Why early decisions matter.
WHAT THE PATTERN IS
A defect or misunderstanding costs more to correct the later it is found.
WHERE IT IS CHEAPEST
In a conversation, before anything is built.
WHERE IT IS MOST EXPENSIVE
In production, after users depend on the behaviour.
WHY THE COST RISES
More has been built on the assumption More people are affected Data may already exist in the wrong shape Other systems may depend on it
WHAT THAT ARGUES FOR
Establishing requirements properly Reviewing designs before building Testing early and continuously Delivering small increments
WHAT IT DOES NOT ARGUE FOR
Attempting to decide everything in advance.
WHY
Some things cannot be known until something exists, and deciding them early means deciding them wrongly.
WHAT TO DECIDE EARLY
Decisions that are expensive to reverse: data models, external contracts, platform choices.
WHAT TO DEFER
Decisions that are cheap to reverse.
WHAT TO ASK OF ANY DECISION
How expensive is this to change later?
WHAT THAT QUESTION DETERMINES
How much thought it deserves now.
