make myself available for questions, on any topic, and I try to treat all questions with respect and humility, in order that people are not discouraged from asking them.

Finally, do I practice what I preach? Unlike pretty much everywhere else in this answer, I can give you a resounding YES here! It's a point of honour for me not to be a hypocrite. I have high standards and a desire to improve tempered with a pragmatic edge and, particularly as a manager, I feel it's on me to measure up and demonstrate the kinds of behaviours I try to encourage in my team, and the value that it brings. I also don't want to lose the enjoyment, credibility, and satisfaction that I get from being a practitioner.

You might ask how I tested this essay. I gave myself 20 minutes to collect and categorise thoughts (a kind of  mini fieldstone approach) and then 30 minutes to write each of those categories into a paragraph. At each stage I was testing that the thoughts were coherent and that I believed them. For example, I asked myself how happy I was to commit to "I always test the same way" and looked for reasons why I might not want to. Here's one I considered: perhaps I'll change my definition of what testing is.

I then spent 10 minutes organising the paragraphs, again looking for coherence, repetition of words or concepts, flow, and, finally, mindful of my time box, I pasted it into Blogger from my text editor and formatted it. I've allowed myself a few extra minutes to find a picture and relevant links and do a final proof read. Apart from typos I won't change anything now.
Image: https://flic.kr/p/m2s7Q

No comments:

Post a Comment