I grew up with jigsaw puzzles – my mother was disappointed when a birthday or Christmas passed without a new jigsaw puzzle somewhere among the presents. I’ve found nothing as effective to help me wind down while on holiday than unpacking a new jigsaw puzzle, sorting out the edge pieces and settling down to start looking for the patterns. And I still remember one weekend when I was scheduled for a brain scan on the Monday after the weekend, and the only way I could think of to get through the waiting was to focus on a new type of circular puzzle I had never built before.

Jigsaw puzzles are not all born equal, and it’s about much more than size. A large puzzle just requires more patience than a smaller one, but need not be harder to figure out. One of the strangest ones I ever saw, and which I regret not buying to this day, had the same picture printed on both sides, in portrait orientation on one side and in landscape on the other, and you had to figure out which side up the pieces had to go. In the round puzzle I mentioned earlier, it was not even possible to identify all the edge pieces from the word go, as many of them did not have obviously straight edges.

But not even the more common rectangular puzzles all yield to the same strategy to solve them. Some have only a few types of pieces – we call them names like H, I, or “man”; sometimes an I with one set of arms much longer than the other looks like an eagle, and so on. When rows of these simple shapes march on evenly, it is often a matter of matching the colours if you want to make progress. Other puzzles consist of more complicated shapes, but if the sides of adjacent pieces have the same height so that you can use the size of pieces to judge fit, most puzzles can be solved logically – matching shape, size and colour often does the trick. Some puzzle pictures contain a mass of small detail – to build those, you need a good memory for where you have seen a particular feature before.

But then there are the ones no amount of rationality can budge. This summer I came up against one that defied all logic – the woodland picture, with a stone bridge over a rushing brook, was somewhat blurred; elements like branches or stones were interrupted by leaves or foaming water and did not often carry over from one piece to the next, large pieces fitted next to small ones without compassion for the puzzler. Apart from building the frame, my analytical mind was of no use whatsoever. Sometimes, my hand went to a particular piece of its own volition, without me being able to explain why this piece would fit. Other times a piece almost called to me, saying that it fit the pattern of what I was doing then, while I was not even looking for it. The whole time I felt like I was grappling with an invisible foe, as if I needed ninja skills to intuit what my next move should be. Great was my satisfaction when I could finally stand back and admire the finished puzzle, all 500 pieces of it!

So I wondered – has anyone ever developed a taxonomy of jigsaw puzzles? Or constructed a Cynefin framework for jigsaws? That woodland puzzle would surely be a candidate for unordered space!

In our work at the Foundation, we have developed a language of our own with phrases like “speaking Greek to the Italians”, “polishing shoes” and others. The phrases function like metaphors and are short-hand for ideas we have discussed at length previously – the one about speaking Greek to the Italians refers to using Cynefin language when speaking to people who are not part of our team and therefore are unlikely to understand what we are talking about. Most families develop private languages like these; they play an important role in establishing membership and identity.

The same applies to fields of knowledge, of course. The discipline-specific discourse provides more precise language tools than the language in common use. Effective use of the discipline discourse is a crucial element of what students must learn in order to acquire an identity, first in the academic world and later as a professional in their chosen field. As an internal consultant, I learnt early on that the skill to rapidly acquire a new discourse is vital in getting access to any new community you want to work with. The downside, as Dave pointed out in his blog earlier this week, is that the use of language can also be used to exclude people from a community.

In a multi-language context, like South-Africa, language in general poses challenges. My grandmother, like Dave’s, also told stories about having to wear a plaque with the words “I am a donkey” when she was caught speaking her own language, Afrikaans, in school in the Colonial days. Language issues played a role in the 1976 Soweto uprisings; in fact, the power issues inherent in language play a political role everywhere. Recently I heard an interesting debate on the radio about the recognition given to the vernaculars of particular social groups and to regional forms of Afrikaans when it comes to what is considered to be the standard form of the language used, for example, in newspapers.

These factors come together dramatically in education, as language is the medium that carries the concepts the children have to learn. The South African education policy allows for mother tongue education for the first three years, with a chosen language of teaching and learning, usually English, added in the next three-year phase. After that, the language of teaching and learning is supposed to be used in the classroom. In reality, urban communities no longer are homogeneous in terms of home language; in fact, many families do not even share a common mother tongue. Most classrooms, therefore, are multi-language environments. In a mathematics classroom, learners must acquire mathematical discourse while being taught in English, while neither the teacher nor any of the class speaks English as a home language. If the teacher wants to revert to a home language to explain a concept, he or she has several languages to choose from. Then in the science classroom, the concepts are expressed in the language of mathematics, which was taught in the mathematics classroom in English, which ….

The tower of Babel, all over again …..

When I was a statistician, I had almost complete professional freedom, as the people I worked for or with did not consider themselves qualified to judge whether the approaches I took were the right ones. In the world of education, the opposite happens – as everyone has gone to school for more than a decade, everyone considers him- or herself an expert on how it should be done, and therein lies the rub. Many an intervention is tried, and even implemented on large scale, because it sounds like a good idea, with very little evidence as to its suitability for the particular context.

So the Foundation I work for set out to generate evidence to support solutions in all our programmes. That turned out to be easier said than done in something like a bursary programme. The sample sizes required to achieve acceptable power and discrimination using traditional statistical methods were simply prohibitive in terms of cost. All the ethical issues familiar in social research presented themselves; for example, how can we not provide the support we believe can make a difference to the bursars’ success just for the sake of having a control group in the experiment? And if we see something is not working, how can we not intervene just to get good data?

The approach to use safe-fail probes appears appropriate and solves some of the problems, but raises others. What we do know about students is that they often don’t realize they are in trouble until it’s too late, and that for many it’s a self-esteem issue to not ask for help, so expecting them to voluntarily join a support programme may not meet their real needs. Although the combination of narratives and quantitative data available from SenseMaker indeed convinces powerfully, the challenge still is to collect the narratives from a diverse enough population to see how the conclusions hold across different contexts before one can hazard any recommendations.

Then we get to the philosophical aspects of what constitutes convincing evidence for policy purposes. When I refer to evidence-based policy making, I often get a most cynical response. Case studies in the medical world show that a strong emphasis on data can lead to over-diagnosis and cause more problems than it solves.

So to get beyond implementing education policies just because they have worked elsewhere or seem like good ideas is not a simple problem; anyone who has experience with it or has an interest in grappling with it is most welcome to help us think about how to do it.

When my daughter was two or three years old, she asked for a story one night, and of course, being a devoted mother, I obliged. But the next night she asked for another story, and the next night, and the night after that…. Soon it started to feel like hard work to come up with a new story each night, so I developed a story-generating algorithm, which went something like this:

* Think of something that happened that day
* Start with “One day, a long, long time ago there was a …”
* Personalize the thing or animal you thought of into a character she could sympathize with
* Create a problem for the character and explore the consequences
* Find a way to solve the problem
* End with a description of how the thing or animal lived after the problem was solved.

This made it much easier to construct an impromptu story. One evening I even ended up with a pink flannel sheep social network. It started with the pink sheep on her green flannel pyjama suit. Of course the sheep was lonely; he had no other sheep within reach to talk to. Until one day (a key phrase when constructing the story), when he realized that where the pyjama top covers the pants, he could get close enough to another sheep on the pant half, and they had a lovely gossip. Then they both found other sheep they could reach across a fold of the fabric, or where the sleeve rests on the pants, and soon all the pink sheep were sharing news by passing messages up and down the pyjama suit, telephone-style.

She often fell asleep before the end of the story, especially on nights when I was at my most creative. The story-telling was a sort of “all is well with the world” ritual that eased the transition from the busy day into peaceful sleep time.

Then she became a teen-ager, and suddenly my stories were too childish. So I threw out the story generator and started anew, thinking up more complex, darker stories that did not always have happing endings. After a few nights, we mutually decided that this did not work – it did not have the same comforting, ritual effect, for one, and it was hard work to think up original stories. For a few weeks, we stopped the stories altogether, but that also was not satisfactory: bed-time wasn’t as much fun anymore. So eventually I went back to the old recipe, choosing more grown-up characters and giving them more typical teen-age problems. This strategy lasted until she went to boarding school at the beginning of the year. When she was scared and alone that first night, I sent her a text message starting with “One day, long long ago, there was a laundry basket, which went to boarding school for the first time”, inviting her to complete the rest of the story. And that was the end of the story-telling ritual – I miss it often!

I wonder though, while the story generating algorithm started life as an effort-saving device for mom, it does represent an archetypal story plot. Will she somewhere in her subconscious, when she grows up, have a programme that says “every day brings a new problem, but never fear, every problem has a solution”?

Well, I suppose a lapsed statistician is a more accurate description of my current status in the field of statistics – I haven’t proven a theorem in a quarter of a century, the last time I tested a hypothesis was two decades ago and as for data-analysis, well for that I now have SenseMaker Explorer!

When I started out as a statistician, there were no personal computers; we made very strong assumptions just to be able to calculate the results; non-parametric methods with fewer assumptions took all-night runs on the university mainframe for a simple hypothesis test. Then came the PC; suddenly exploratory statistics became possible – not that it was considered rigorous enough to be proper statistics back then. But I loved it – it was like being a detective, looking for structure in a mass of data, using dimension-reduction techniques to squash the information down into two or three dimensions that could be visualized and interpreted, looking for patterns in graphs and finally, when the data set gave up its secrets, finding out what the patterns I saw could actually correspond to in the real world! Unfortunately, in the first stage of analysis, the pattern often corresponded to a management decision nobody bothered to tell the poor statistician about, but that’s another story.

Sounds familiar? When I first saw the scatter plot matrix in Explorer, I felt like I’d come home – even better, I could now do the same kind of analysis on “soft” data too!

I was taken aback initially when I was told that statistical analysis is best applied in the chaotic domain, though – I had always thought of it as a very rational, analytical tool to help make sense of uncertainty. Well, that’s not contradictory then, is it? Agents acting independently… independent observations as a basic assumption for statistical methods … ok, so statistics could make sense of chaos ….

To my statistical mind, fitting linear (e.g. polynomial) regression models to data is an admission that you don’t know anything about the underlying mechanism, otherwise you would have used a more explanatory model. Linear models only represent correlations and no causality can be inferred from them; they only apply to the range covered by your experiments and cannot be extrapolated to different conditions, but they can indicate promising directions in which to search for better solutions – so again, isn’t this what one does in the chaotic domain?

So I have to acknowledge, statistical methods are applicable in the chaotic domain, especially the empirical type of methods (functional models and many other techniques fit better in the complicated domain). But I’ve never ever simply discarded an outlier, they are the most interesting species of data point!

In our work at the Foundation, we have experienced one example after the other of happy coincidence. We would go to an official meeting, only to find that the chairman of the meeting is a childhood friend, or university roommate, or close colleague from a first job, or … I would identify a suitable partner for a project, only to learn that the Foundation’s director, Mpho, helped him to set up his business. We would decide to tackle a certain issue in a particular way, and then discover that the ideal opportunity to present the case is coming up in two weeks and we can still get on the agenda …

Now I’ve long ago discarded the “assumption of intentional capability”, especially the version that implies that when something unpleasant happens, someone intended harm. Such a position is simply not tenable if you view the world as a collection of systems at different levels, going about their business and interacting with each other when the circumstances require. That some of these events will have effects that I experience as negative, is just a simple inevitable fact of life – no malicious intent is required to explain it (I’m also not a fatalist – some of the agents in the systems I’m part of may indeed intend harm, but that’s quite rare in my experience).

Strangely though, I find it harder to interpret such synchronicity as impersonal when the outcomes are favourable, as in the examples I started with. There is a real temptation to interpret such happy coincidences as good omens, signals that we’re on the right track, or that someone or something is helping us.

I suspect the facility for happy coincidences we have experienced has more to do with having a receptive eye for opportunity than with extra-ordinary fortune. I’ve often used the teaching example of exploring possible choices for a new car – once a particular model of car has become “top-of-mind”, suddenly you notice hundreds of them on the roads where you never had before. The “gift” of serendipity may require nothing more than being open to the possibilities of events as they unfold – of course, being connected to many different networks also helps!

Then again, a famous South African gholfer, Gary Player, is reputed to have said: “The more I practice, the luckier I get” …

The last year or so has been a fascinating roller-coaster ride of excitement and frustration as a small team of us worked to set up the Sasol Inzalo Foundation, whose goal is to be a pioneer and leader in STEM (science, technology, engineering and mathematics) education systems reform in South Africa. We’ve met new wonderful, passionate people, but also rediscovered an amazing number of friends from the past; we’ve experienced so many unbelievable coincidences that we’ve stopped to be surprised by serendipity, and the intellectual challenge of engaging with complex ideas has more than made up for the frustrations of grappling with the South African education system and its stakeholders.

As this journey forms the backdrop to much of what I want to write about, let me start with some background about the Foundation: it was set up recently by Sasol, a large South African petrochemical company, and the world’s largest producer of synthetic fuels, hence the STEM focus. However, the Foundation has an independent board and our mandate is to focus on STEM skills development and capacity building for South Africa, not on Sasol’s future talent needs (although Sasol will benefit indirectly too, of course).

The current reality of the South African STEM education landscape is rather bleak. As in many other democratizing countries, there has been a huge drive to get all children into school, but the education system simply did not have the capacity to produce good results in the face of massification. South Africa performs poorly on almost all international education benchmarks, and there is a general despondency that the education system is not delivering on expectations. Moreover, despite significant public and private spending, the picture has not really changed for decades. So our team decided to not throw good money after bad by giving more grants, but rather to position the Foundation as a pioneer and leader in Maths and Science education systems reform, collaborating (as an operating Foundation) with others who share our vision and engaging with influencers and decision makers, in order to have a meaningful impact on the system.

But we needed a fresh approach – the traditional ones did not seem to work and many change agents were getting exhausted. We also wanted to put gathering evidence and impact assessment at the core of what we do, as many existing interventions were based on what could at best be called anecdotal evidence. When I met Sonja and Aiden from the Narrative Lab and Mpho Letlape, the Foundation’s MD, realized she knew Dave from IBM, things just fell into place. We purchased a SenseMaker site license, and plan to use it to underpin all our work. Wherever we share our plans, people get excited – our different approach has provided an injection of energy for many.

One of our first projects is a bursary scheme for engineering and science undergraduate students, with the aim of understanding how to make a diverse range of students successful (throughput is a major issue in South Africa). We are currently putting the final touches to the signifier design to track the affective aspects of the students’ experiences with SenseMaker – more about this in future blogs.