thinking like a scientist

A reflection on how the habit of not fooling yourself — testing assumptions, naming things carefully, updating beliefs — shows up in programming and in everyday decisions.

I was trying to explain to someone how a zipper works.

Not how to use one. I’ve been doing that since I was a child, without looking, without thinking.

I couldn’t explain it.

I said something about teeth, then the slider, then I ran out of words. I’m not sure either of us knew I had stopped explaining and started throwing random words in the hope they would fit together.

That stayed with me. So I looked it up.

Here is what is actually happening.

Each tooth on a zipper has a tiny hook on one side. The slider — the thing you pull — has a Y-shaped channel inside it. When you pull it up, the Y squeezes the two rows of teeth together and the hooks click into place, one by one. When you pull it down, the pointed tip of the Y wedges between the teeth and pops them apart.

That is it. A Y-shape that either squeezes or wedges, depending on which way you move it.

I have used this every day for decades. I learned that just now.


I have been a programmer for a long time. In that time I have learned a lot of vocabulary and concepts.

There is a specific kind of confidence that comes from vocabulary. You start to feel fluent. You can read documentation, join conversations, follow arguments. And for a long time I mistook that fluency for understanding.

The gap only shows up when someone asks you to explain.

Richard Feynman said that knowing the name of something is not the same as knowing the thing. That the name is just a handle. The understanding is something else entirely.

I think about that often. Not because of who said it, but because I have felt it — that specific moment when the explanation runs out and you realise the words were doing more work than you knew. Like someone mid-sentence who forgot what they were actually saying. Maybe mid-sentence I realised that I did not understand something as well as thought I did. The Dunning-Kruger effect.


I solve a programming problem the same way I couldn’t explain the zipper.

I form a hypothesis early. Usually before I’ve looked at much. Then I go looking, but I’m not really looking for the answer. I’m looking for confirmation.

I read the logs in the direction of what I expect. I check the places I already suspect. I interpret ambiguous output in the way that fits.

Sometimes it works. The bug is where I thought it was.

But sometimes I spend an hour in the wrong place. When I trace back what happened, it’s usually the same thing: I didn’t test the hypothesis. I assumed it.

Someone who can pull a zipper perfectly but has no idea why there is a Y-shaped channel inside the slider, and a debugger who cannot explain their own assumption, are doing the same thing. Operating on a mechanism they trust but cannot describe.

That is fine for daily repeated activities. It is less fine for critical problem solving.


I’ve worked alongside engineers who knew roughly what I knew. Same frameworks. Same technical jargon. It’s like we spoke some secret language, but what I said might have been totally different from what they would have understood.

Some of them would hold a broken assumption much longer than others.

Not because they weren’t smart. Because they were attached to being right on the first read.

The ones who found things faster weren’t the ones who knew more. They were the ones who held their assumptions lightly. Who expected to be wrong somewhere and went looking for it early. Who updated without friction when something didn’t fit. They were ready to look at data or information they had and update the assumption they held.

That’s a different thing from knowledge. That’s probably the relationship with uncertainty.


We might do the same thing in conversations. Someone is still speaking and we’ve already decided what they mean. We listen for evidence that we were right. When something contradicts our expectation, we find a way to explain it away.

This is efficient. The mind has to be. You can’t scrutinise everything.

But it means there is a whole class of mistakes that feel like understanding. Ones that are invisible precisely because they feel so familiar.


A scientist would tell you this is not a personal failing. It is how the mind is built.

Pattern recognition evolved to keep you alive, not to find the truth. When early humans saw a predator, they didn’t run a controlled experiment — they ran. The instinct to protect a working model is efficient. It has served us well for a very long time.

The zipper is a small reminder of that. I didn’t know how it worked because I never needed to. The mechanism was invisible precisely because it worked.

Neil deGrasse Tyson said that the universe is under no obligation to make sense to you. The data doesn’t care whether you look at it. Reality will be exactly what it is, regardless of how confident you feel about what it should be.

The scientific method is a formalised habit of distrust toward your own conclusions. Every hypothesis is written as an invitation to be proven wrong. A scientist who has been shown wrong hasn’t failed — they have been handed something more accurate in exchange. That is not embarrassing. That is the whole mechanism.

What I find useful in that framing is this: the discomfort of being wrong doesn’t go away. But what it means changes.

It stops feeling like a flaw and starts feeling like the work.


I am still doing all of this. Regularly.

I still explain things with words I don’t fully understand. The zipper was one. There are thousands more that haven’t come up yet.

The difference now, maybe, is that I notice the moment when I run out of real explanation and start making up concepts I don’t understand. Only sometimes. But that’s a start.

Occasionally I stop mid-sentence and realise I’ve been holding a word like it was a thing. The word is not the thing.

And then I have to go back to the beginning. Which is uncomfortable.

And also, I think, how you actually learn something.


I keep coming back to a simple question.

Not what do I know?

But can I describe a bicycle to someone who has never seen a bicyle?

If I can’t, I’m probably holding the handle again.

That question costs almost nothing to ask.

And sometimes it completely changes where I look next.