A Stranded Constructivism
The course had me in the first ten minutes. Some time later I finished it bored, ambivalent, and fairly sure I would not be using most of it. This is the account of what happened in between, and why I think it matters for anyone who teaches.
It was free, it was from a well-regarded university, it was on coaching students through the kind of conversation I have several times a week, and it opened by drawing a line I found immediately useful: coaching is both a mindset and a skillset. I wrote simple brilliance in my notes — the pleasure of grounding future learning from a stable conceptual footing. Then it asked me to name a problem I was actually facing. I chose choosing between curriculum specialisations, and the course used my own problem to model a coaching conversation back to me. I loved the straight-to-practice imperative. This evoked a passionate haste for me to dig into the course and acknowledge the model tutor as a worthwhile source of expertise.
I want to be careful about what I am criticising. This is not a bad course, and the people who built it are not incompetent. What happened to me between that opening and the end is a pattern in how online courses are made, and it is worth naming, because I think it explains a lot of teaching that looks right on paper and dies in the room.
Two ways of teaching, briefly
There are broadly two ideas about what teaching is.
The first is that knowledge is a thing an expert has and a learner does not, and teaching is the transfer. Explain it well, check they received it, move on. Call this instructivism. It is not a slur — direct explanation is often the fastest, kindest thing you can do for someone, and the research on worked examples is genuinely good.
The second is that knowledge is something the learner builds, out of what they already have. It is constructed rather than transmitted, subjective, building on prior constructions. Teaching, on this view, is arranging the conditions under which someone builds. Call this constructivism.
Most decent teaching uses both. The question is never which one, it is whether the support you supply gets taken away as the learner gets stronger. A worked example earns its advantage only where the support it supplies is progressively withdrawn. That is the whole trick. And here nothing was withdrawn at all.
The contradiction at the centre
Early on, the tutor says something that stopped me:
That is the course's entire premise, stated well. And I couldn't help but wonder that if the premise of inner wisdom is that the learner already holds the solution, and then that contextuality is instrumental to enabling their realisation, why is coaching delivered as settled doctrine? The course provided no format or dialogic space sufficiently dynamic to encourage an interdependent arbitration of meaning, and this deep furrow stole a good deal of the momentum rightly put to the model's progression.
Every module ended with a checkbox quiz: unlimited attempts, a show-answer option, and bare correctness feedback. I wrote in my notes, very straight forward, lack of critical thinking required. Being wrong was resolved by clicking a different prompt, or sourcing the identical answer-phrase from the previous segment. Right answers advance, wrong answers demand remediation — which is the textbook test for behaviourist design, sitting inside a course about inner wisdom.
The quiz registers recall. It did not register learning.There was a pair of video simulations, a student seeking support from a first dismissive, then engaged tutor. The second was genuinely good — brilliant authenticity, enough to move me. Then it was reduced to a proscriptive set of question answers: choose 3 terms that best describe the educator's approach: distracted, respectful, curious, impatient, judgemental, non-judgemental. Obvious antonym pairings. And because no diagnosis had taken place, no reliable assessment of learner proficiency was integrated into the course at all. The course could not tell whether I knew this already. It had no way of finding out.
The discussion space, when it finally came, replicated the simulation of the previous module. I felt unenthusiastic about contributing, a feeling which compounded when the presentation and content of the invitation to discuss was generic, uninviting and loose, its requirements hardly signposted. I submitted anyway, and had completed the course before I was responded to.
That last detail is the whole problem in miniature. The only genuinely constructivist thing in the course happened after the course was over.
Why this is a tutor's problem, not a reviewer's
I did not take this course as a critic. I took it because I sit across from students and try to get them to find things themselves, and I wanted to know whether these were competent strategies to apply in my professional practice.
Some of it is. The mindset/skillset distinction is good. Perceiving students as possessors of their own solutions is good, and I wrote at the time that the modelled approach provided a wonderful next step, while I would probably hurry to some didactic solution without scaffolding a way for self-motivated learning (and the lovely habits that come along). That is a real thing the course did to me, and I have used it since.
It is worth saying that the people who built the course have published their own evaluation of it, and they reach a warmer verdict than mine — reporting that students came away with high intrinsic motivation, critical reflection and an intention to apply what they had learned (Maini et al., 2024). I do not doubt those reports. I doubt what they measure. There is a recurrent problem in simulation-based training of non-technical skills, where what gets assessed is how acceptable participants found the experience rather than whether they acquired the skill (Bracq et al., 2019). I finished the course reporting interest too. I could not have demonstrated competence, and at no point was I asked to.
But my worry, sitting there, was practical and it never got answered: there was no way to analyse whether a student did indeed possess the knowledge, or did need domain-specific help — which is an immediate throughline I consider daily in my work. That is the actual job. Everything depends on telling those two students apart, and a course that models coaching without modelling diagnosis has skipped the part that is hard.
So my objection is not that the course was too instructivist. It is that the instructivist parts were poorly built and never retracted, and the constructivist parts had nowhere to land. My reflection attributes my frustration, and even my ambivalence, to the architecture lending itself to the cheap achievement paradigm of the massive open online course. The result is what I would call a stranded constructivism: the right instincts, with no structure to register that any learning occurred.
What would fix it
Not more content. The fix is that discussion becomes the place where learning is registered, rather than an afterthought appended to a recycled video.
The obvious objection is that asking more of learners costs you completion. The evidence runs the other way. Fidalgo-Blanco, Sein-Echaluce and García-Peñalvo (2016) ran a hybrid model across two MOOCs, pairing the e-learning platform with a social one and structuring cooperation deliberately rather than simply opening a forum. Completion reached 27.8% and 28.2%, against a platform average of 13.47% and 13.95% and a global range of 5% to 15% — roughly double in both cases. Participants of quite different profiles reported similar perceptions of learning quality, which suggests the gain came from the model and not from who enrolled.
A course that never asks anything of you has not earned your attention either, and attention is what completion actually measures.
One qualification, because I am arguing for the withdrawal of scaffolding and would be a hypocrite to demand none for myself: dialogic space needs its own scaffolding, and by the argument I have been making here, its own withdrawal. Opening a channel is not the same as specifying one.
The thousands of people in a course like this are the resource. The failure is not one of resources but of routing: the course never puts learners in contact with one another at the point where contact would do work. Grouping them by professional background, instead of leaving them undifferentiated, would make scale an asset rather than the liability it currently is. Clumped in undifferentiated and blind to the professional knowledge of my peers, I was bored and made to rely on a fixed protocol of knowledge transference.
Which is, I think, the thing worth taking away, and it applies well beyond one course. If your teaching says the learner holds the answer, the design has to leave somewhere for the answer to come out. Otherwise you have told them they are capable and then arranged the room so they cannot demonstrate it — and they will notice. I did.