Pedagogical Deadlock

Sometimes a student gets stuck in a way the software has no idea how to notice.

The common kind of stuck is obvious. The learner gets a question wrong, the system serves easier material, they try again, they still get it wrong, and the platform keeps looping back through the same topic. The adaptive logic is working as designed. The student is also, in the ordinary meaning of the word, trapped. They have been sent around the same circle so many times that the whole thing has started to feel less like learning and more like being told they are stupid. A good human teacher would have noticed by the third time and changed something completely, a different example, a different medium, a conversation about what is actually going wrong, a break for tea. The software just keeps serving the next remediation, because from its point of view, the process has not yet failed. Nothing is broken. It just is not working.

This is what pedagogical deadlock means. A system can be executing its rules perfectly and still be doing active harm, because it has no representation of the category "this is not going to work, no matter how many more times I try." A proper adaptive platform needs to be able to escalate itself out of its own loops, either to a different strategy, or to a human, or to a quiet suggestion that the learner come back tomorrow. Most platforms cannot do this, because they were never designed with the possibility of their own failure as a first-class concept. In the last situation where you were really stuck, was it the sort of stuck where more of the same would have helped, or the sort where only something completely different would have worked?

Last week we said resistance is often well informed. This week we look at the kind of stuck a learner can be in that the software has no way to recognise.