The thing my dad kept telling me at 8. The thing my history teacher kept telling us at 15. The thing I kept not getting until I started merging PRs my coding agent wrote.
When I was maybe 8, I found tracing paper. Put it over a drawing in a book, follow the lines, get a "drawing" that looked exactly like the original. It felt magical.
Around the same time, stencils showed up at school. The cool kids had them. Sheets with cutout shapes. Flowers, cars, animals, geometric things. You'd move your pencil inside the cutout and get a perfect shape. Even fancier than tracing paper because you didn't need a book to copy from. The shape was already in the stencil.
My dad caught me doing the tracing thing. He was upset. Not because I'd done something wrong exactly. Because I'd skipped the part where I actually learn to draw. The hand doesn't get trained. The eye doesn't get trained. You just have a piece of paper with lines on it.
I didn't really get it at the time. I had a drawing. The drawing looked good. What was he mad about.
Later. School. History assignment. Individual research. Everyone in the class, including me, opened Wikipedia and copied the same three paragraphs. Different fonts. Different margins. Same paragraphs.
The history teacher walked in with all the assignments stacked and asked us if we thought he was stupid. Then he told us the point wasn't the paper. The point was going through the sources and coming out the other end with something that was actually yours. We had skipped the going-through part.
Also didn't really get it. I'd handed in something. It looked like an assignment. It said the right things.
Nielsen wrote something in 1997. How Users Read on the Web. 79% of users don't read. They scan. Only 16% read word by word.
I re-read that essay this week in the middle of merging PRs my coding agent wrote.
Nielsen was worried people were skimming. What I'm doing is one level below skimming. I'm not reading at all.
The agent reads the failing test. Writes a patch. Writes a summary of the patch. I look at the summary. Click merge.
I didn't read the code. I didn't read the diff. I read the agent's paragraph about the diff. Same as reading a Wikipedia paragraph and calling it my research.
Same as tracing.
The common thread across all of it. Tracing paper. Stencils. Wikipedia paste. Agent merge. Output without thinking. Output without creativity. And nothing of me in any of it. Not really mine. Just a piece of paper with lines on it.
Ship fast. Move on. A stack of PRs before lunch. My hand didn't get trained. My eye didn't get trained. But something got shipped, and it looks like work.
For an ADHD brain especially, this is the drug. Nothing to hold in your head. Nothing to slow you down. Just the shape of productivity without the tax of it.
I've been doing this for months. It ships. It compounds. It also hollows out.
Or it will if I keep going. This piece is an alarm clock, not a report. I'd rather catch it now than write it from further gone.
Prod goes down at 2am. Customer says the app renamed their folders. Security review flags a query you don't remember writing because the agent wrote it.
Now you have to actually read.
You open a file you've never actually read even though your name is on the commits. You open a PR from three weeks ago and realise you don't remember what it does. Your eyes slide off the agent's plan doc because you never had to comprehend one, you only had to approve them.
This is the moment my dad and my history teacher were warning about. Not the moment where the tracing paper fails. The moment where you actually need to draw.
The question I keep having to ask myself. When I merged that PR without reading it, was I optimising for speed, or was I being lazy?
Some of the time it's speed. Real deadline, small stakes, tracing the flower is fine for the birthday card.
Most of the time, honestly, it was laziness dressed as speed. The reading was hard. The clicking was easy.
And it's easier to be lazy than ever.
All of us walking around with attention spans hacked. Ten-second videos. Bursts of stimulation between anything you try to do. Low vitamin D from sitting inside. Undiagnosed ADHD for some of us. Diagnosed for others.
Short-form video isn't even that old. Five years feels like a decade because covid locked us inside and the phones did the rest. Acceleration.
I sometimes wonder if big tech pushed short videos on us to cause the brain rot so they could sell us AI later. Probably not that clean. Probably just the way incentives lined up. But the timing looks convenient.
I got diagnosed with ADHD in March 2025. Some things had gone wrong by then. Work. Relationships. I know who felt it before I did.
The chase for novelty and the constant need for stimulation make a lot of things hard. Fine print reads like a wall. Laundry sits folded in my head, never in the basket. Keys, wallet, phone, gone every five minutes. Talking over people because the turn takes too long. Closing my eyes on every app's loading screen to skip the wait. Snapping at the people closest to me.
Anger without processing the emotion. Fast reward. No work. Dopamine again.
The medication does wonders.
I'm not selling meds as the answer here. What I am saying is the environment we're being lazy in hits differently depending on what's going on in your head. For some of us the pull is stronger.
Either way. Being lazy is the path of least resistance right now for basically everyone. Which makes reading the diff harder than it would otherwise be.
Addictions are easy to pick up. Smoking. Casino. Stock market. Same shape. Fast reward, no work.
I lost 37k in 2020 doing intraday trading. That was the year I could have thrown any random money at Nifty50 and made a lot. I didn't. Long-term is boring. Weekly swings are boring. Intraday is where the dopamine lives.
Same as tracing a flower. The chart moved, I clicked, the number changed. Something happened. It felt like work.
It also stole focus from actual work. I only noticed later.
The way out wasn't a stock chart telling me I was wrong. It was people I look up to. Slow conversations. Someone asking "what are you actually doing here" without making me feel small about it. Enough time to accept that I was chasing the wrong shape.
I mention this because the pattern with agents is the same. Fast reward, no work, feels like productivity. I don't know if I would have caught the trading thing on my own. Might need the same kind of people around this one too.
Realising the pattern isn't enough. Knowing the words for it isn't enough. Reading a hundred articles online. Some right, some wrong. Meeting doctors who get it and doctors who don't. Inclusive people, and people who aren't. You have to sort through it. You have to decide.
Then you have to actually make the move.
For me the move was booking an appointment. The good doctor never had a slot the same day or the next day. It was always two days out. I want everything now. Two days may as well have been a month. I put it back on the list. Then off it. Then back on. Then off. Months of this. What eventually got me to book was strangers on a Slack channel typing me their own stories.
That's the executive function bit. Not thinking about the fix. Doing the first thing that gets you to the fix. Which is the same muscle that's already broken.
Same trap with agents. I know reading the diff is what I need to do. Knowing doesn't get the diff read. Sitting down and reading it does. And sitting down and reading it is exactly the thing the last few years have made hard.
The fix isn't complicated. You have to read. Actually read. Slow down at the diff. Read the plan before approving it. Read the code the agent generated before you merge it. Read the failing test the agent claims to have fixed.
Simple. Not easy. Because the whole point of the agent was to save you from that. Going back and doing it feels like undoing the productivity gain.
But it's not undoing it. Reading is what makes the agent useful instead of a liability. Reading is the hand training. The eye training. It's the part that was supposed to actually be yours.
He was writing about people reading marketing copy and news articles. Low stakes. If they scanned and missed a nuance, whatever.
The stakes for us are different. Approving code that runs in production. Approving migrations that touch millions of rows. Merging PRs we haven't read.
The old habit worked for the web. Scan the page. Get the gist. Move on.
The new habit is not fine for code. Accept the summary. Ship the PR.
You have to read again. Or you find out what you missed the hard way.
I still maintain a website for my dad's office. evescentrekannur.com. Small clinic in Kannur. Profile pictures. New staff photos. A doctors list that changes every so often.
Started years ago. I built it in WordPress so he could update it himself. He never did. He kept asking me. I ended up paying the hosting too. Eventually I realised he didn't need WordPress. He needed something someone else could edit for him. Ripped it out, made it vanilla HTML.
These days Claude does the edits. He asks me for a change. I ask Claude. Claude opens the file, makes the change, deploys it.
Last time he asked for a change I was sitting next to him. I made a show of it. Look, I don't have to work anymore. It's easy. The changes appeared on screen like magic.
He was not impressed. He looked at me and said: what about your skills now? Wouldn't it all fade if you don't practice?
Twenty-something years since he caught me with the tracing paper. Exact same question.
I didn't have a good answer this time either.
I told him getting stuff done is what I'm paid for. That's the job. And this was just clerical work on a small clinic website anyway. Not that serious. I let it go.
I've been thinking about that answer since.
I might go back to playing the old PC games. Mario. Dave. The ones I played in LKG, before HTML. The ones he put in front of me on the first computer he brought home.
Not because gaming is a fix. Because doing something small and slow and yours, without an agent between you and the thing, might be the muscle I need to get back.