For as long as I can remember, I have always been a very pragmatic person, feeling great discomfort when not working towards a greater goal.

The terminal goal of any sentient rational being should be to happy. That’s how happiness is defined.

And what makes me happiest the most is helping others, having a positive impact in this world. And, naturally, the biggest ones come from bettering societal systems. Given their reach, even slight improvements are impactful.

The one I found the biggest opportunity in improvement was the education system. I’ve always disliked it, primarily for its inflexibility but also inefficiency, lack of quality, among others.

This blog post is my attempt at consolidating ideas I have been having throughout my life, it describes not only a better School but a better Education System.

Surprisingly, some of my ideas have been implemented. Geographically, the closest example being IAL. However, they still have to follow MEC, and, thus, cannot be a perfect implementation of my ideas.

Specification

Firstly, what is an Education System’s primary goal?

To educate, to give people knowledge. Students lack a certain knowledge, you give them a lecture about it, they pay attention, they study, they learn.

Lecture

A Lecture’s goal, like that of the general Education System, is to give the student knowledge. It only differs in scope, where it focuses on a single thing.

Completing (watching, more generally, experiencing) a Lecture should be synonymous to having a certain Knowledge.

In its most basic form, it achieves it’s goal by transferring knowledge

• from the teacher, who is more knowledgeable
• to the student, who lacks knowledge

sequenceDiagram
  participant Student

  box provided by
Education System
    participant Teacher
  end

  Note over Student: lacks knowledge
  Note over Teacher: possesses knowledge

  Student-->>Teacher: Attention
  Teacher->>Student: Knowledge

  Note over Teacher,Student: possesses knowledge

(warning: highly unscientific)

Project-based Learning seems to be the most popular (revolutionaraly) for its soft skill development benefits.

However, for my goals (mostly quality), it is impractical.

As such, I currently lean towards one-to-many.

Quality

To ensure optimal quality, classes should be scripted, pre-recoded and edited. (If not possible, they should, at least, roughly follow some kind of agenda, practiced by the Teacher beforehand.)

Time is one of the most, if not THE most, important resource available to us humans.

As such, learning speed is paramount.

Lectures should be as intuitive as possible, to minimize time wasted re-explaining. No one wants to sit through a 3 hour long unedited lecture, specially one that could have been condensed down to a 10 minute video or read.

Because of that, lectures must be dense, they should convey as much information as possible, as quickly as possible.

Flexible Duration

Duration can vary. Although you can aim for a certain value, it is fundamentally unpredictable, and should be treated as such.

• More complex concepts take longer to explain.
• More basic concepts take shorter to explain.
• More intelligent students grasp concepts faster.

To fix a lecture’s duration is to downplay it’s value.

Progress

Traditionally, progress is measured in the unit of an entire year, where you either pass or fail.

I find that absolutely ridiculous, there is no world in which you absolutely master or completely fail a whole year.

Instead, progress should be done on the smallest scale possible. Optimally, a single lecture.

Failing

Sometimes, students can’t grasp a concept. And that’s totally fine! It just means they need more time and another chance.

Recorded lectures, so they can be re-viewed by the student, and/or additional material, such as personalized special classes with the teachers, should be provided.

Grading

As failing doesn’t mean you might have to re-do an entire year of your life, and lose all your friends, there won’t be as big of a pressure to get good grades, opposed to the traditional education system. Instead, you might only have to study a bit more, commit a bit more, go to a few special lessons and re-take the test, all while being constantly motivated by teachers and students.

As a result, there won’t be as big of a reason to cheat on a test. Cheating only means your grade will be worse as a proxy to your knowledge, which only negatively affects you.

Because of that, tests won’t need to be monitored, as students do that by themselves.

Tests should also be as efficient as possible, having quick to answer questions that accurately test for a specific knowledge or skill.

More experimentally, LLMs could also be used, providing faster, more thorough and cheaper feedback, at the cost of a small risk of hallucination. I only recommend this for online implementations, as it seems antisocial to use them in a physical setting.

Personal Development

Thankfully, Personal Development is already something worked upon by the traditional education system. However, I might have slightly different ideas on how to do them.

These play in nicely with Grading, and have been partially explained there. Because of this, I won’t be going as in-depth here.

Autonomy

Autonomy allows one to much more directly reach their goals, without the need of help from others.

Autonomous students are also great for the school, as they need much less care and resources.

Students should always be encouraged to study, by themselves, beyond what the school gives. With their extra knowledge being integrated into classes.

In topics in which the student is already knowledgeable, they can aid the professor or even direct the class, which helps in the development of soft skills.

Culture

Culture is very powerful, and is a power that the traditional education system does not harvest.

I believe that cheating and scorekeeping behavior, among other things, shouldn’t be normalized.

I have very good arguments against both:

• for cheating (although this should have been basically solved by Grading):

  • “It’ll make you know less how much you actually know.”

• for scorekeeping behaviour:

  • “It’ll lead to a worse experience for all.”

  • “An eye for an eye will leave everyone blind.”

Ethics

Ethics allows for a better experience of everyone involved, by each participant optimizing not their own individual enjoyment, but instead, the collectives.

Some believe ethics is subjective, and while I agree some parts of it could be considered as such, I believe there is a clear and objective way to be ethical and beyond. And that is, to always consider what is your goal.

For now, it can be “make future me proud of my current actions”, which leads to a myriad of repercussions:

• Generally, teaching others is something to be very proud of, so, making sure that those around you take the most out of every interaction becomes key.
• A lot more time is spent being productive instead of wasted deciding what to do next or in unimportant activities.

Grouping

Traditionally, students are grouped by chronological age, which, honestly, makes absolutely no sense. Not only do different people have different knowledge and learning speeds, but also for each individual subject.

As such, as has been done with progress, it should be done in the smallest scale possible. Optimally, a single lecture.

But what about socialization? How will they form a group identity?

Students that share the same interests, not necessarily the same age, will naturally bond together and form groups. However, they will be much more flexible, varied and organic than traditional imposed ones.

Implementation

My goal with this implementation is to provide aid to those that would benefit from learning at their own pace but don’t have the resources for homeschooling.

My ideas may be too revolutionary to be implemented as a school in any county, as they require a fundamental redesign of education itself.

However, it does not necessarily need to be implemented as a school.

If I actually invest blood, sweat and tears into this project, it will probably be in the format of a section of this exact website. Freely available to anyone interested. But not just free as in free beer, free as in freedom!

Making it an online-first experience has a few pros and cons:

Pros

• It can reach a much wider audience. You don’t need to live geographically close to a specific place, you only need a device (even a cellphone should be enough) and access to the internet. (which 92.5% of Brazilians had in 2023)
• Content creation is much easier. You only need to record a single video, that can then be watched by millions of people.
• Teachers can work from the comfort of their house, instead of having to go every day to the school.

Cons

• It is much harder to have the same kind of reciprocal learning that IAL might have, as students experience lectures at a different time then the teachers.
• Lectures are going to be much less personalized.

However, local (or even an official) groups with students and teachers can be used to send feedback as well as answer questions.

In the end, I choose an online-first approach because I can have a much bigger impact:

• This is a single person effort, and, using this system, I need to perform lectures only once, with them having the potential to have a much wider effect then otherwise.

• IAL, as long as many others, already exist. This way, I can reach a much less explored audience, one that is in higher demand for a better education system.

A hybrid approach might have been better, but I wouldn’t be able to do it alone.

Specification

Having Specification (Outer) and Specification inside Implementation (Inner) might be confusing, but:

• Outer specifies my ideas for how any Education System could be better.
• Inner specifies how I’d implement it (Implementation).

The Project would be available as a section of this website.

Lecture

Lectures are identified by a number, which grows sequentially in accordance to chronological time.

A Lecture can have any number of Lectures as it’s dependencies, which means it’ll assume the student already knows those things / has already watched those lectures.

Lectures can only depend upon Lectures that came before it, as to maintain a directed acyclic graph topology.

Lectures can have multiple immutable versions. Only the latest of which needs to be available.

Each version has a written script and, optionally, a recording of such.

The Lecture must be available in written (script) and/or video (recording) form.

Feedback

Feedback will be implemented via a comment system, similar to Reddit, where students and teachers can ask and answer questions. (I still haven’t decided how to do the ranking just yet)

Grading

Tests are collections of multiple choice questions. (with “I don’t know” as an answer, thanks IAL)

Multiple Choice Questions were chosen because:

• they provide more instant feedback

• are automatable, teachers can focus their energy more into creating new lectures instead of grading

Each question should test for a specific knowledge, that way:

• it is easier (more automatable) to know where the student needs to improve. (if they answer wrongly, it’s what the question was testing for)

• you minimize the number of questions the student has to go through

Although, in specific situations, it might be preferable to have discursive answers. Those would have to be graded by LLMs.

Fun Additions

Lecture Visualization

It should be possible to visualize all Lectures as a directed acyclic graph.

Example

---
title: Lecture Dependency DAG
---
graph LR
    subgraph Math["1 - What is Math?"]
        Abstraction(Abstraction)
    end

    subgraph Counting["2 - Numbers (Counting)"]
        Collection("Collection
(Tuple / List / Group / Set)")
        Succession(Succession)
    end

    subgraph Operators["3 - Basic Binary Operators"]
        Operator(Operator)

        Function(Function) -->
        Inversion(Inversion) & Repetition(Repetition)

        Succession(Succession) -- Repetition -->
        Addition(Addition) -- Repetition -->
        Multiplication(Multiplication)

        Succession(Succession) -- Inversion --> Predecession(Predecession)
        Predecession(Predecession) -- Inversion --> Succession(Succession)

        Addition(Addition) -- Inversion --> Subtraction(Subtraction)
        Subtraction(Subtraction) -- Inversion --> Addition(Addition)

        Multiplication(Multiplication) -- Inversion --> Division(Division)
        Division(Division) -- Inversion --> Multiplication(Multiplication)
    end

    Math --> Counting --> Operators

Numbered Rectangles are Lectures.

Rounded Rectangles are topics taught in those lectures.

Edges represent dependencies (Lecture A -> Lecture B means you need to experience Lecture A before Lecture B)

This could also incentivize students by visually showing their progress.

State of the Art

Although I should have done this first, I didn’t put a lot of research into the topic before starting, and plan on expanding this section later.

There exists a class of software called Learning Management Systems that are worth looking into. Here is a cool list a found.

My ideas sound extremely similar to mathacademy, although I only learnt about it very far into the writing process. neetcode is also similar.

Mastery Learning sounds a lot like what I’m trying to do and Bloom’s 2 sigma problem is absolutely shocking.