Cheap code gets expensive fast¶
Code is easier to produce than ever.
That doesn’t make it cheap.
Specs → code → garbage¶
Unclear input degrades over time:
Specs → code → worse code → garbage
AI doesn’t fix ambiguity.
It amplifies it.
Code is not cheap¶
The cost isn’t writing code.
It’s:
- understanding
- testing
- changing
- trusting
AI lowers the first.
Not the rest.
Before you code: align¶
Reach a shared design concept first.
Be clear on:
- what you’re building
- boundaries
- key decisions
Otherwise you’re just generating guesses.
Language matters¶
Ambiguous language → ambiguous systems.
Clear terms → better code, better tests, clearer ownership.
That’s DDD in practice.
Good codebases are easy to test¶
Hard to test usually means:
hard to understand
Testing forces clarity:
- what is a unit?
- what behaviour matters?
These are design decisions.
Deep modules¶
Prefer:
- simple interface
- complexity hidden
Avoid: - complex interfaces exposing everything
AI works better when boundaries are clear.
The takeaway¶
Cheap code gets expensive fast.
If you want momentum:
- align before coding
- use clear language
- design for testing
- keep boundaries tight
Otherwise you’re not moving faster.
You’re just creating problems faster.
A side effect: storage grows faster than understanding — and eventually, disk space runs out.