Categories
Blog posts in English

Immanuel Kant and the Hallucinating Tester

Quality is an illusion. That may seem like a bold statement, but there is a deeper truth to it that I will discuss in this blog. I’ll also discuss how we can approach the real.
We can think of testers as doctors, scientists, or researchers whose job is to research, explore, or examine some software, gather, analyze, and communicate factual knowledge based on observations.
But science teaches us that when we research and observe things, including software, what we “see” is not reality. At TED 2017, University of Sussex neuroscience professor Anil Seth called what we see “hallucinations”.
This gives the hopefully scientific tester some severe epistemological challenges: As she is a person, and is hallucinating, how can she (or we) trust her observations?
The problem for her is that the images that she experiences as real are a synthesis, an intuitive product of her observances based on a minimal amount of sensory data. The critical mindset is important in testing but doesn’t help by itself.
Fortunately philosophy has a solution for her (and us). Before I explain it, let me share a daily life story about intuitive illusions and assumptions.
 

Walking on Black Ice

I was out walking my poodle Terry a few days ago. A car came against us, but as we were on the sidewalk and the car on the road, the situation was intuitively safe.
Unfortunately, my intuition turned out to be wrong as only a moment later my foot slipped on the sidewalk and I realized that the wet road was not wet; both the road and the sidewalk were covered in black ice.
When another car approached I was aware of the danger, and made sure to keep myself and my dog safe.
There could be a moral in this story about always being cautious about cars and roads, but it might end up in over-cautiousness of the type that grandmothers sometimes impose on their grandchildren.
Instead I consider it a reminder that we don’t see things as they are: The road was wet until my foot slipped and I realized it was icy.
Already the Stoic philosophers in Rome 2000 years ago had figured this out.
 

Immanuel Kant’s Model of Mind

In 1781 the German philosopher Immanuel Kant published his mammoth work Critique of Pure Reason in which a key concept is the transcendental, which can be thought of as a bridge between the real and the hallucination.
Let me explain: Something that is only realized by intuition, dreams, myths etc, and which doesn’t link to experience, is transcendent. Something realized by pairing sensing and experience is transcendental.
Kant’s model is simple and straightforward, as Kant was pedantic, but it still needs some explanation:
Outside us is of course the objects which we sense. Kant calls them “the things in themselves”. It could be the road I was walking with my dog.
Kant thinks of us as rational beings who act on the basis of the thing in itself, and that has caused much debate. Skepticism will claim that the thing in itself is not available, and that there is only the subjective experience. Logical positivism will claim that the thing in itself doesn’t exist at all. Realism will doubt the subjective. We can probably all appreciate that the always rational human doesn’t exist.
But Kant’s bridge is interesting. What he says is that even though “the thing in itself” is not available to us, we can still say rational things about what we see.
So the mind is connected to the real in a way so we can gain and share experience. Does it sound weird? In a way it is, but Kant’s arguments has generally stood the test of time and critical philosophers – and even neuroscience.
So let me tie this to testing.
 

Belief as a Premise

There are different ways to test: In exploratory testing, we do it interactively by operating and observing the software. In automated testing we “outsource” the actions to another piece of software, and our task is then reduced to making sense of data from the tests and suggest and possible implement changes to the test-software. Scripted and chartered testing sits somewhere in-between the two “extremes”.
However,no matter how we practice testing, we need to make sense of what is observed. And since observing is subjected to sensing, the only thing we have available is our intuitive image about the thing we are testing.
James Bach is quoted as saying “Belief is a sin for testers.” I like the quote as it is an important reminder to be careful what we think: It’s not reality. The road might not only be wet. The software probably doesn’t always do what it did this time. I probably missed something. My mind is hallucinating.
So with a bit of wordplay in Kant’s home language, German, I’ll say that “die Sinne ist die Sünde.”
Our senses are the sinner, but as they are also our only hope to see some tings about reality belief is not an option. It’s a premise.
But since we know, we can establish the transcendental: Think the real rationally by testing our beliefs.
In other words: The realist approach to testing is to test the product. The transcendental approach is to test beliefs.
 

On Common Terms

There is something missing in the above as so far I’ve only talked about sensing, imagining, and experiencing. The brilliant part of Kant’s philosophy is that he explains how we can collect experiences.
Kant develops four categories of terms that we think by, and argues how they are given to us a priori, i.e. before we experience anything. He argues how they come from the existence of time and space. Back in his time Newton had just published his theories. Today, we’ve progressed, and it probably makes better sense to think of the terms as a result of the experience of space and time.
But what’s important is that although our experiences vary, we’re on common terms, so to speak.
This is important as it means we can think and express our knowledge about experiences generally.
Let me give some examples: I told you about the black ice on the road above, and while cannot be certain what I said is true, you can understand my experience. I can also share a testing problem, and we can imagine solutions together. I can try them out afterwards, and share experiences with you. We can even talk about testing in general, and imagine solutions to certain testing problems in general.
In other words: The terms allow us to relate, connect, discuss, collaborate, learn, reflect, prospect etc.
This makes the transcendental model of experience complete: We can sense, imagine, think, and express our thoughts into words and actions that we can share with others, who can do the same.
 

The Two Things I Want to Say

So what do I want to say with all this? I want to say two things:
The first is that yes, we are trapped in hallucinating minds. We might theoretically be able to escape them if we subject our testing to strict scripted procedures, make sure what we do is repetitively accurate, and only communicate what we can verifiably record and therefore objectively observe. But we’ll essentially be turning ourselves into machines and miss intuitive and tacit knowledge. And one way or another, we’re still stuck in a mess where at every and any judgement and decision made will be based on hallucinations.
But we’re not lost as we can explore the product and our intuitive ideas about it transcendentally, i.e. by realizing that both are in play when we test. Although we can’t get access to the “thing as it is”, we can experience it. Our expeirences do not have to be transcendent, i.e. disconnected from real, but can be transcendental.
And this is the second thing I’ll say: Since we are not alone in the trancendental, our roles as testers become clearer.
People are different, but I think a fundamental, perhaps even genetically coded, qualification for testers is to be sensitive people with intuitions which are easily disturbed by reality. On top of that, great testers need critical thinking skills, i.e. courage to doubt intuitive illusions, and creativity to come up with test ideas useful in the context. The rest is about interaction and communication with teams and stakeholders so that the good hallucinations about the software that we develop through our testing are shared.
 

Testing Transcendentally

In the spirit of Anil Seth, the neurology professor, let’s be honest: Software quality is a hallucination.
We can’t escape our minds and the apparent mess created by the hallucinations we think of as real. But we can experience quality transcendentally by testing.
To me testing is not so much an exploration of a product.
I see testing first and foremost as the transcendental practice of exploring of our own, and our team colleagues’ and stakeholders’ hallucinations about the product.

References