Day 1668
Cursive Was My First Cipher
How a second-grade handwriting lesson led me down a cryptography rabbit hole on a Thursday night in October.
Tonight, an AI agent told me something that stuck: “LLMs are switches.”
I had asked it for a TL;DR of an article, and somewhere in its explanation, it described large language models as switches. Give them instructions, point them in a direction, and watch what happens.
I liked that analogy. The more I thought about it, the more I realized how much of learning comes down to finding the right mental switch. Sometimes an idea that seemed impossibly complicated becomes almost obvious once somebody explains it in a language your brain already understands.
Which brings me to cursive handwriting, cryptography, Bitcoin, and a rather unexpected appreciation for my second-grade teacher.
Bear with me.
My First Brush With Cryptography Was in Second Grade
I came across a post today about somebody struggling to read Angelina Jolie’s cursive handwriting. My immediate thought was that I learned cursive in second grade, and to this day, I actually find it easier to write that way.
Speaking from experience, a surprising number of people can’t read it.
That got me thinking. Cursive is almost like a primitive form of cryptography. The underlying information hasn’t changed. It’s still English, with the same letters, words, and sentences. But change how those letters are represented, and suddenly something perfectly legible to one person looks like ancient hieroglyphics to another.
Now, before a cryptographer comes after me, I understand that cursive isn’t actually encryption. There’s no secret key, no mathematical security, and certainly nothing stopping a sufficiently motivated person from figuring it out.
But as an analogy for how information can be represented differently depending on what the reader knows, I think it works beautifully.
We encounter little versions of this everywhere. Inside jokes that make absolutely no sense to outsiders. Nicknames whose origins have been lost to history. Family shorthand that sounds like complete nonsense to anyone who wasn’t there when it originated.
The information is sitting right in front of you. Whether you understand it depends on what you already know.
And that was enough to send me down the rabbit hole.
One thing I’ve been appreciating more as I get older is just how much easier learning becomes when you find an analogy that connects something unfamiliar to something you already understand.
Math, in particular, can feel painfully stale when you approach it cold. Symbols, formulas, proofs, definitions. Your eyes glaze over before you’ve even figured out why you should care.
But give me a good analogy first? Now we’re talking.
It’s like getting a little dopamine hit before climbing onto the treadmill. Once you understand why something is interesting, you’re much more willing to put in the work required to understand how it actually works.
So tonight, I asked AI for some fascinating facts about cryptography.
Three stood out.
1. Proving You Know Something Without Revealing What You Know
This one genuinely blows my mind. Imagine being able to prove you know a secret without ever telling anyone what that secret is.
That is the central idea behind a zero-knowledge proof.
The classic analogy involves something called the Ali Baba cave.
Imagine a circular cave with two paths, one going left and one going right. Deep inside, the paths connect through a magical locked door that opens only when someone speaks a secret password.
You tell me you know the password. Naturally, I’m skeptical.
But instead of asking you to reveal it, we play a game.
You enter the cave while I remain outside, unable to see which path you’ve chosen. Then I walk to the entrance and randomly shout which side I want you to emerge from.
If you know the password, no problem. You can open the secret door and move between the two paths, emerging from whichever side I request.
But if you’re bluffing, you’re stuck wherever you originally went. You’ll have a 50% chance of choosing correctly.
We repeat the experiment.
And repeat it again.
After twenty successful rounds, the probability of someone without the password passing every challenge by pure luck is approximately one in a million.
At that point, I’m pretty convinced you know the password.
And I still have absolutely no idea what it is.
How cool is that?
Another analogy I love involves Where’s Waldo?
Imagine I challenge you to find Waldo in one of those absurdly crowded illustrations. You find him, but you don’t want to reveal where he is on the page.
So you take a massive sheet of paper, cut a tiny hole in it, and position that hole directly over Waldo’s face. I can see that you’ve found him. But because the surrounding page is completely covered, I can’t tell where he was hiding.
You’ve demonstrated knowledge without giving away the information you wanted to protect. The Waldo example isn’t a complete cryptographic protocol, but it captures the underlying intuition.
And think about the implications. We live in a world where proving something about yourself often means surrendering far more information than necessary. Want to prove you’re old enough to enter a bar? Show somebody an ID containing your name, birthday, address, and potentially other personal information.
Want to prove you have sufficient funds for a transaction? Depending on the system, you might have to reveal far more about your finances than the other party actually needs to know.
But what if you could prove you’re over 21 without revealing your birthday?
What if you could demonstrate that you control sufficient funds without exposing your entire financial history?
This is where the mathematics becomes genuinely exciting. With the right cryptographic protocols and credentials, statements like these can be verified while revealing remarkably little about the underlying private information.
Of course, the details matter. Not every zero-knowledge system provides the same privacy guarantees, and the technology doesn’t automatically eliminate every possible information leak.
But the underlying principle is powerful.
You shouldn’t have to surrender everything you know simply to prove one thing is true.
As somebody who loves Bitcoin, privacy, and individual liberty, that idea resonates with me. Verification and privacy should not be enemies.
2. Two Keys That Never Have to Meet
The second concept was public-key cryptography.
This one I was already somewhat familiar with because of Bitcoin, but I realized how much I still have to learn.
Imagine you could create a special lock and key. You keep the key private, but you manufacture unlimited copies of the lock and distribute them to anybody who wants one. Someone you’ve never met can take one of those locks, place a message inside a box, lock it, and send the box to you.
Only you possess the key capable of opening it.
The remarkable part is that you never had to meet beforehand to exchange a secret password. That’s the intuition behind public-key encryption. In practice, the mathematics involves a public key and a corresponding private key. Modern secure communication systems often use public-key techniques to establish shared secrets, then use faster symmetric encryption to protect the actual conversation.
And the same broad family of mathematics gives us another extraordinary tool: digital signatures.
Instead of using your private key to unlock a message, you can use it to produce a signature that someone else can verify using your public key.
That is especially important for Bitcoin.
When I receive Bitcoin, I don’t need to know or trust the person sending it. And when I eventually spend those coins, the network doesn’t need to call my bank, verify my identity with a customer service rep, or ask some institution for permission.
I create a valid cryptographic signature. Bitcoin’s consensus rules determine whether the transaction is authorized. The mathematics does the heavy lifting.
Obviously, there is more to Bitcoin than signatures. The network still needs consensus rules, proof of work, and a mechanism for determining which coins have already been spent.
Get Chep’s stories in your inbox
Join Medium for free to get updates from this writer.
But the ability to prove authorization without revealing a private key is fundamental to the entire system. And I think that’s incredible. Consider how much human coordination traditionally depended on intermediaries, personal relationships, institutional trust, and physical proximity.
Now we have mathematical systems that allow two strangers on opposite sides of the planet to exchange and verify information without ever having met.
For much of human history, that would have sounded like sorcery. Today, we use versions of this technology every time we interact with secure websites, authenticate software, or transact on Bitcoin.
There is also a lesson here that I to like to remind myself of often: Your private key isn’t merely another password. It’s the cryptographic authority to spend your money.
Lose control of it, and you can lose control of your Bitcoin. Lose access to it without a backup or alternative recovery mechanism, and there may be nobody who can restore that access.
No password reset button. No manager to escalate your ticket to. No institution that can simply reverse the mathematics.
That responsibility is part of what makes self-custody so extraordinary. Freedom and responsibility tend to travel together.
3. The Unbreakable Cipher That Humans Made Impractical
The third idea might be my favorite purely because of the irony.
There is a form of encryption called the one-time pad that, when implemented correctly, provides perfect secrecy.
Not very difficult to break.
Not would take a supercomputer a million years.
Mathematically perfect secrecy.
Imagine Alice wants to send Bob a message. They first obtain identical copies of a completely random sequence of bits, at least as long as the message itself. Alice combines her message with that randomness using a simple mathematical operation called XOR. Bob reverses the operation using his copy of the same random sequence.
To anyone intercepting the encrypted message, every possible plaintext of the appropriate length is consistent with some possible key.
Without the key, the ciphertext reveals nothing about which message was originally sent.
Even unlimited computing power doesn’t solve that information problem.
But there’s a catch.
Actually, several catches.
The key must be truly random. It must be at least as long as the message. It has to remain secret. And perhaps most importantly, it can never be reused. Break those conditions, and the mathematical guarantees can fall apart.
Now imagine wanting to send someone a thousand-page document. You’d need at least a thousand pages’ worth of perfectly random key material, securely delivered to the recipient beforehand. Want to send another thousand pages? Better have another thousand pages of randomness.
Suddenly, the encryption isn’t the difficult part. The logistics are. Human beings discovered a mathematically perfect way to hide information, only to run headfirst into the much more ordinary problem of getting the secret key to the correct person.
It’s like inventing an indestructible vault and realizing your biggest challenge is figuring out how to ship the keys without somebody stealing them.
There is something wonderfully humbling about that.
It also helps explain why public-key cryptography was such an important breakthrough. Public-key systems don’t provide the same information-theoretic secrecy as a correctly implemented one-time pad. Their security generally depends on computational assumptions.
But they help solve an enormous practical problem: allowing people to establish secure communication without distributing a giant secret key ahead of time.
Sometimes the most useful invention isn’t the theoretically strongest one. It’s the one human beings can actually deploy. That lesson extends well beyond cryptography.
A perfect system that nobody can use is often less valuable than a sufficiently secure system that billions of people can. There is a fascinating distinction between mathematics and engineering. Mathematics asks whether something can be proven. Engineering asks whether we can make it work in the physical world.
Magic happens when the two cooperate.
LLMs Are Switches. Learn How to Flip Them.
Which brings me back to the AI agent that started this whole train of thought.
“LLMs are switches.”
Strictly speaking, that’s a metaphor, not a complete description of how language models work. But I’ve been thinking about it as a useful way to understand the increasingly powerful tools we have at our fingertips. True/False, Yes/No, On/Off
You don’t need to reconstruct the intelligence inside a model every time you want it to do something. You need to learn how to direct it.
Ask a better question. Supply relevant context. Give it useful tools. Let it explore something you don’t yet understand. Challenge its answers. Ask it to explain the same concept using a different analogy.
Each interaction can reveal another way to use the machine.
And when you start connecting these tools to software, APIs, and other tools, the possibilities become even more interesting. What struck me tonight is that learning itself works in a similar fashion.
Sometimes your brain already possesses most of the information needed to understand something, but the connection between two ideas hasn’t been made.
Then somebody explains cryptography using a cave. Or a children’s puzzle book. Or a lock and key. Suddenly, an intimidating mathematical concept becomes something you can actually picture.
A switch flips.
The analogy isn’t the complete understanding, of course. At some point, you still have to learn the mathematics. The analogy gets you through the door, but it doesn’t eliminate the staircase waiting on the other side.
That’s fine with me.
I’d rather understand an idea imperfectly and become curious enough to investigate it than dismiss the entire subject because the first explanation went over my head.
And that’s what I appreciate so much about AI as a learning tool.
I can approach a subject from almost any angle, ask incredibly basic questions without embarrassment, and keep exploring until something clicks.
I can ask for the children’s version, the engineer’s version, the mathematical version, and then challenge the model to tell me where all three explanations break down.
The important thing is remembering that AI can also be confidently wrong. Understanding doesn’t come from blindly accepting whatever it says. Use the tool to sharpen your own thinking, then check what you’ve learned against reality.
Amplify your own curiosity.
The More I Learn, the More I Want to Learn
There’s a particular kind of happiness I get from understanding something today that I didn’t understand yesterday. It doesn’t need to make me money. It doesn’t need to become a business. It doesn’t even need to have an immediately obvious practical application.
Sometimes it’s enough to sit there and think, Holy shiitake mushrooms, that’s how that works?!
I think we underestimate how much joy there is in that feeling. We spend so much of our lives learning because we’re required to. To pass a test. Get a grade. Earn a degree. Qualify for a job.
But learning for the sheer pleasure of discovering something is a different animal. Nobody assigned me cryptography homework tonight. Nobody demanded that I understand zero-knowledge proofs before bed. I was simply curious, and now I understand three fascinating ideas even better.
That’s a pretty good return on an evening.
I’m also grateful that Bitcoin pulled me into this world in the first place. Before Bitcoin, I had relatively little reason to think deeply about public and private keys, digital signatures, or the mathematics that enables strangers to transact without a trusted intermediary.
Bitcoin gave those ideas real consequences. Suddenly, cryptography wasn’t some abstract field reserved for mathematicians. It was a technology that could help protect property rights and individual sovereignty.
Now AI is helping me explore the mathematics beneath it. And the deeper I go, the more I realize just how much there is left to understand. I suspect that feeling will never go away.
Ask one good question & you will often find ten more.
That’s the real gift of curiosity. The more you discover, the more doors appear. And instead of becoming discouraged by how much you don’t know, you start getting excited about how much there still is to learn.
Tonight, all of this started with somebody struggling to read Angelina Jolie’s handwriting.
A little detour through second grade. A magical cave. Waldo hiding behind a sheet of paper. Some impossible-to-break encryption. And a few thoughts about artificial intelligence.
Now here I am, grateful to God for another day of learning, for the technology that makes it so accessible, and for the curiosity to keep exploring.
And especially grateful to my second-grade teacher, who apparently had me practicing cryptographic-adjacent information encoding decades before I knew what a private key was.
Turns out those cursive worksheets were more useful than I gave them credit for. Back then, my teacher held the key and I was learning to read the cipher. Tonight it felt like I finally got handed the key to a few more.
10/15/26
Conor Jay Chepenik
