Thursday, July 08, 2010

Nature of Problem(s), we testers (should) investigate (or solve)

I am puzzled about the nature of problems that software testers, today are juggling with or should be losing their sleep over - more than the problems themselves. As humans, we love simplicity (not sure about about other species - not sure if there is any way to check if they love simplicity too). In every walk of life, we seek and embrace simplicity. Many experts tell the stories about how they have solved mysteries of the world through and give us the tools, techniques to conquer/dominate the world around us. We now have quantum computers even though there is barely anyone who has fully understood Quantum theory. We can make machines, apply our partial knowledge but make a mistake of claiming glory of scientific knowledge saying “this theory is true as I have experimentally verified it and there are practical applications”. The mistake that we make is about “completeness of knowledge” – making a machine based on a theory does not amount to complete understanding of the theory.


Alfred North Whitehead, the famous English mathematician and Philosopher warned us “Seek simplicity and learn to distrust it”. Simplicity is so tempting that it fools our minds and thinking and makes us glued to it. It appears to me that we have not learnt to distrust simplicity at all – worst many even does not know that we should “doubt” it. There is “banana” principle, attributed to Jerry Weinberg. For many things like simplicity, we have only learnt how to start but have horrible lost the trick “when to stop”.

Many in software world believe that software testers are expected to answer questions that, simply lead to “pass” or “fail”, “works” or “does not work”. These questions are like asking “is 4+4 =8? Is this an isosceles triangle? Is the temperature of this body 86 degree centigrade? What is National Stock market Index today?” Engineering, mathematical and scientific questions right?


In our software profession, we love to pose “engineering/scientific” questions that lead to binary answers. After all we are software ENGINEERS and software is ENGINEERED. Hence I started this post with simplicity and our obsession with simple models of complex things. What is the true nature of questions/problems that software testers investigate and solve? Let me make a bold conjecture here. I think they are similar to the questions that we pose in the realm of social sciences. For example, questions like “Is homosexuality linked to or mainly due to some mental illness?” or “Should burqa be banned in public places” or “Is there a God”. These questions are asked in a broad social setup and are linked various material practices of the people in the society, about their culture, history etc. Much in a similar way, software operates in a social context and people are mainly concerned about how software alters their day-today work. The nature of questions/problems requires non-binary, qualitative answers and solutions. The questions about quality are based on emotional, subjective feelings (there is no binary 0 or 1 about software quality).

As Michael Bolton says “decisions about quality are always emotional and political”. How can there be binary mathematics in emotions and politics? That is the point these software engineers and process people miss and insist on quantified definitions of quality and other abstract entities associated with Software. It appears to me that they are constantly conditioned by the business environment to believe so. Coming back to simplicity, software engineers (process/CMMI/Six Sigma) people definitely think that the problems that software testers/engineers solve (investigate- actually) are purely engineering/mathematical in nature. Hence they believe in absolute answers, causality, objectivity and certainty.


I think they are wrong !!!! When do we learn when to stop saying “banananananana..?” While you are reading this and interested in debating about “nature of testing problems”, many budding software testers are happily busy in churning out hundreds and thousands of test cases and blissfully shouting “pass” or “Fail”. Well, you might say do they have options … I am sure they do.

Shrini