Critic Thinking

It's not a typo, the title. I've been thinking about how - like a professional critic - a professional tester frequently needs to bridge the gap between the connoisseur and the consumer, to take the desires, constraints, needs, sensitivities and complaints of both into account when trying to make sense of and assess a product.

I used to spend a lot of time on music: I'd listen to tens of new records every week and go to gigs all the time. I could, and did, talk about subtleties of a drum sound, listen to tracks because of who produced them, argue about whether this or that 12" single was dark acid techno or acid dark techno or dark/acid techno or whatever. I would play records because I appreciated technical aspects of the sounds on them as much, if not more than, the music. I would deliberately seek out ever more subtle or challenging sonic experiences, such as  Dog Pound Found Sound (which I actually played on the radio). Or records made by someone linked to someone who had once played with someone I had once liked a record by. I was a connoisseur.

My wife often thought the records sounded the same, or were simply crap. What she did like she just knew she liked, usually a decision based purely on how what she heard made her feel. Any aspect of how or why or where it was made was irrelevant to her. She was a consumer.

When I wrote about or broadcast a piece of music I, as the critic, would attempt to respect the respect the depth of knowledge and zealousness of the connoisseur and try to place records in some kind of context that would make sense to the consumer while at the same time not compromising the integrity of either party. Similarly, when testing, it can be important to bear in mind both the internalised view of the product and the end user view (or rather views) of it. Being able to do this, and trade them off against one another, is a skill, a craft, an art.

One example; here's the bones of a particular kind of conversation I've had many times over the years:

Person 1: When I do A, the product does B. This was a surprise to me.
Person 2: Ah, yes. It does B because of X, Y, Z. It's how we were asked to do it. It's expected.
Person 1: But as far as I can see, A has nothing to do with X; Y is not visible at this point; Z is a concept that only exists in the product internals.
Person 2: Yes, but that's how it was requested and implemented. It's expected.
Person 1: But when I do A, getting B looks wrong when all I have to go on is X.
Person 2: Yes, but given how we coded it - which was according to the requirement - it's expected.

I have played both roles in this going-nowhere-fast dialogue. When I was a software engineer - or, to avoid offending actual software engineers, when I wrote the code - I was more often Person 2, a specialist with detailed knowledge and perhaps an entrenched position. (The specification as a shield. Discuss.)

As a tester I have sometimes found myself as Person 2 when discussing a feature I'm working on with someone who isn't working on it. When I've invested time, effort and emotional capital I can become defensive; closer to the end of the cycle I can become more of an apologist for flaws. Even on a feature I feel free to criticise myself - and probably have done at length - I can feel uncomfortable with others' criticism of it. Even when I agree with the criticism I can feel obliged to justify the observed behaviour.

In my experience the Person 1 role seems to come naturally to those with less baggage or, perhaps better, background. They can be gloriously indignant at the slightest provocation, focussed on the detail that offends them to the exclusion of all others and unreceptive to any justification. It seems to be a natural trajectory that, once they've been around the product for a while and inevitably been inducted into the world of Person 2, it becomes hard for them to decline its comforting, if stifling embrace. (The specification as blinkers. Discuss.)

At the other end of that trajectory, I find that testers with more experience can tend to become more comfortable maintaining both Person 1 and Person 2 positions simultaneously, expressing both without prejudice, balancing them out in context. As it happens, I don't think experience is necessary for this duality, but I do think self-awareness is and that frequently comes with experience.

So, what kind of person will you be today? A connoisseur, a consumer, a critic?
Image: http://flic.kr/p/5VdZka

If You Want to Get Ahead

"Testing helmets the old fashion way" the tweet said.

"It'd be better if that was a brick wall" one of my team said.

"Yeah, that is what the specs asked for" I said.

And how we all laughed, for just a little too long, those sad chuckles of shared recognition.
Image: tweet

Testing is an Onion

The other day Michael Bolton tweeted this

"Art" is the activity of directing attention to things and providing affordances for interpretations. Which is why testing IS an art.

I love this tweet not least because it itself affords so much opportunity for interpretation. I'm intrigued by its possibilities. I found myself picking at it, perhaps even factoring it. A few thoughts ...

"Art" is quoted. Are these scare quotes? Do they suggest some uncertainty, disagreement or ironic intent? Or are they merely an alternative to some stylistic markup such as bold font that you might see in a dictionary definition?

"Is" is emphasised in the second sentence. On that verb, that kind of marker might signify a refutation of some other assertion. Could that be the case here? What else might it be contributing?

Both "art" and "an art" are used. The former is frequently defined in terms of beauty whereas the latter is usually applied to a task that requires skill (e.g. Oxford Dictionaries). It's possible for one thing to be both. Is the use of both deliberate? Is it significant?

Natural languages contain much exception and idiom and do not lend themselves to rules and conventions. Which is why testing IS a language.   

"Affordances" is a relatively uncommon and quite technical word. Wikipedia describes an affordance as "a quality of an object, or an environment, which allows an individual to perform an action" but points out several variant usages. Might it have been chosen because it is the only word that gives the precisely nuanced meaning that was desired? Or because its rarity provides memorability to the whole? Or something else? Whatever the reason, it's a sore thumb here and potentially brings more than its core meaning.

The first sentence might be a definition of the term "art". It might be metaphorical  or analogy. It might be explication. It might merely be attributing some property to the concept. Which is it? It is any? Could it be more than one?

The format of the tweet is syllogistic, probably a variant of the Fallacy of the Undistributed Middle  in which there's an implicit premise, "testing is the activity of directing attention to things and providing affordances for interpretations". If you interpret the text literally, then it is subject to a potential logical fallacy and the extent to which you accept it depends on the extent to which you believe that the set of all art encompasses the set of all testing.

Even if your interpretation is not literal the implicit premise is hard to avoid.

Semantics is the study of meaning. Compositional semantics builds the meaning of the whole from the meaning of the parts. For instance, you can work upwards from the meanings of words via the grammar of each sentence to the meaning of the entire tweet. Restrictions are placed on higher-level potential interpretations by concrete interpretations of the components.  But it's both a bane and a beauty of natural language that the information transmitted can be a gestalt: more than the sum of the words.

"Pragmatics" is the identification and application of context to observation. Which is why testing IS pragmatics.

Pragmatics is a layer of meaning above semantics where context is added to the interpretation.  With the right context, any and all aspects of meaning can be changed, however logical or illogical they may appear to be without it. Stylistic considerations and communicative intent can form part of this wider context. What kinds of contexts might wrap around this tweet? And need they be exclusive? A few more thoughts ...

Perhaps the structure of the tweet is simply a rhetorical device for attributing characteristics to testing. It doesn't do to overlook the obvious. Sometimes.

There's a tradition of discussion about whether testing is an art and/or a science, or art or science, or artistic or scientific. The tweet locates itself in that tradition by its subject matter.

But maybe it's also part of some specific chain of dialogue or discussion that we aren't seeing the rest of in the Twitter timeline. We've said the emphatic "is" could be a response to some other statement, something like this:

"Art" is an activity that is driven by aesthetics. Which is why testing is NOT an art.

Perhaps it is sarcastic, trying to illustrate the way that carefully chosen wording can associate two distinct concepts to justify some position.

Michael has a self-declared Mcluhanite tendency to say things for their provocative effect, to make others think. So perhaps I've fallen into his trap and I'm self-yanking my chain here.

Perhaps it's just a throwaway tweet and doesn't bear any inspection: it's not intended to have any meaning beyond recording a thought in the moment it was originated.

Language can be a noisy channel for communication and Twitter as a medium naturally emphasises this because of its limit on length. But in other places other constraints apply - time, budget, the expected reader, the skills of the writer and so on. In those places and at those times where we need to minimise noise and maximise signal the onus is on us to do so by considering the kinds of messages we're sending or receiving: the words themselves and the context in which they're bundled.

Onions are made up of layers which can be difficult to uncover individually without skill and effort (and they can make you cry). Which is why testing IS an onion.

P.S. Michael was kind enough to criticise an early draft of this post and my motivation for writing it. Somewhat ironically, but entirely correctly, he identified that both could be clearer.
Image: http://flic.kr/p/9EEfp1

It's a Syntax

More quotes from The Complete Plain Words:

George Saintsbury denounced the futility of trying to 'draw up rules and conventions for a language that is almost wholly exception and idiom'. 

Jespersen preached that the grammar of a language must be deduced from a study of how good writers of it in fact write, not how grammarians say it ought to be written.

Forget the debate about whether testing is art or science. I'm wondering: is testing a language?
Image: phpSyntaxTree

How-abouts and What-ifs

Happenstance. Ideally, I wouldn't have been reading two books at once. But I'd already started the one when the Dev manager chucked the other onto my desk as he stalked past, wild-eyed from some meeting, en route to the kitchen for a caffeine salve to the throbbing vein in his forehead. Intrigued (with the book not the vein, I've seen that plenty of times now), I started flicking through it, got hooked and then alternated between it and my own over the next few days, seeing the connections that both made to the, ah, idea that ideas (or lack of them) can be a problem.

The Dev manager's book, Fooled by Randomness by Nassim Nicholas Taleb, talks about consequences of an insufficient grasp of potential outcomes. Taleb's domain is financial markets and the instruments that populate them and he attributes the false confidence of many traders - and the population at large, by generalisation - to a lack of understanding of first possible scenarios and second the ability to make sensible estimates of their probabilities. Oh and, third, a flawed sense of their own prowess based on a misunderstanding of the extent to which their past performance was down to them or to chance.

In the the other book, Lateral Thinking by Edward de Bono, the proposal is that logical or vertical reasoning can only take you so far, that there are classes of solution that are unlikely or impossible to find by starting from a known state and simply reasoning. Lateral thinking, for de Bono, is a way to generate what-ifs and how-abouts rather than following existing lines of thought. He makes connections between lateral thinking, creativity, insight and humour, observing that a cornerstone of all is the ability to recast a set of circumstances. For example, humour frequently comes when pulling the rug from under a set of assumptions, exposing a different way of viewing a scenario. He asserts that the operation of the mind favours reinforcements of existing thought patterns and - perhaps less intuitively - that those patterns are significantly influenced by the order that information was first encountered. (Taleb has a similar concept of path dependence.)

Coincidence. Only that week I'd been talking to a couple of members of my team about trade-offs between (a) time spent thinking about a problem, researching it, exploring possible lines of attack, sketching out potential consequences, identifying commonalities and differences amongst the various approaches and (b) thinking of a plausible solution, diving in and just doing it. The knowledge gained from the former has the potential to significantly improve whatever action is ultimately taken. But it might also turn out to be worthless and still have consumed your budget. You might win big choosing the first idea, if it just works, or you might end up with a compromise when, late, you realise there's something significant you missed that you feel you could've and should've thought of. Barriers to trying the former can include functional fixedness, the open-ended nature of it, managers putting pressure on, fear of burning through the budget for a project without producing anything and the fact that JFDI can produce something good enough.

Taleb talks about the downsides of a narrow vision, saying that because of the relative infrequency of paradigm-shifting events - the kinds of thing a superficial or non-existent analysis would miss - and the large numbers of traders, many traders with little or no skill can do well by following a trend and/or having good fortune. When an unexpected event - a Black Swan, as in a later book of his  - does appear, a proportion will just happen not to be affected, will appear prescient or have unwarranted strategic smarts attributed to them, will themselves reason out an explanation for their success - so-called hindsight bias - and perhaps even pursue what they think they have been doing more aggressively afterwards.

To contextualise Taleb's example: assume that in any given year there's a 50% chance of a journeyman tester not being found out after missing some potentially serious bug. If we start with 1000 testers then, even after ten years the chances are that a handful will still have been fortunate enough not to have had adverse consequences (1000 in year 1, 500 in year 2, 250, 125, 64, 32, 16, 8, 4, 2, 1).

As a tester, as someone who invests time trying to balance the cost of applying effort, and where, against the risk of not applying it, it is humbling and worrying to think that the results of my work may be as much, or even more, down to luck as to anything I do or did. I may miss rare issues, glaring issues, trivial and severe issues but if they happen not to be encountered by someone who matters or at a time that matters or with an effect that matters then I may still be thought of as a success and I may still think of myself as a success and continue on in whatever approach I've been using. The one which assists me in missing those issues. A tester could go through a whole career making poor decisions but never seeing failure. You could be interviewing that person, with a stellar CV, right now, for that test manager position in your organisation.

So what is Taleb's prescription? First, it's selfish; to take advantage of it by invoking a strategy that aims to win big when the rare event occurs but at worst lose small at other times. More generally he suggests being aware of the human proclivity for being fooled means that you can at least take account of it. Further, being able to envisage situations other than the status quo will enable the probabilities of such events to be considered. Next, he counsels that:

Maximing the probability of winning does not necessarily maximise the expectation from the game when one loss is catastrophic

He's talking here about the fact that probability is insufficient on its own for evaluating risk. It has to be associated with some measure of cost in order to provide an expectation. For instance, suppose that as a trader the probability of losing £10000 on any given day is 1/1000. The cost is extreme, but the probability is small (to a human mind thinking along standard paths). Let's say that the probability of making £1000 is 4/1000 and of making £1 is 9995/1000. So in the course of 1000 days, or around three years, our expectation is a significant loss although almost every day shows some profit. If the trader never thinks of the 1/1000 event the expectation over the period would be an apparently safe, if modest, profit.

9995 x  1          £9995

4    x  1000       £4000

1    x -100000  -£100000

----             ------

1000             -£86005

       
Ideas are the currency of creativity. The fuel of furtherance. The driver of disruptive actions. Without ideas there's no route off the beaten track barring accident, and a strategy that relies on timely accidents for its sole source of innovation is setting its user up to fail at some point. Of course, Taleb is not the originator of expectation and de Bono is not alone in thinking about ways to provoke ideas (and he's moved things along since this particular 1970s work too) but, regardless of the sources, these notions can be useful to testers at both micro and macro levels.  Brian Eno famously has his Oblique Strategies cards, for example, and  various testers have suggested using similar schemes to assist with test idea generation.

Over the years I've come to value up-front exploration (including physical and thought experiments, proof-of-concepts, prototyping) to build a model of the problem space before implementation. Frequently, even where the budget for a project is tiny, I'll try to isolate some time, however small, for it. In particular, I've learned the hard way that my first idea for the solution to a problem is usually wrapped up in my notion about how I'd implement it given what I know about other implementations. You can get out of that mindset by consciously disentangling the what from the how, but you can step back further to deliberately make time to consider the what-ifs and how-abouts and it's here particularly that lateral thinking can help.

For de Bono, lateral thinking is a skill that you can choose to apply. His book is all about techniques for the generation of ideas, about disrupting thought patterns, the provocation of thinking outside of the norm, the understood. For him, by generating more suggestions you have a wider pool of starting points to consider. Some of them may, on evaluation, be clearly absurd or impractical, but may cause you to think of something else which isn't or spur someone else on to another chain of ideas which don't have the same flaws.

Logical reason and lateral leaps can be interleaved in any way, there is no need to stick exclusively to one approach; feedback from one round of work can and should influence the next round. He is enough of a realist to talk about stopping heuristics for a round of deliberate lateral thought (it's not rocket science: use time boxes, create a certain number of ideas ...).  Although the analogy is not complete, there's some parallel with the view that Rapid Software Testing has of the potential interplay of exploratory and scripted testing.

 

If you bring in the notion of expectation too (where the cost could be the risk of failure), you have some basic machinery for generating and comparing possibilities, for rudimentary prioritisation.

One application for lateral thinking that seems to get relatively little attention is that of identifying alternative solutions to something that already works. Frequently, we're approaching tasks from the point of view of an identified problem: test this new implementation, fix that abhorrent behaviour, improve the performance of that operation and so on. Letting yourself look at something extant and considered sufficient with a lateral eye can be productive (if not necessarily always popular).

On the other hand, lateral thinking may be less immediately applicable when investigating failures. This is often a place where reasoning from the known or observed is the best starting point. Looking at logs, inspecting customer reports and so on will often provide enough evidence for traditional logical reasoning to narrow down the problem. When you're involved in a live support call with a customer, suggestions from way left-field may not be appropriate, at least not until the obvious options have been exhausted.

Synchronicty. I came across the de Bono book while browsing the Pelican section of the Book Barn near Bristol (that day I also picked up Crosby's Quality is Free and a box of other stuff including Julie Burchill's caustic Love it or Shove it  all at a quid a pop!). I wasn't looking for it, but reading it put me in a position to make some new (to me) connections. Quite apart from the actions you take when confronted with a problem, that exposure to the ideas and experiences of others is another valuable way to increase your chances of enumerating possibilities. You never know when they'll come in, when the now will remind you of the then and spur that thought, that crucial what-if or how-about.

And so some time this week as I pace rapidly down the office, fresh from some meeting or other, on my way to walk briskly round the block, the vein in my temple pulsing and plum-coloured, I'll slap The Complete Plain Words onto the Dev manager's desk,  muttering "what if I just told them to ..." as I disappear round the corner.
Images: http://flic.kr/p/ca1gm, Amazon, http://flic.kr/p/4qCTgp, RST v3.1.3

Plain as a Pikestaff

Honestly? I usually find non-fiction books hard work but I'm reading The Complete Plain Words like a thriller (see also Long Time Gone). I've only just started but I love quotes like these on clarity:

What appears to be a sloppy or meaningless use of words may well be a completely correct use of words to express sloppy or meaningless ideas. 

Mistiness is the mother of safety.

We need to choose the right words to make our position clear. If we don't, the words will make clear our position.
Image: http://flic.kr/p/627MJn 
Sounds: If I Could Only Remember My Name 

Migrate Idea

Of course (of course!) we were delighted to be asked to provide test support for an internal project which was upgrading an important and widely-used application and rationalising the data associated with it. Of course (of course!) the many application instances are intended to be standardised but, despite best efforts to be consistent, over time we've built up a load of special cases, workarounds, emergency patches and so on, all of which make this kind of migration task more difficult.

Of course (of course!) the deadline was both close and ambitious but there was enough time for us start to write automation to obtain a client-side view of the data and configuration on each instance (via an API) to provide an initial audit and then agree policy for migration with the various stakeholders based on that. We then extended the test code to generate an expectation of the equivalent client-side view for each server on its migrated instance which we'd check (via equivalent API) in the new software.

Of course (of course!) the migration policy turned out to be incomplete and ambiguous in places once the physical migration attempt began and we encountered discrepancies in the observed data, found that some of the data wasn't what we thought it was, identified errors in the migration, identified errors in the test code, were forced to modify policy, were forced to modify expectations, were forced to rework parts of the migration and so on as we went along.

Of course (of course!) across several rounds of implement-test-analyse-fix we converged on a stable system, the automation giving a broad brush regression test at a scale and speed impossible in manual testing but backed up by a team of testers interacting with the application. By the final iteration the automation wasn't in step with the implementation, but we were confident we knew where the major exceptions were and had plans to fix them.

Of course (of course!) we met the deadline and of course (of course!) there was no more test budget because the next priority was immediately upon us. But, of course (of course!) we weren't happy to accept that because the updated software hadn't failed at the point of switchover it would be just dandy once in production. So we looked for a cheap way to get some kind of quick stability test.

Of course (of course!) without budget our options were limited, so we decided to try recycling the test automation by leaving it running on a loop against each of the servers for several days. We had each experiment log to a unique file and write a pass percentage into it. At any given time we could simply grep the logs, import into Excel, and plot the pass rates for each instance of the application for each test set over time.

Of course (of course!) we knew that pass rates wouldn't be 100% in many cases because we never got back to fixing up the expectations of some of the checks. But in this round what we were interested in was the deltas: without some external action, the results should not change. If the results don't change, we have some evidence that the system is stable. When results did change, we investigated, but didn't attempt to change the test code's expectations. Instead we simply looked for further changes to the new pass rate.

In this chart, time point 4 looks like a transient infrastructure issue since all servers experienced a change in behaviour at the same time and then recovered. Server C's data disappeared at 6 suggesting that the test code might have failed in one iteration, but we don't know why without looking. The same server experienced a sharp delta at 9 while none of the other servers were affected, so probably an issue on that server to investigate. Server A changed behaviour at 7 and stayed there: perhaps a genuine data change, or perhaps some persistent failure.

Of course (of course!) there were some positive results: we identified one back-end server which had to be recommissioned due to disk problems and another which was not being load-balanced correctly.  And of course (of course!) there was some noise and some false results which, if nothing else, on investigation confirmed that the system was working as expected. For instance, we found that contention for licenses prevented some suites from running occasionally. And then of course (of course) there was the growing confidence that constant usage wasn't of itself a problem for the new installations.

Of course (of course!) we did have to expend some resource on this work, but the analysis wasn't particularly onerous and we think the benefits were worth the cost. We tweaked things based on the early round of test to reduce the noise, identify patterns which could be safely ignored and so on. We also started to parallelise the same set of tests and so put more background load on the system as we became happier with the way it was functioning.

Of course (of course!) this kind of approach won't work everywhere and depends on some conditions being met, for example that changes in the data and configuration being checked are uncommon across the test period. It should be noted that it may also not expose different issues, or combinations of issues, that happen to have the same pass rate. Investigation of issues will probably be made more efficient if logging includes diagnostic information and time-stamps so that it can match up with server-side data when there's a need for finer granularity than the simple pass rates and, in fact, we found that grepping (although I'm using ack more these days) this richer data was a good way to perform quick investigations and look for patterns across servers not shown up by the basic pass rate.

Despite the caveats the method worked well on this project, and I'd use it again when I thought it would provide benefit. The implementation team liked it and thought we did a job of the highest quality, were the best testers they'd ever worked with, had great rapport with them, great team spirit, good fashion sense, great hair and smelt nice. OK, they didn't actually say that but they were thinking it, of course. (Of course!)

Image: http://flic.kr/p/4Jrzjx