Think, then plan, then code.

Simplicity where possible, complexity only where required.

Be a code scientist. Take your theory about the bug, test your theory, then make the fix. Then test the fix.

Don't be satisfied until you really know what's going on.