Education is one of those topics, which everyone has an opinion on, and reforms are proposed every other day. In Germany it is even worse, as education is state’s jurisdiction, so multiply the mess by 16. So, is this just another proposal which claims to be much better than any others proposed before?
I would definitely have an opinion to offer. But writing down this opinion is fruitless if I can not substantiate my claims. And this is fundamentally the problem of every proposal. No one has any real evidence. Where would you even get it from? Sure, there are a couple of pilot projects from time to time, where volunteering schools try out a new education model. But even those are not very reliable. Why? Because they volunteered.
You see, often the difference between an awkward theatrical performance and a comedy is the confidence with which the actors sell it. And it is painfully obvious if a teacher heavy handedly tries to implements the latest innovation taught in the last seminar they attended. I remember the time when we suddenly had to draw a mind map as an introduction to the topic in every class. So if all pilot projects are conducted by volunteers (i.e. people who believe in the new method) you are going to get much better results than you will get in practice, when the unmotivated teacher next door has to implement them.
So one of the biggest problems in education is the lack of data. Additionally you can not make arbitrary experiments out of ethical concerns. You are playing with the future of the kids you are experimenting with after all.
Obtaining Data with Positive Side Effects
In my school there was a hallway decorated with quotes. One of them was roughly
a teacher helps you solve problems you wouldn’t have without them
this harmless joke has a true and problematic core: The teacher is the one, who designs the examination and ultimately grades the students. This is problematic for multiple reasons:
a lot of students will try to sabotage lessons by distracting the teacher, since material not covered will not be part of the exam. From a top down view they are of course sabotaging themselves as they prevent themselves from learning. But people are generally not patient enough to look that far ahead. So their objectives are grades. And sabotage is a shortcut to better grades.
exams posed by teachers are not comparable. This is ultimately unfair to the students, as anyone looking at the grade will have no idea what that grade actually means. One example from my own experience:
We had to choose between Latin and french in 6th grade. Latin was the choice of anyone who didn’t really want to learn another language as it promised to be a little bit of history and less vocabulary. This of course meant that our Latin class was atrocious from a language standpoint. To keep grades in a unsuspicious range, exams were ultimately translations of texts we already translated in class, with tons of vocabulary provided in a table on the side.
To prevent people from simply memorizing the entire text, some sentences were dropped from the original text. But using the words provided and remembering the story from class, you could reliably puzzle together the translation without really knowing any Latin at all.
So now I have a “good” Latin grade. But does that mean I am actually good at it? Of course not.
bad teachers can simply cover up their subpar teaching skills with easier exams.
In the last two years of school, the relation between teacher and students greatly improved. Why? Well because we were preparing for the final examination. An examination which was provided from a central authority - well… central to the state within Germany. Suddenly the teacher was the coach, helping you pass the exam instead of causing your problem. And this flipped the teacher/student relationship on its head.
Central Examinations
There is no reason not to do central examinations. The cost of mailing around exams is negligible at this point. Simply randomly shuffling exams around the country
- improves the teacher student relationship
- makes grades more comparable
- provides data on the effectiveness of teachers and their teaching methods
- removes a ton of bias against/for students by virtue of anonymity
- dampens the tendency for grade inflation over time
The only valid concern I heard so far, was that central examinations remove the flexibility teachers have with the syllabus. Since there is value in teachers adapting the syllabus around children’s interest. And nothing would be more unfortunate than a teacher shooting down curious questions because they can not afford the distraction from the syllabus.
But there is a relatively simply fix for that: Provide a menu of exercises in the exams so that students can pick an area they are strong in/where the teacher focused their lessons on.
A potential problem with that approach is, that students might always pick the same topic (e.g. always pick probability in mathematics and never learn the basics of algebra and analysis).
Points not Grades
To avoid this problem you could fundamentally redesign exams. What if you did not have a maths exam, but a probability, analysis and algebra exam and you would only have to take each of them at some point (in no particular order)? Finishing school would then not mean finishing a particular year, but rather achieving a minimum score in all required exams. You could have a progression of increasingly difficult exams in one subject (probability I, probability II,…) or you could be even more radical and have everyone write the same exam with a selection of exercises of various difficulty. Beginners would do the easier exercises at the beginning, while advanced students would do the more difficult exercises with more points.
There is an important side effect of this approach. You can compare a students performance to their past performance more easily as they would be retaking “the same” test. This has multiple benefits:
in general every retake will result in a new high score. While grades measure where you should be by now, scores on such a gradual test would reflect where you are. And it is much more motivating to achieve a new high score than to just “stay on the same grade” year over year.
As students will have to achieve a certain threshold to finish school, there is still pressure to reach certain milestones. And students will inevitably compare themselves to each other. But there is still a much bigger sense of progress, if your score increases every time you retake the test.
(I would bet that this is one of the reasons why sports are more popular than other subjects)
Since there is a minimum threshold of points students need to achieve in order to finish school, they will accumulate roughly this number of points over time. So what if we take the current average cost of teacher salaries per student, divide this cost per student by the total number of points, and pay teachers per point gained of their students?
At this point the interest of the student and teacher are perfectly aligned. And aligned interest result in a much more productive collaboration.
Since we now decoupled the measurement of progress from the school system, it is possible to loosen the screws on the syllabus. Teachers are free to come up with novel ways to teach their students. As they have a financial stake in the outcome, they will tend to be a bit conservative with changes - which is a good thing, as children’s futures are at stake - but if they are sure about something, they can just do it.
Of course there will be some issues at the start: Maybe gaining points is easier for beginners than for advanced students, causing a windfall for entry level teachers. Or vice versa. Another problem is, that short term point gain might not be conductive to long term learning.
This can be balanced out by schools. So instead of paying teachers directly, the schools would be payed the amount all of their teachers earned. Then the school could redistribute these earnings to the teachers. They could just pay every teacher the amount of points their students earned, but they can also introduce additional measures to balance out the highs and lows of point gains over time. Since the school has the macro view as their students stay with them for longer than a year, it will have an incentive to invent measures which cause teachers to prioritize long term learning over short term learning.
In total this will result in a rapid trial and error process, where unsuccessful strategies are quickly weeded out, ultimately resulting in much more effective education.
An Optimization Algorithm
We have designed an optimization algorithm, which optimizes the amount of points gained by students over their time in the school system. It is quite important to note that schools will optimize for points and not “learning”. And while the points try to measure “learning” progress, as soon as a measure becomes a target, it ceases to be a good measure (Goodhart’s Law).
For this reason this measure needs to be a moving target. It needs to be constantly readjusted to actually measure learning instead of “something else” which might be an artifact of the current measurement.
So the measurement should be designed by stakeholders in the outcome. E.g. Universities might desire students to have certain skills when they enter. So they are good candidates to create such measures as they will constantly see the results from students entering their halls. If they notice some skills were neglected, they can adjust the tests such that they test for these skills, which will mean the next round of optimization will put a greater emphasis on these skills.
Companies could also top up funding for certain skills, making points in these categories worth more, causing schools to focus more attention on these skills.
Why Grades are Unavoidable
Okay, so if we have Grades at all, this might be a good idea. But what if you have the opinion that Grades are detrimental to students in general? Did I not just make an opinionated suggestion contrary to what I stated I would do in the beginning?
Well, if your school believes that Grades are bad, they can delay grading students up until the last year. Then students would take all tests at once and collect all the necessary points at once. This results in a windfall for the school at the end of a students time at school.
That is a bit risky to fly blind up until the end you say? Well, it is also risky to never asses a students capability and hoping everything will turn out fine. But if you are convinced that this will improve students performance, you can do it.
Okay, but what if you wanted to have no grades at all? Well how do you expect universities to select their students or companies to select their employees? If the school isn’t doing the testing, then the universities/companies will have to do it themselves. And they will not be able to observe candidates for years. In other words, their judgement will likely be based on a single test. Great way to reduce the pressure…