- Image by sunbs35 via Flickr
When you think of art that was produced in 1917, the heyday of Renoir and Monet, a men’s urinal is probably the last thing to come to mind. Yet, Marcel Duchamp submitted this “masterpiece” called “Fountain” as his statement about the quality of a certain art show. (Read more about him here and here.)
Is this great art? Is it art at all? What about it is or is not art? Duchamp did not make this himself. He bought it. What I love about the whole episode is that it was so incendiary that people still argue about it. People argue about this in much the same way the software testing community will be arguing about the meaning of quality until we’re all dead and the aliens are trying to decide if a McDonald’s hamburger wrapper should be catologued as that-which-the-strange-creatures-called-“art”.
This is why I love languages, writing, art and music. They deepen the meaning of context. They find the core of our humanity and our attempts to relate to one another.
There was a lively discussion on twitter today about art and quality:
@chris_mcmahon was asking if this is a high quality painting. @lanettecream doesn’t think so. @shrinik pointed out that quality is personal. @michaelbolton thinks that it’s impossible to mention quality without also connecting the quality to someone specific (an idea of Jerry Weinberg’s in Intro to General Systems Thinking. I guess I should read that.) I said: is that a trick question or what? That’s like asking someone to define art.
When I asked @lanettecream why she didn’t like the painting @chris_mcmahon linked to, she said it looked, “boring and too male.” It’s obviously not of high quality for her.
@chris_mcmahon followed up with:
it’s a painting by Franz Kline that sold for $5,122,500 in Nov 2008. does that change your mind?
Note that this is exactly the type of question and argument that goes on in any art history class.
So if quality is subjective and perception is reality, where does this leave Duchamp’s “Fountain.” Duchamp obviously didn’t think it was of high quality. In fact, he submitted it under the name “R.Mutt” because he didn’t want it permanently associated as his art (haha). While it was accepted for the art show, it was never displayed. The art world is divided. People pay millions for reproductions of it…what does that say? How do we define quality at all? How do we define art?
Although it is easy enough to say “context,” I question whether that is an oversimplification. What happens when the context is shared by hundreds, thousands or millions of people over time? My honest opinion of defining quality is that there is no one definition. Even in defining quality and art, any levels that we give them such as high or low will eventually become superfluous much in the way that “with a long enough timeline, the survival rate for everyone drops to zero.”
Here is a youTube of a performance by the performance art group, Survivial Research Labs. They build robots and unleash them upon each other until everything is completely destroyed. I find it oddly hypnotic and comforting. I think I’ll watch it again while I ponder the meaning of testing vs. checking.
Update: Michael Bolton left a really great comment. Those of you using a reader might want to click through.
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_e.png?x-id=49c871fd-49c8-4479-9f0e-99ad571c26bf)
To be clear, Jerry’s insight is that quality is not an attribute of something, but a relationship between the person and the thing. This is expressed in his famous definition, “quality is value to some person(s)“.
You and your readers might be very interested in Louis Menand’s recent New Yorker article, ostensibly on Andy Warhol, but also about the ongoing question of what is and isn’t art. (Alas, the abstract doesn’t discuss this much, but I found the full article very interesting. Menand’s stuff is of very high quality… uh, to me.)
I would argue that art, like quality, is a special case of the general Relative Rule (a general systems law that I named), which says that for any abstract X, X is X to some person. Quality isn’t an attribute of something, but a relationship between some person and the something in question. So too is art.
That allows for a general definition of quality (art, value, complexity, jargon, bug, interesting…) that is subjective and uncertain for the specific case. If we, as testers, want to help people know what they have and get what they want, we have to probe what’s important to them. That in turn requires us to learn more about things, people, and their relationships to each other.
Some find that effort daunting or overwhelming. I see it as an opportunity, a mandate to explore. I loved your line, (languages, art, writing, and music) deepen the meaning of context. They find the core of our humanity and our attempts to relate to one another. I couldn’t agree more.
Done well, testing does that too.
—Michael B.
Michael Bolton liked my blog post!? How did Chris McMahon so elegantly put this type of situation…”EXCUSE ME WHILE I GO HUFF INTO A PAPER BAG.”
In the case of relating to one another, the question, “what is art,” often gives very deep insight into who someone is, in much the same way that a tester’s definition of “quality” says a lot about who they are as a tester. The fact that these questions bring out so much personality fascinates me and is part of the reason why I like software testing so much.
Harnessing the depth and power of these questions is something I hold as a high level goal in my efforts to be a better tester.
Thanks for the great comment…and now to go find that paper bag.
This impressed me a lot when I first read it: http://www.exampler.com/blog/2007/05/22/context-driving/
I’ll get back to that in a minute, but first a digression about that Kline painting. (A digression with very little explanation, because there is a ton of background required.)
Kline and fellow Abstract Expressionists accomplished their best work during the time the New Criticism was at the height of its popularity. And from a New Critical standpoint, the Kline painting is a very good painting indeed.
And the Kline painting occupies an important place in the story of the Abstract Expressionists and the New York School, so from a Structuralist standpoint, it is a very good painting.
From a Postmodern standpoint perhaps it is not such a good painting. It takes itself very seriously.
The problem that I am starting to have with context-driven software testing is that, taken literally, it is very difficult to actually declare any software to in fact be bad. There is much software that supplies value to people that is, actually, very bad software. It seems to me that a lot of context-driven software testing does not examine software itself, but instead is forced only to consider any available value that people can be coaxed to realize from the software.
So in order to actually point to some software and be able to say “this is bad software”, one would have to have some sort of defensible critical stance from which to make such a judgment.
As Marick points out in his “context-driving” piece, agile development provides such a critical stance: if the software cannot be made to have 100% unit test coverage, then it is by definition bad, and the work that faces us is to change the software so as to make it good, and the definition of good is to implement a design so as to achieve 100% unit test coverage.
What I begin to suspect is that there are other valuable critical stances available to us, but that we as an industry literally do not even suspect that they exist. Occasionally we stumble into a critical stance; the Wiki Way may be such a critical stance: to the extent that software does not enable all users to collaborate equally, it is bad.
In the art world we as critics may adopt any critical stance we care to in order to do the best work that we can do. But we as software testers have an appalling lack of available critical stances to adopt from which to actually examine the software itself.
“we as software testers have an appalling lack of available critical stances to adopt from which to actually examine the software itself.”
and that’s exactly the perspective I’ve been pushing myself and this blog towards.
Professional critique is a skill to be learned in any area such as art, music or architecture. It takes just as much training as actually creating the art and goes much deeper than love/hate. How can we, as testers, leverage this skillset for our evaluation of software?
There is much software that supplies value to people that is, actually, very bad software.
Bad to you, perhaps. Bad to the customer, maybe. Or in a more nuanced way of looking at it, there are things that support its value to the customer, and things that threaten its value to the customer. The tester’s role, like that of the critic, is to use our knowledge of the product, its context, and the audience and to identify those things.
It seems to me that a lot of context-driven software testing does not examine software itself, but instead is forced only to consider any available value that people can be coaxed to realize from the software.
I’d consider that to be a pretty serious mischaracterization of what context-driven testing is about. What have you seen or heard to suggest it?
if the software cannot be made to have 100% unit test coverage, then it is by definition bad, and the work that faces us is to change the software so as to make it good, and the definition of good is to implement a design so as to achieve 100% unit test coverage.
I think that’s a terribly simplistic view; not only unagile, but also unAgile if we use the Manifesto as a rubric. There’s a big difference between “bad” and “not as good as it might be”. It might be okay to say, “that’s a bad attribute of the software”. Yet a piece of software can perfectly satisfy a customer without having 100% unit test coverage—and the Manifesto is explicit that “we value working software over comprehensive documentation”.
We as software testers have an appalling lack of available critical stances to adopt from which to actually examine the software itself.
I don’t think that’s true at all. We have an infinity of stances available. The problem, as I see it, is that the stances are available but the in general, the craft appears not to opt for them. The context-driven community exists largely with the goal of recognizing precisely that point.
—Michael B.
I studied some aesthetics in Philosophy many years ago, and the “what is art?” question was intensely discussed.
The best answer for me was, and is:
“Art can’t be defined, but I know if it is art when I see it.”
“good art” was never discussed, probably because it was obviously subjective.
Quality is more like “good art” than “art”, but anyway:
I can tell what “quality to me” is when I see it.
I can tell what “quality to others” is when I see it, if I know a lot about the intended usage and users.
@Michael
I agree that we have an infinity of stances available. For my growth as a tester, I want to use all of the wealth that professional critique of the arts has to offer to improve the way I perceive software and how I relate to others through software. That’s what I identified with in Chris’s response.
It is for this reason that I think this type of critique should be used to train testers. Teaching people how to be better testers or how to be testers at all has been a huge topic of late. For me it can be difficult to focus/defocus or to think outside of I know and have seen before, yet this is crucial for testing. To critique art, especially modern art, it is necessary for me to reach outside the boundary of my personal comfort zone and find a way to appreciate and judge something new and, perhaps, uncomfortable that I then have to communicate effectively to someone else. When I say that I have to appreciate something that might make me uncomfortable, that means that even if a painting is not something I would want in my house because I don’t like it, that doesn’t necessarily mean it’s a bad painting. In that case, I have to learn how to separate my idea of quality from someone elses. All of this is in addition to learning how to carefully observe what’s in front of me. Shrini Kulkarni blogged about that skill a couple of months ago.
It is a big plus for the context-driven community if this type of exploration is happily leveraged.
I like Weinberg’s definition of quality, but I’m not convinced that it is sufficient for a general definition of quality.
Off the top of my head I can think of two concepts that I suspect are important to quality that it doesn’t seem to address: perfection and fulfillment of purpose.
By definition, perfection is ultimate quality. Therefore something that is perfect in same way must, in some way, possess/produce quality. Consider then a perfect sphere. Its perfection is in its sphericity, which is it’s shape and shape is an attribute of the sphere. Does this not mean that quality can also be an attribute of something?
Also is something’s ability to fulfill its purpose sufficient for it to have quality even if doing so provides no value to anyone? For example, assume x & y are both sprockets and that x fulfills the purpose of sprockets, but y does not, because y has some defect that prevents it from doing so. Can we not reasonably state that x has quality and y does not (or at least that x is of higher quality than y), purely on their ability to fulfill the purpose of sprockets, irrespective of whether x or y specifically or sprockets generally provide anyone with value? I think people do this all the time.
Perhaps quality, like most things, requires more than one definition.
@Michael Bolton
Assuming that quality is the “relationship between the person and the thing”. Can “the thing” be anything? I strongly suspect not.
@Chris McMahon
When you say “a lot of context-driven software testing…is forced only to consider any available value that people can be coaxed to realize from the software.” I get the impression that by value you mean benefit. Is that correct? Surely if the costs outweight the benefits, you have grounds to declare the software bad (if you acknowledge that the costs & benefits are context dependent).