"Our engineers were building a bridge across a river, when suddenly there was a flash flood and everything got washed away. What did they do? They protested, saying it was an out-of-syllabus question" -- An old joke among engineering college teachers.
Some problems with our current modes of teaching are well known. Education is too exam and metrics oriented; there is a tendency to study for exams, rather than study to learn something; coursework reduces to a competitive activity, while learning is typically a collaborative activity; and so on..
To be sure, our current model of homogenized, standardized, calibration and metrics-based system, has its merits. From a country, which was reeling under rampant illiteracy and drought, a century ago, our education system has created one of the world's largest talented pool of engineers and other professionals.
A standardized, process-oriented system is somewhat like a factory -- it can mass produce, acceptable quality outputs at acceptable costs.
A process-based system also goes through several phases of evolution. Anyone familiar with capability maturity models can recognize the following phases of process maturity: Ad hoc, repeatable, measured and optimized.
Early systems of pedagogic processes were ad hoc in nature -- typically dependent on coherence between the teacher and the pupil. But as education needs increased, ad hoc processes became no longer feasible. We needed to scale, and for this we needed processes that gave satisfactory results and were repeatable and replicable in several places. Once, a repeatable process is in place, it is not long before we start calibrating and measuring different elements of the process. And once we have metrics, we can optimize on different elements and produce the best possible output for a given input.
However, process maturity does not stop at optimality -- and this is often overlooked. Unbridled optimization invariably eventually leads to overfitting. Overfitting is a state where the system becomes hypersensitive to metrics and starts reacting to, and trying to optimize on every measurable element.
Somewhat similar concerns must have prompted Andy Grove to say, "Every successful organization contains the seeds of its own destruction."
Success breeds overfitting. And this is not just with education. Consider media overfitting to TRP ratings, and you see what I mean. Overfitting gives specific narrowly defined metrics, much more importance than the underlying spirit of the activity.
The solution to overfitting, is to shift the focus from processes to principles. We need to shift the focus from asking how to conduct a course, to what is the course meant to achieve.
The class as a whole (including the teacher) works towards fulfilling a given mandate. The teacher contributes to the mandate by providing lectures, facilitating discussions, etc. Students contribute to the mandate in several ways -- by offering talks themselves, writing essays, leading and contributing to forum discussions, contributing to the classroom glossary, etc.
Each mandate runs for about 4-5 weeks. At the end of this time, the class reflects back on the mandate by summarizing what has been learned. The teacher then does a random sampling from the class to assess how the class as a whole, fares with respect to the mandate. If the class as a whole is found to be satisfactory with respect to the mandate, the current mandate is closed and the course moves to the next mandate.
And that is the overall idea. There is no individual-level quiz or exam at the end of a mandate. The teacher does ask questions -- but anyone could be asked. So the class as a whole needs to be prepared for the end-of-mandate session.
Grading is based on the evidence shown by the students (on an online platform like Moodle) towards their contribution to the mandate. The more evidence that students show that they are helping the class meet the mandate, the better it is for their individual grades.
At the end of the course, typically we have a (summer/winter) break for a month. The last mandate in the course is a reflective mandate, and is open through the break. Students can use this time to reflect back on the course and try and implement something based on what they have learned through the months, and contribute it back to the rest of the class.
The design objectives behind the above model are as follows:
The model has met considerable success so far, but I have thus far refrained from quantifying its "goodness" against conventional models, to prevent overfitting. Thus far, the goodness of the model is apparent by the (formal) feedback received by students and the (informal) observation that students who have gone through this model in the past seem to be applying their knowledge well, in their professional lives.Some problems with our current modes of teaching are well known. Education is too exam and metrics oriented; there is a tendency to study for exams, rather than study to learn something; coursework reduces to a competitive activity, while learning is typically a collaborative activity; and so on..
To be sure, our current model of homogenized, standardized, calibration and metrics-based system, has its merits. From a country, which was reeling under rampant illiteracy and drought, a century ago, our education system has created one of the world's largest talented pool of engineers and other professionals.
A standardized, process-oriented system is somewhat like a factory -- it can mass produce, acceptable quality outputs at acceptable costs.
A process-based system also goes through several phases of evolution. Anyone familiar with capability maturity models can recognize the following phases of process maturity: Ad hoc, repeatable, measured and optimized.
Early systems of pedagogic processes were ad hoc in nature -- typically dependent on coherence between the teacher and the pupil. But as education needs increased, ad hoc processes became no longer feasible. We needed to scale, and for this we needed processes that gave satisfactory results and were repeatable and replicable in several places. Once, a repeatable process is in place, it is not long before we start calibrating and measuring different elements of the process. And once we have metrics, we can optimize on different elements and produce the best possible output for a given input.
However, process maturity does not stop at optimality -- and this is often overlooked. Unbridled optimization invariably eventually leads to overfitting. Overfitting is a state where the system becomes hypersensitive to metrics and starts reacting to, and trying to optimize on every measurable element.
Somewhat similar concerns must have prompted Andy Grove to say, "Every successful organization contains the seeds of its own destruction."
Success breeds overfitting. And this is not just with education. Consider media overfitting to TRP ratings, and you see what I mean. Overfitting gives specific narrowly defined metrics, much more importance than the underlying spirit of the activity.
The solution to overfitting, is to shift the focus from processes to principles. We need to shift the focus from asking how to conduct a course, to what is the course meant to achieve.
*~*~*~*~*~*
This kind of background reasoning led me to experiment with a new kind of learning model, after having a teaching experience of over 8 years, in the conventional model of lectures, tutorials, assignments and exams.
This model is what I call: Mandate-oriented learning. I have been implementing it for the past three years and it has evolved over time with greater clarity.
The underlying assumptions behind the model are the following:
- This model is applicable for a higher-education setting, where the teacher can assume a reasonable level of maturity and background proficiency from the students
- This model assumes that the students are genuinely interested in learning the subject, and have taken the course out of their own free will, with a desire to learn, and not out of peer-pressure or any other extraneous motivations
The class as a whole (including the teacher) works towards fulfilling a given mandate. The teacher contributes to the mandate by providing lectures, facilitating discussions, etc. Students contribute to the mandate in several ways -- by offering talks themselves, writing essays, leading and contributing to forum discussions, contributing to the classroom glossary, etc.
Each mandate runs for about 4-5 weeks. At the end of this time, the class reflects back on the mandate by summarizing what has been learned. The teacher then does a random sampling from the class to assess how the class as a whole, fares with respect to the mandate. If the class as a whole is found to be satisfactory with respect to the mandate, the current mandate is closed and the course moves to the next mandate.
And that is the overall idea. There is no individual-level quiz or exam at the end of a mandate. The teacher does ask questions -- but anyone could be asked. So the class as a whole needs to be prepared for the end-of-mandate session.
Grading is based on the evidence shown by the students (on an online platform like Moodle) towards their contribution to the mandate. The more evidence that students show that they are helping the class meet the mandate, the better it is for their individual grades.
At the end of the course, typically we have a (summer/winter) break for a month. The last mandate in the course is a reflective mandate, and is open through the break. Students can use this time to reflect back on the course and try and implement something based on what they have learned through the months, and contribute it back to the rest of the class.
The design objectives behind the above model are as follows:
- View learning as a collaborative activity, rather than as a competitive sport. The entire class is assessed against a mandate and students have to help one another to ensure that the entire class makes it to the next mandate
- Encourage reflection and assimilation. Usually, in conventional coursework, stress builds up to paranoid levels by the time of exams and right after exams, students tend to unwind completely and forget everything that they learned. The spillover mandate helps in reflecting back on their course after the semester is over, so that the learning lingers on
- Allow the student to discover and develop their unique mode of learning. Not all students are equally good at answering well-defined questions, and not all of them are equally good at implementing something. Learning a subject involves approaching the subject from our own unique vantage point.
This post here is written to introduce the background thinking behind the mandate model and to solicit comments and suggestions on how we can improve it.