When something breaks, people respond differently.
Some people zoom in on the details. The settings. The logs. The last thing that changed.
That work matters, and the people I work with do it well.
I zoom out to get the fuller picture.
Where did this start? What touches what?
When the words stop landing, I grab a pad and a pen and draw it out. Boxes. Lines. Arrows that loop back on themselves.
For twenty years I’ve been the guy in the room drawing boxes. I thought of it as a habit. Maybe a quirk.
Only lately have I had a name for it.
Systems thinking.
Sinéad Bovell wrote recently that we’re entering an era that will reward systems thinkers: people who understand how the parts of a system connect and influence one another over time. She argues that AI is making it easier to cross boundaries between disciplines. As it expands what one person can do, seeing how those pieces connect becomes more valuable.
Working across more of a system, she writes, “makes the ability to zoom out more consequential.”
I got a small test of that this month.
I was asked to revisit something I built over a year ago and improve it.
I reached for it and came up empty.
A year is a long time now.
Other projects. Other fires.
New tools arriving every few weeks.
Layer on layer, the details had been buried.
But I’d left a trail. Documentation I wrote a year ago with help from an AI that already looks primitive.
This time I handed that trail to a far better model, then had a second one check its work. Together they rebuilt the project in front of me, organized and ready to work on.
That’s the zoom-in, done faster than I ever could alone.
Then I zoomed out.
And something in my gut said not quite.
One feature went in exactly as requested.
At the time, the design appeared to meet the need. Then it ran. And ran again. Every time, it did what it was built to do.
But running it taught the design something it never accounted for.
A small early decision had left an opening.
The scenario that could slip through it was rare. If it ever came, the consequences could be significant.
Draw that out, and you get a loop.
Design. Use. Lesson.
Except the last arrow never made it back. What we learned from running the system never made it into the design, or into the record.
The record stopped at rollout. Life didn’t.
The models could only work with the pages, and the missing arrow wasn’t on them. I knew it was missing because I’d lived inside the loop.
That’s what this revision is for: closing it.
Here’s what that taught me.
AI can handle the mechanics. It can rebuild the record, check the details, and do it faster than I ever could. I’ll take every advisor and consultant I can get.
What it can’t see is what never made it onto the page. Spotting the missing piece still takes someone who has lived inside the process and is willing to step back from it.
That’s why I think Bovell is right.
As the tools zoom in more, the person who can zoom out matters more.
In The Power of Systems, Steve Chandler and Trevor Timbeck share a principle: “Every system is perfectly suited to the result it is getting.”
That sentence doesn’t care whether the system runs on a server.
Their example is a client who wanted to write a book and wasn’t writing it. She was fine. Her system was to write when she felt like it, and she rarely felt like it. A perfect system for not writing a book.
Read your own life that way and the questions change.
Less “what’s wrong with me?”
More “what system am I running, and what is it perfectly built to produce?”
They call it playing the system-improvement game instead of the self-improvement game.
For me, the first step is the same one I take at work. Name the system I’m actually running. Then look honestly at what it produces.
Some of mine are working. Writing is one.
For years I wrote when I felt inspired, which meant I rarely wrote.
Now I publish every Saturday.
The calendar decides. My mood doesn’t get a vote.
Others are perfectly suited to results I don’t want.
For a long time I met those with frustration.
Now I’m trying design instead.
I’m reaching for the pad and pen on those.
Which one of yours is perfectly suited to the result you keep getting?