Decisions feel like progress. You were undecided, now you're decided. Forward motion.
But here's the thing: every decision creates future work.
Someone will need to understand the decision. Someone will need to maintain what was decided. Someone will need to change it when circumstances change. Someone will need to explain it to new team members.
That's liability, not asset.
The Decision Burden
Consider a technical decision: "We'll use PostgreSQL for the database."
Seems simple. But that decision now requires:
- Understanding: Future developers need to know this was decided and why
- Maintenance: PostgreSQL needs to be kept running, updated, backed up
- Explanation: New team members need to understand the choice
- Context: When someone asks "why not MySQL?" there should be an answer
- Potential reversal: If circumstances change, someone faces migration work
The decision didn't create value directly. It created future work.
Why This Matters for AI Development
AI multiplies decisions. Every prompt iteration involves micro-decisions. Every generated code block embeds decisions. The volume of decisions exploded.
If each decision is a liability, you're accumulating liabilities faster than ever.
Most teams don't think about this. They make decisions, move on, don't document, don't track. The liabilities pile up invisibly.
Then someone asks "why does this work this way?" and nobody knows. The liability comes due.
Treating Decisions as Liabilities
If decisions are liabilities, you should:
**Make fewer decisions.** Don't decide things you don't need to decide. Keep options open where possible.
**Make reversible decisions.** Prefer choices that can be changed over choices that lock you in. Reversibility reduces the liability.
**Document decisions.** When someone needs to understand the decision later, documentation pays down the liability. Undocumented decisions have the highest carrying cost.
**Review decisions.** Periodically revisit decisions. Are they still valid? Do circumstances warrant reversal? Active management of the liability portfolio.
**Assign ownership.** Every significant decision should have an owner who can explain it, maintain it, and decide when to revisit it.
The Documentation Connection
Documentation is how you manage decision liability.
A documented decision: - Can be understood without finding the original decision-maker - Can be explained to new team members - Can be evaluated when circumstances change - Can be reversed deliberately rather than accidentally
An undocumented decision: - Requires archaeology to understand - Creates confusion for new team members - Gets changed accidentally when someone doesn't know it existed - Becomes unexplainable mystery over time
Documentation doesn't eliminate the liability. But it converts a high-maintenance liability into a lower-maintenance one.
The Practical Impact
Thinking of decisions as liabilities changes behavior:
**Before a decision:** "Should I decide this? What's the liability I'm creating? Can I defer? Can I keep it reversible?"
**During a decision:** "Let me document this so the liability is manageable."
**After a decision:** "Who owns this decision? When should it be reviewed?"
This isn't bureaucracy. It's realistic assessment of what decisions actually cost.
Accumulating Thoughtfully
You can't avoid decisions. Building things requires deciding things.
But you can be thoughtful: - Decide what needs deciding, not more - Prefer reversible over irreversible - Document so future maintenance is cheaper - Review and retire decisions that no longer serve
Decision management is liability management. The teams that get this right build maintainable systems. The teams that don't accumulate mystery.