The Two-Week Test
The Test You Can Run in Two Weeks
There's a test you can run in two weeks that will tell you more about your relationship with AI tools than six months of self-reflection.
It's simple: take two weeks off. Real off — no coding, no AI tools, no keeping up with what's new. Just... absence from the thing.
Then come back.
What the Return Tells You
The first day back, you'll feel the friction.
Not the friction of getting back into work — the friction of getting back into your own head. The mental models that felt smooth before will feel rough. The decisions that used to be fast will feel slow. You'll notice your hands don't quite remember the keystrokes.
That feeling — that's the signal.
If the friction feels like rust and you can feel the skill waiting underneath, you're okay. The model is still there. It just needs to be warmed up.
If the friction feels like you're reading someone else's handwriting — if you look at a problem you used to own and it looks like it was written in a language you used to speak — that's the more important signal.
The Two-Week Threshold
The two-week threshold matters for a specific reason.
One week is enough to feel the mental load lift. You sleep better. The anxiety about falling behind fades. That's the rest signal — and it's real, but it doesn't tell you much about the skill.
Two weeks is enough for the skill signal to surface. When you remove the cognitive load of daily work plus the cognitive load of AI-assisted work plus the anxiety of keeping up — when all of that settles — what's left of your actual capability?
That's the two-week test.
The Problem With "I'll Just Check In"
The engineers who most need this test are least likely to run it.
"I'll just check the Slack."
"I'll just see what's been merged."
"I'll just stay current."
Every check-in resets the clock. The anxiety of falling behind reasserts itself. The skill signal never gets the silence it needs to surface. You come back from the "time off" as tired as when you left, because you never actually left.
The rule
The two-week rule requires something most engineers are not great at: trusting that two weeks won't permanently derail you. That the gap won't become structural. That you'll come back and the ground will still be there.
The engineers who've run this test report something consistent: the ground is still there. It just looks different than they expected.
The Return Inventory
When you come back, take inventory in a specific order:
Day 1
Look at a piece of code you wrote before the break. Not with nostalgia — with the question: do I still know why this was written this way?
Day 3
Pick a problem you used to be able to solve in your head. Try to solve it in your head again. Notice what you have to reach for.
Day 5
Write a function — a real one, something that does something — without AI assistance. Not to prove anything. Just to see what's there.
Day 10
Take a problem to AI like you normally would. Notice whether you still have a strong opinion about whether the answer is right before you run it.
Each of these is a diagnostic, not a test. There's no pass/fail. But the pattern of answers tells you something real.
5 questions · 4 tiers · See where you stand