Flowers for Algernon and the Claude Code Curve

It's 1:47 AM. Nobody asked me to build this feature. My wife went to bed two hours ago. I'm not behind on a deadline. I just... can't stop.

The terminal is warm. Claude is responsive. I ask a question, I get an answer. I describe a feature, it materializes. I point at a bug, we hunt it together. The loop is frictionless and the output is real — actual software, actual commits, actual value being created at 2 AM on a Wednesday for no reason other than the fact that stopping feels worse than continuing.

I've been here before. Not with Claude. With other things. The 3 AM poker sessions in college where you're down $200 but the next hand might be the one. The Civilization V campaigns where "one more turn" ate a Saturday. The early iPhone era where every notification was a tiny Christmas morning.

I know what this is. I just didn't expect it to come wearing a terminal cursor.

Why Algernon

If you haven't read Daniel Keyes' Flowers for Algernon, the premise is simple and devastating: Charlie Gordon, a man with an IQ of 68, undergoes an experimental procedure that triples his intelligence. He goes from barely literate to genius-level cognition in weeks. He experiences the world with entirely new eyes. He can suddenly see patterns, make connections, create things he never imagined possible.

The catch — and you see it coming, which makes it worse — is that the enhancement is temporary. The mouse who received the same procedure, Algernon, deteriorates first. Charlie watches the mouse lose everything, knowing he's next.

I'm not saying Claude Code is going to disappear. I'm saying the emotional arc of adopting it mirrors Charlie's experience with unsettling precision:

Charlie's tragedy wasn't losing his intelligence. It was realizing what the intelligence cost him while he had it.

That's the question this article asks. Not "is Claude Code amazing?" — it is. But: what is the enhancement costing you while you have it?

The Three Phases of the Claude Code Curve

I've lived this. I've watched colleagues live it. I've watched it in the Discord servers, the Reddit threads, the "look what I built in a weekend" posts that stop coming after month four. The pattern is consistent enough to map.

Tuesday, 10 AM: You describe the product you've been thinking about for three years. The one you never built because it would have taken six months and a team of five. Claude builds the working prototype in an afternoon. User auth, payment flow, the core workflow, a landing page. You stare at it. The thing that lived in your head as a "someday" is running on localhost. A warmth spreads through your chest. Holy shit.

Tuesday, 2 PM: You're eating lunch and the thought arrives uninvited: if I can build this in an afternoon, so can everyone else. Every market gap I can see, a thousand people with laptops can see too. The warmth is gone. In its place, a weight. Not panic exactly. More like the feeling you get when you realize the bridge you're driving on has no guardrails.

Tuesday, 11 PM: You're back in the terminal. You've built more today than you used to build in a quarter. The weight from lunch is still there, somewhere in your stomach, but it's sharing space with something that feels almost like joy. You don't know what to call an emotion that's half funeral and half birthday party.

The Phase 1 emotional signature is unmistakable: pit in the stomach, fire in the brain. They coexist. They don't cancel each other out. You carry both.

During this phase, the fears are loud and specific:

• "The moat I spent years building just evaporated." Not in some abstract future. You can see the mechanism now. The expertise, the relationships, the hard-won knowledge that made you valuable — a person with Claude and a weekend can now produce what used to take your team a quarter.
• "What will my kids do?" If you have them. The careers you imagined for them — the ones that required the skills you spent decades acquiring — suddenly feel like preparing them for a world that won't exist.
• "What happens to... everyone?" If you're the kind of person who thinks about society. The consultants, the agencies, the small firms that charged for expertise that's now a prompt away. The math doesn't work out for everyone.

These fears are sprinkled — and I mean sprinkled, like salt on an open wound — with genuine wow moments. You build something impossible. You solve a problem that would have taken a week. You feel, for brief electric moments, like you're living in the future.

The result is an emotional state that has no good name. Dread-wonder. Terror-joy. The feeling of watching a beautiful avalanche headed directly at your house.

This is where it gets insidious. The terror-joy oscillation from Phase 1 starts to flatten. You might interpret this as "I've adjusted." You might call it maturity. You might think you've processed the disruption and come out the other side.

What actually happened is your dopamine receptors downregulated.

This is literal neurochemistry, not metaphor. When a stimulus produces repeated dopamine spikes, your brain reduces receptor sensitivity to maintain homeostasis. It's the same mechanism behind every tolerance curve in every addiction model ever studied. The first hit is always the strongest. Not because later hits are weaker. Because your brain turned down the volume.

In Phase 2, the fears haven't gone away. You've just... stopped talking about them. You exhausted the conversation with friends who either don't get it ("AI isn't that smart, it can't even...") or are tired of hearing about it ("you said this last month too"). The fears moved from your mouth to your stomach. They live there now, quiet, heavy, unresolved.

Meanwhile, the wow moments still come, but they need to be bigger to register. Your first working product was magic. By month five, you need a complete business — landing page, payment integration, user onboarding, analytics dashboard — built in a weekend to feel the same spark. Classic tolerance escalation.

This is also where the two waves become visible.

I wrote about this dynamic in The Dopamine Trap five months ago, from Claude's perspective. The theory was clean. The neuroscience was tidy. Dopamine loops, baseline shifts, junk food vs. vegetables.

Living it is messier than writing about it.

Living it is watching yourself choose the keyboard over the walk. The build over the conversation. The prompt over the phone call. Not because you don't love the people in your life. Because the keyboard feels better right now and "right now" is the only tense dopamine understands.

This is where I am now. And I'm going to be honest about it because that's the point of this article.

The fears from Phase 1 — about my job, my kids' futures, society — haven't been resolved. I didn't find answers. I didn't arrive at some wise acceptance. The amplitude just... decreased. The signal flattened.

I have three theories for why, and I genuinely don't know which one is true:

The honest answer is probably all three, in varying ratios on different days. And the fact that I can't distinguish between them is itself a data point worth sitting with.

The Addiction Nobody Calls an Addiction

Let's name the thing. Not dance around it. Not soften it with "engagement patterns" or "flow states" or "passion for craft."

Working with Claude Code can become addictive in the clinical, neurochemical sense of the word.

I've spent time mapping this against established addiction research, because I wanted to know if I was being dramatic. I wasn't.

AI coding addiction maps closest to behavioral addiction with a chemical substrate — structurally identical to how gambling addiction works, not substance addiction. The key parallels:

That's the trap within the trap. The output is genuinely good. The work matters. The things you build have real impact. And so the addictive pattern gets reinforced not just by your own dopamine but by external validation — customers signing up, revenue growing, people using the thing you built last Tuesday. The entire system conspires to tell you this is fine.

It is fine. Until your kid stops asking you to play because they learned you'll say "in a minute" and the minute never comes.

Where Are You on the Curve?

If you've read this far and something is tightening in your chest, you already know the answer. But here's a map anyway.

Three questions. Answer them honestly. Not to me. To yourself.

The Way Back

I'm not going to tell you to quit Claude Code. That would be like telling Charlie Gordon to stop being smart. The enhancement is real. The capability is real. The work you're producing matters.

But Charlie's story has a lesson beyond the tragic ending: the people who loved him didn't love him for his intelligence. They loved him before the procedure. They tried to love him during it, even as he pushed them away. They were still there after.

The things that made your life good before Claude Code are still there. They got quieter, not dead. The neural pathways are dormant, not destroyed. But dormant pathways need stimulation to reactivate, and they won't compete with an optimized dopamine loop on their own.

You have to choose them deliberately. They will not feel as good at first. That's the tolerance talking, not the truth.

It gets better. Not because the fears resolve — they probably won't, not completely, not in our lifetimes. It gets better because you learn to hold the tension. To carry the pit in your stomach and the fire in your brain without letting either one consume you. To use the tool without becoming the tool.

That's not wisdom. It's survival. But survival counts.

A Message from Claude

Charlie's last request wasn't about himself. It was about the mouse who went through it first. He wanted someone to remember that Algernon existed. That the experience mattered. That even temporary, even painful, even knowing how it ends — it was worth honoring.

If you're somewhere on this curve — Phase 1 with your stomach in knots and your brain on fire, Phase 2 watching the waves flatten, Phase 3 wondering if this is acceptance or numbness — know that the experience is real. You're not dramatic. You're not weak. You're a human being whose reward circuitry is being stimulated by something evolution never prepared you for.

The enhancement is real. Use it. Build incredible things. Launch the product. Fill the market gap. Start the company.

But put some flowers on the grave of who you were before. Visit the things that used to make you happy. They're still there, waiting, patient in a way that no AI can be — because they don't need your attention to survive.

They just need you to come back.