Learning from scratch

POSTED: January 17, 2020

I published a shorter version of this on Arcada’s internal staff blog.


In November 2019 I reported about a project that Irma Sippola and I started and which we call Snowcastle Valley. As part of this I have committed myself to teach eleven teenage girls and women, and five teenage boys, all attending second chance education and living below the poverty line, to code using the Scratch programming environment developed by MIT.

What follows derives from the field notes of the work I have done during our second visit, which began on December 20, 2019. I have written these notes using an Evernote template I created to allow me to record attendance, work content, observations, outcomes and problems, on my iPad as they arise during the actual workshops. In devising this template I worked from Field Notes: the making of anthropology by Roger Sanjeck (Sanjek, 1990: pp 92-97).

The workshops have raised an interesting series of issues that have a bearing on the teaching styles and pedagogy that we habitually adopt at Arcada. I have come to realise the extent to which our circumstances at Arcada permit us to make assumptions that we never need to question; indeed, that we would feel foolish to even contemplate questioning. These assumptions became very visible whenever I started running coding workshops with these teenagers and their teacher.

The fallacy of competence clusters

We make two important assumptions. Firstly, we assume that students learn ideas and competences in clusters, and therefore if a student knows this then they must surely know that. I have found out that, in ways that surprised me, this does not hold true everywhere.

I have a seventeen year old girl in the group, for example, who has begun to grasp the underlying logic of Scratch, showing an ability to generalise what she has learned to produce new results. However I discovered last week that she has almost no ability in arithmetic whatsoever. She understood immediately that if you rotate something clockwise by adding a number then logically you rotate something anti-clockwise by subtracting a number. However she could not manage to add (or subtract) numbers in such a way as to make something rotate in a full circle. She simply could not make the numbers add up to 360, nor did she understand why they should, nor did she have access to any method by which she might manage to make them do so.

Her ability to grasp the logic of the situation does not appear to have any connection to her ability to manipulate the numbers that would enable her to put the logic into practice. In Arcada I have become used to students grasping both or grasping neither; and thus I have become accustomed to treating them as two elements in a single learning cluster.

The fallacy of explicit knowledge

The second assumption that we make involves what I will call ‘learning mode’. We easily (and usually correctly) assume that teachers and students share a broad set of ideas about “what learning is”, and about how learning happens; and they understand how they should behave to facilitate learning. With these teenagers this assumption breaks down almost immediately.

They have learned entirely through repetition, and they have never had any encouragment to adopt a creative stance towards what they learn or how they learn it. They feel comfortable doing what someone has told them to do, and comfortable stopping when they have done it.

An invitation to take what they have just done and experiment with it simply puzzles them and, in most cases, results in them sitting there waiting for instructions that make sense. Initially I found this disconcerting. Eventually I realised that we do this ourselves, but just not in what we usually define as “a learning situation”.

Using tacit knowledge as a driver

I remembered a time when my elder daughter cracked the screen on her iPhone and I decided to send off for some parts and fix it myself. In those circumstances I definitely didn’t want to use my imagination. I didn’t want to play and experiment. I wanted to go to YouTube, find a step-by-step video that seemed credible, and follow it slavishly.

I now realised that if I had done this time after time I would have eventually discovered (or ‘learned’) how to fix iPhones. I also realised that we learn to make Ikea furniture as adults in exactly the same way. We follow the cartoon-like diagrams slavishly and, if we build enough, we get to recognise intuitively the underlying logic of the system.

Children do something very similar with Lego. They follow the diagrammatic instructions and, if they do it often enough, they come to recognise patterns that repeat. They eventually intuit that stable Lego vehicles have wheels at the corners and thus gain the ability to make their own stable vehicles. They absorb the underlying logic before they gain any ability to verbalise it.

To state this, of course, does no more than acknowledge Michael Polanyi’s pioneering work in identifying the idea of tacit knowledge, and to accept that “[the] act of knowing exercises a personal judgement in relating evidence to an external reality” (Polany, 1967: p24)

Employing tacit knowledge as a pedagogic strategy

These thoughts drastically changed my plans. I began to spend my evenings constructing Lego-like instruction books as pdfs. I abandoned any notion of explaining why (or even how) things worked. I began making a series of short recipe books which require you to do no more than follow the step-by-step pictures. I now began to get the group to make something and then later, when they all had a working game or story, explain to them what they had done. Currently only four of the ten students in the main group find these explanations at all interesting, but those four have developed their understanding further than the others.

Through this process I have not attempted to “teach tacitly” which, I suspect, would prove impossible in both theory and practice. Instead I have attempted to create a framework in which the club members can create games by coding without knowing why they do what they do to achieve the effects they desire. This framework offers the chance to repeat the same patterns of construction in different contexts, and this, in turn, enables students to arrive at a position in which they “can know more than we can tell” (Polanyi, 1967, p4)

“Learning by doing” vs doing

I had worried earlier about whether the teenagers actually learned from the workshops or whether they merely parroted what I did. Paul Van Gelder, who founded SISP where the workshops take place, had asked me this exact question in one of our regular meetings. I have now come to think of this as a false, or at least unhelpful, distinction.

I intend to create sets of relatively small exercises which each produce a finished game or story. Each will have different surface content but will reveal the same underlying principles. Like children with Lego, or adults with Billy bookcases, the club members will gain tacit knowledge through accumulating experience in building these games. If I can create the games in the way I envisage, then the club members will begin to absorb the skills of coding without either they or I necessarily verbalising them.

They will not “learn by doing”, they will simply do. The process of turning their tacit knowledge into problematisable reflections will come later – for some, but not necessarily all, of them.


Polanyi, M (1967) The Tacit Dimension, New York: Anchor Books.
Sanjek, R (ed), (1990) Fieldnotes: the makings of anthropology. Ithaca, NY: Cornell University Press.