Series: Gsd (3/7)
Done > Perfect: The Core GSD Mindset, from Planning Paralysis to Actually Moving
"Just one more tweak."
If your work history contains a pile of "90% done" things, you've probably said that phrase more than once.
The problem isn't that remaining 10% — the problem is that the cost of improving that 10% is far higher than you imagine. High enough that you'd rather leave it in drafts than face the moment of "submit it, then wait for a response."
Perfectionism isn't a personality trait — it's a risk avoidance strategy. The price of avoidance just accumulates as a pile of unfinished things in your to-do list.
Done > Perfect Doesn't Mean Doing Shoddy Work
Let's clear this up first, because it's the most common misunderstanding.
Done > Perfect doesn't mean "do it carelessly, quality doesn't matter." It means: shipping something in the state where people can actually use it is more meaningful than it sitting on your hard drive forever as a perfect plan with zero real-world impact.
A 70-point thing shipped can:
- Get real feedback, letting you know which of the missing 30% actually matters and whether it's worth filling
- Let the other party or user start using it, accumulating real data
- Free your brain from the background noise of "this thing isn't done yet"
A 95-point draft that never shipped only offers:
- You continue knowing it exists, continuing to drain your attention
- Nobody gets any value from it
Do the math — it's clear.
Why "Perfectionism" Is More Dangerous in the AI Era
Before, doing something had friction: revising cost time, redoing cost effort — that friction, in some sense, forced decisions.
AI has almost eliminated friction. You can ask AI to help you generate five versions, each making good sense, each making you think "let me think about it more, maybe the next one will be better." AI generates quickly; your decision speed doesn't speed up in parallel — the result is you lose direction in infinite possibilities.
What's more troubling: every version AI gives you looks good and professional — which makes "which one to pick" harder, not easier. No obviously bad options to eliminate; only "anxiety of choosing between good options."
AI is an execution accelerator, not a decision accelerator. The latter requires Done > Perfect to fill the gap.
Three Tools for Resetting "Good Enough"
1. Define the Completion Condition (Definition of Done)
Before starting any task, answer this question clearly: "What state counts as done?"
Not "what state counts as perfect" — but "what's the minimum viable state to ship this and let it serve its purpose."
Examples:
- Writing a technical article → DoD: structure is clear, core argument is supported, reader can follow along and try it themselves (not: every word is perfect, every detail is cited)
- Developing a feature → DoD: main flow works, no blocking bugs (not: every edge case handled)
- Preparing a proposal → DoD: problem is clear, solution is reasonable, recipient knows the next step after reading (not: every slide is beautifully designed)
Write the DoD down. When you start "just one more tweak," check back against it — if the DoD is already met, you're doing work beyond what's necessary.
2. The Time Budget Method
Set a fixed time limit for the task. When time's up, ship in its current state (or make the decision to push it forward).
This isn't slacking — it's using a psychological fact: Parkinson's Law — work expands to fill all the time you give it. Give something two hours, it usually takes two hours; give it one hour, most of the time one hour is enough.
AI can help you here: when starting a task, tell AI "I have 45 minutes, please help me stay focused on the most important part" — its response style will adjust accordingly.
3. "Good Enough" vs. "Better"
Before every "let me just fix this" impulse, ask yourself: "Is this change to make it 'good enough,' or to make it 'better'?"
"Good enough" changes are worth doing — they're filling real gaps.
"Better" changes are, very often, satisfying your own anxiety, not actually giving users more value.
The feeling that lies to you is: "just one small step away from perfect." — that one small step usually has no endpoint.
Starting Matters More Than Perfecting
The other half of Done > Perfect is about starting.
Many people wait until "they're ready" before beginning, but the feeling of "I'm ready" never truly arrives — because the more you prepare, the more clearly you see what else isn't prepared.
GSD's starting principle is: use what you have now, do what you can now, then correct as you go.
This is especially important for using AI tools. Don't wait until you have a clear problem before asking AI — bring in a fuzzy question and let the conversation help you clarify the problem. That itself is a valid way to start.
"I don't know how to do this yet" is a reason to start, not a reason to wait.
One Sentence to Remember This Article
Done beats perfect. Shipping beats sitting there waiting for the right moment.
Next Step
With the mindset in place, the next question is: okay, I've decided to move — but the task is still vague. "Getting this thing done well" doesn't equal "knowing what the next step is."
How do you turn a fuzzy intention into a list you can execute today? That's what Breaking Down Tasks with AI: From Vague Requirements to an Actionable List solves.
Back to Series Overview
This is the second stop in the GSD series, covering the Done > Perfect mindset and starting philosophy. For the full reading path and other article topics, return to the GSD Beginner Course.