ABOUT THE PROJECT
In a world where artificial intelligence, coding, and machine learning are rapidly shaping everyday life, this project was created to help primary and junior students explore these complex concepts in a meaningful, hands-on, and memorable way.
At the heart of the project is the M.E.N.A.C.E. Machine — an early form of machine learning originally designed to play Tic Tac Toe using simple physical components like beads and matchboxes. Despite having no computer or digital programming, the machine is capable of “learning” through reinforcement and probability, making it a powerful and accessible introduction to the foundations of artificial intelligence.
Our students have already explored the M.E.N.A.C.E. concept through classroom prototypes built from handmade matchboxes and condiment containers. While these early versions sparked tremendous curiosity and excitement, they also revealed the challenges of coding and machine learning in a very real way. Beads spilled, “code” was corrupted, and entire games were ruined by small mistakes — providing students with authentic lessons about precision, persistence, debugging, and problem solving.
This grant seeks to transform those early experiments into something truly extraordinary.
Working alongside a community partner from the skilled trades, students will help design and build a large-scale wooden M.E.N.A.C.E. Machine that is both visually striking and fully functional. The completed installation will serve not only as a playable Tic Tac Toe machine, but also as a living demonstration of reinforcement learning and probability in action. Over time, students will “train” the machine until its beaded code consistently learns to tie or defeat human opponents.
The project is designed to extend far beyond a single classroom. Students will have opportunities to invite classes from across the school to experience the machine firsthand and may even travel to other schools to share their learning and demonstrate how machine learning can be understood through creativity, collaboration, and play.
To deepen student understanding, the project will also bring in a community expert with a background in mathematics, coding, artificial intelligence, or computer science — ideally from a post-secondary institution such as the University of Toronto. This guest speaker will help demystify topics like reinforcement learning, coding, and AI, showing students that these fields are accessible, exciting, and filled with opportunities for future innovation.
Ultimately, the purpose of this project is not simply to build a machine. It is to inspire young learners to think critically, experiment boldly, embrace mistakes as part of learning, and recognize that the technologies shaping the future are things they, too, can understand and create.
At the heart of the project is the M.E.N.A.C.E. Machine — an early form of machine learning originally designed to play Tic Tac Toe using simple physical components like beads and matchboxes. Despite having no computer or digital programming, the machine is capable of “learning” through reinforcement and probability, making it a powerful and accessible introduction to the foundations of artificial intelligence.
Our students have already explored the M.E.N.A.C.E. concept through classroom prototypes built from handmade matchboxes and condiment containers. While these early versions sparked tremendous curiosity and excitement, they also revealed the challenges of coding and machine learning in a very real way. Beads spilled, “code” was corrupted, and entire games were ruined by small mistakes — providing students with authentic lessons about precision, persistence, debugging, and problem solving.
This grant seeks to transform those early experiments into something truly extraordinary.
Working alongside a community partner from the skilled trades, students will help design and build a large-scale wooden M.E.N.A.C.E. Machine that is both visually striking and fully functional. The completed installation will serve not only as a playable Tic Tac Toe machine, but also as a living demonstration of reinforcement learning and probability in action. Over time, students will “train” the machine until its beaded code consistently learns to tie or defeat human opponents.
The project is designed to extend far beyond a single classroom. Students will have opportunities to invite classes from across the school to experience the machine firsthand and may even travel to other schools to share their learning and demonstrate how machine learning can be understood through creativity, collaboration, and play.
To deepen student understanding, the project will also bring in a community expert with a background in mathematics, coding, artificial intelligence, or computer science — ideally from a post-secondary institution such as the University of Toronto. This guest speaker will help demystify topics like reinforcement learning, coding, and AI, showing students that these fields are accessible, exciting, and filled with opportunities for future innovation.
Ultimately, the purpose of this project is not simply to build a machine. It is to inspire young learners to think critically, experiment boldly, embrace mistakes as part of learning, and recognize that the technologies shaping the future are things they, too, can understand and create.
| grant_proposal.pdf |
HOW M.E.N.A.C.E. WORKS
|
|
|
MENACE is a machine
The word MENACE is an acronym.
The M stands for Matchbox/Machine.
The E stands for educable which means capable of being educated!
The N, A, and C stands for Noughts and Crosses which is a synonym for Tic Tac Toe.
Lastly, the E stands for engine which is a synonym in this case for computer.
MENACE operates through 304 containers. Each container has a label on it that represents 1 of the 304 possible game states or scenarios that one faces in a game of Tic Tac Toe.
Here is how it works!
The computer plays first and uses the “O” symbol.
A human assists the computer. First, the human finds and opens the container that has a label showing an empty Tic Tac Toe board. This container is always the first to be used!
A coloured bead is randomly taken from the container. The colour of the random bead represents one of 9 places where an O can be placed within the Tic Tac Toe board.
The human assisting the MENACE computer records the O symbol matching the place on the game board.
Then the opponent represented by the X symbol decides where to place their symbol.
It’s MENACE’s turn now. The container with the current game state sticker is found and is opened and another bead is randomly selected telling the human helper where to put the O symbol.
This back and forth goes on and on until a winner is determined.
After the game has been completed, if the computer lost the game, one of every coloured bead that led to that loss is removed from all of those containers. We are consequencing the computer for its randomly drawn beads leading to a loss.
If the computer draws with its opponent, 1 of every coloured bead that led to that tie is added. Tie’s are all right! They are a slightly better outcome. We’re rewarding MENACE slightly.
Lastly if the MENACE machine wins a game, 3 of every coloured bead that led to that win is added. This is known as reinforcement learning!
At first, randomness determines MENACE’s choices. But over time, MENACE’s randomness slowly disappears due to reinforcement learning. Eventually, the likelihood of a win or tie will become certain because only beads leading to wins or ties will remain within the containers waiting to be drawn.
So little coders, let’s put it all together!
MENACE parallels with the basic computer coding that you are currently learning in school!
Imagine you're teaching your computer friend to play Tic Tac Toe. At first, your computer friend might not know much about the game, but you want it to learn and get better each time it plays.
You start by telling your computer friend the rules of Tic Tac Toe: "To win, you need three X's or O's in a row, like this!" You show it how the game works.
Now, here comes the cool part! You want your computer friend to remember which moves are good and which ones aren't so great. You create a special list, kind of like a magical notebook, where your computer friend writes down what happens during each game.
For example, if it tries a move and wins, the computer friend writes, "Yay! This move worked!" If it tries a move and loses, it writes, "Oops, this move wasn't the best."
Over time, your computer friend looks at its magical notebook and figures out the best moves to make. It learns from its wins and losses, just like you get better at games by playing them.
The code is like the set of instructions you give your computer friend to read its magical notebook and choose the best moves. It's like saying, "Hey, if you see three X's in a row, that's a good move!"
And there you have it – your Tic Tac Toe game learns over time how to tie or even beat a human. It's like having a little friend that gets smarter and smarter with each game you play together!
The word MENACE is an acronym.
The M stands for Matchbox/Machine.
The E stands for educable which means capable of being educated!
The N, A, and C stands for Noughts and Crosses which is a synonym for Tic Tac Toe.
Lastly, the E stands for engine which is a synonym in this case for computer.
MENACE operates through 304 containers. Each container has a label on it that represents 1 of the 304 possible game states or scenarios that one faces in a game of Tic Tac Toe.
Here is how it works!
The computer plays first and uses the “O” symbol.
A human assists the computer. First, the human finds and opens the container that has a label showing an empty Tic Tac Toe board. This container is always the first to be used!
A coloured bead is randomly taken from the container. The colour of the random bead represents one of 9 places where an O can be placed within the Tic Tac Toe board.
The human assisting the MENACE computer records the O symbol matching the place on the game board.
Then the opponent represented by the X symbol decides where to place their symbol.
It’s MENACE’s turn now. The container with the current game state sticker is found and is opened and another bead is randomly selected telling the human helper where to put the O symbol.
This back and forth goes on and on until a winner is determined.
After the game has been completed, if the computer lost the game, one of every coloured bead that led to that loss is removed from all of those containers. We are consequencing the computer for its randomly drawn beads leading to a loss.
If the computer draws with its opponent, 1 of every coloured bead that led to that tie is added. Tie’s are all right! They are a slightly better outcome. We’re rewarding MENACE slightly.
Lastly if the MENACE machine wins a game, 3 of every coloured bead that led to that win is added. This is known as reinforcement learning!
At first, randomness determines MENACE’s choices. But over time, MENACE’s randomness slowly disappears due to reinforcement learning. Eventually, the likelihood of a win or tie will become certain because only beads leading to wins or ties will remain within the containers waiting to be drawn.
So little coders, let’s put it all together!
MENACE parallels with the basic computer coding that you are currently learning in school!
Imagine you're teaching your computer friend to play Tic Tac Toe. At first, your computer friend might not know much about the game, but you want it to learn and get better each time it plays.
You start by telling your computer friend the rules of Tic Tac Toe: "To win, you need three X's or O's in a row, like this!" You show it how the game works.
Now, here comes the cool part! You want your computer friend to remember which moves are good and which ones aren't so great. You create a special list, kind of like a magical notebook, where your computer friend writes down what happens during each game.
For example, if it tries a move and wins, the computer friend writes, "Yay! This move worked!" If it tries a move and loses, it writes, "Oops, this move wasn't the best."
Over time, your computer friend looks at its magical notebook and figures out the best moves to make. It learns from its wins and losses, just like you get better at games by playing them.
The code is like the set of instructions you give your computer friend to read its magical notebook and choose the best moves. It's like saying, "Hey, if you see three X's in a row, that's a good move!"
And there you have it – your Tic Tac Toe game learns over time how to tie or even beat a human. It's like having a little friend that gets smarter and smarter with each game you play together!
HOW MACHINES LIKE CHATGPT LEARN THROUGH REINFORCEMENT
How Large Language Models Work
Imagine a giant library filled with millions of books, articles, and conversations. A large language model (like ChatGPT) is trained by reading through this library—except instead of reading with eyes, it processes words using math and patterns.
Feeding the Machine Information – Scientists and engineers give the machine a huge amount of text from books, websites, and conversations. This helps it learn how people write and talk.
Breaking It Down into Patterns – Instead of remembering entire books, the machine looks for patterns in words and sentences. For example, it might notice that the phrase “How are you?” is often followed by “I’m good, thanks!”
Storing What It Learns – The machine doesn’t store exact sentences like a notebook. Instead, it remembers connections between words, so it can predict what words might come next when answering a question.
Feeding the Machine Information – Scientists and engineers give the machine a huge amount of text from books, websites, and conversations. This helps it learn how people write and talk.
Breaking It Down into Patterns – Instead of remembering entire books, the machine looks for patterns in words and sentences. For example, it might notice that the phrase “How are you?” is often followed by “I’m good, thanks!”
Storing What It Learns – The machine doesn’t store exact sentences like a notebook. Instead, it remembers connections between words, so it can predict what words might come next when answering a question.
How Reinforcement Learning Improves It
Now that the machine has read all this information, it needs help figuring out which answers are best. That’s where reinforcement learning comes in!
The Machine Starts Knowing Nothing – Just like a beginner learning a game, the machine starts by making guesses when answering questions.
It Tries and Makes Mistakes – If it gives a wrong answer, it learns from that mistake. Think of it like playing tic-tac-toe: If you lose by making a bad move, you won’t make that move again next time.
Humans Give Feedback – Trainers (real people) look at the machine’s answers and give thumbs-up for good ones and thumbs-down for bad ones. This teaches the machine what works and what doesn’t.
It Repeats and Gets Better – The more it practices and gets feedback, the better it becomes. Over time, it learns to choose answers that people like the most.
It Uses Patterns to Predict Answers – After millions of examples, the machine gets really good at guessing what words should come next in a sentence. That’s why it can answer questions in a way that sounds natural!
It’s Always Learning – Even after training, the model can improve. New words, phrases, and ideas keep getting added, so it stays up to date.
Just like your Menace Machine learns to play tic-tac-toe by rewarding good moves and avoiding bad ones, ChatGPT learns the best answers through practice, mistakes, and feedback.
The Machine Starts Knowing Nothing – Just like a beginner learning a game, the machine starts by making guesses when answering questions.
It Tries and Makes Mistakes – If it gives a wrong answer, it learns from that mistake. Think of it like playing tic-tac-toe: If you lose by making a bad move, you won’t make that move again next time.
Humans Give Feedback – Trainers (real people) look at the machine’s answers and give thumbs-up for good ones and thumbs-down for bad ones. This teaches the machine what works and what doesn’t.
It Repeats and Gets Better – The more it practices and gets feedback, the better it becomes. Over time, it learns to choose answers that people like the most.
It Uses Patterns to Predict Answers – After millions of examples, the machine gets really good at guessing what words should come next in a sentence. That’s why it can answer questions in a way that sounds natural!
It’s Always Learning – Even after training, the model can improve. New words, phrases, and ideas keep getting added, so it stays up to date.
Just like your Menace Machine learns to play tic-tac-toe by rewarding good moves and avoiding bad ones, ChatGPT learns the best answers through practice, mistakes, and feedback.
| menace_explanation_script.docx |
SUPPORTING LINKS/DOCUMENTS
https://www.drooker.ca/coding.html (Password: 4321)
https://www.drooker.ca/dataprobability.html (Password: 4321)
| grant_proposal.pdf |
| menace_machine_grant_summary.docx |
| boxes0.pdf |
| boxes1.pdf |
| boxes2.pdf |
| boxes3.pdf |
| enlarged_colour_chart.pdf |
| menace_colour_guide.pdf |
| menace_guiding_questions.docx |
| tic_tac_toe_strategy.docx |
GALLERY
(Expressed written consent has been obtained for all posted images and videos)
PROJECT SUMMARY
We facilitated an inquiry-based project centered on the construction and programming of a physical MENACE (Matchbox Educable Noughts and Crosses Engine) Machine—an analog model of reinforcement learning designed to demystify artificial intelligence for young learners. With the support of a community partnership through NewMakeit, a local makerspace in Newmarket, Ontario, students engaged in an ambitious, hands-on initiative that fused woodworking, coding, and AI theory into a single meaningful experience.
Participate
Students took part in a full-cycle design and build process to create a wooden MENACE Machine installation. With mentorship from tradespeople at NewMakeit, they used tools such as laser cutters and woodworking techniques to bring their analog computer to life. Following the build, students coded the machine using color-coded beads to simulate the decision-making logic of AI reinforcement learning. Once completed, students wheeled the Menace Machine from classroom to classroom throughout the school, facilitating live demonstrations and discussions on the connection between historical analog computing and
modern-day AI.
Reflect
As students deepened their understanding of AI through the MENACE Machine, they engaged in thoughtful reflections at multiple stages. They evaluated the accuracy of their bead-based coding by testing game outcomes and troubleshooting errors. A significant part of the reflection focused on the labeling of the machine’s 304 unique game states, exploring how some configurations were mathematically equivalent due to rotational and reflective symmetries in tic-tac-toe. This led to rich discussions about optimization, efficiency, and pattern recognition—core aspects of AI logic and real-world computing.
Apply
Students applied their knowledge by presenting the functioning MENACE Machine to peers and educators, using it as a physical proof-of-concept to explain abstract concepts like reinforcement learning. They fielded questions, demonstrated the machine’s operation, and articulated how a mechanical system like MENACE parallels more complex AI systems, including large language models. Their ability to translate foundational learning into confident public presentation demonstrated deep understanding, technical skill, and metacognitive growth.
In Summary
This project enabled students to engage deeply with artificial intelligence, not just through digital abstraction, but through tangible, hands-on experience. They came to appreciate the crucial interplay between trades and technology: coding met carpentry, and AI theory became embedded in laser-cut wood and beads. By constructing, programming, and presenting their MENACE Machine, students gained meaningful insight into how human problem-solving, creativity, and design remain at the heart of even our most advanced technologies.
Participate
Students took part in a full-cycle design and build process to create a wooden MENACE Machine installation. With mentorship from tradespeople at NewMakeit, they used tools such as laser cutters and woodworking techniques to bring their analog computer to life. Following the build, students coded the machine using color-coded beads to simulate the decision-making logic of AI reinforcement learning. Once completed, students wheeled the Menace Machine from classroom to classroom throughout the school, facilitating live demonstrations and discussions on the connection between historical analog computing and
modern-day AI.
Reflect
As students deepened their understanding of AI through the MENACE Machine, they engaged in thoughtful reflections at multiple stages. They evaluated the accuracy of their bead-based coding by testing game outcomes and troubleshooting errors. A significant part of the reflection focused on the labeling of the machine’s 304 unique game states, exploring how some configurations were mathematically equivalent due to rotational and reflective symmetries in tic-tac-toe. This led to rich discussions about optimization, efficiency, and pattern recognition—core aspects of AI logic and real-world computing.
Apply
Students applied their knowledge by presenting the functioning MENACE Machine to peers and educators, using it as a physical proof-of-concept to explain abstract concepts like reinforcement learning. They fielded questions, demonstrated the machine’s operation, and articulated how a mechanical system like MENACE parallels more complex AI systems, including large language models. Their ability to translate foundational learning into confident public presentation demonstrated deep understanding, technical skill, and metacognitive growth.
In Summary
This project enabled students to engage deeply with artificial intelligence, not just through digital abstraction, but through tangible, hands-on experience. They came to appreciate the crucial interplay between trades and technology: coding met carpentry, and AI theory became embedded in laser-cut wood and beads. By constructing, programming, and presenting their MENACE Machine, students gained meaningful insight into how human problem-solving, creativity, and design remain at the heart of even our most advanced technologies.
| menace_machine_grant_summary.docx |