Skip to content

reversible-irreversible.md

Latest commit

 

History

History
93 lines (64 loc) · 4.35 KB

reversible-irreversible.md

File metadata and controls

93 lines (64 loc) · 4.35 KB
Raw

Reversible vs. Irreversible Decisions

The single most useful mental model I've adopted as a manager. Before investing time in analysis, consensus-building, or escalation, one question: can this be undone cheaply?

If yes, decide fast and move. If no, slow down and get it right. Most leadership dysfunction comes from getting these backwards.

Type 1: One-Way Doors

You walk through and can't walk back. The decision locks you in for years.

You know it's Type 1 when:

  • Reversal costs more than the original implementation
  • External parties are affected (customers, contracts, regulators)
  • People's jobs or team structures change
  • Multi-year capital or license commitments are involved

Examples I've encountered:

  • Choosing the primary database engine for a platform
  • Signing a 3-year enterprise license
  • Making a public API commitment to customers
  • Letting someone go
  • Choosing a cloud provider

How I treat these:

  • Gather data. Find someone who's made this exact choice before and ask what they'd do differently.
  • Write an ADR. Force myself to articulate alternatives honestly.
  • Sleep on it. If I'm still comfortable tomorrow, proceed.
  • Get input from someone who disagrees with me. If I can't find that person, I'm not looking hard enough.
  • Set a decision deadline anyway. Type 1 doesn't mean Type Never.

Timeline: days to weeks. If it's taking months, something else is broken.

Type 2: Two-Way Doors

You walk through, look around, and walk back if you don't like what you see.

You know it's Type 2 when:

  • Reversal takes hours or days with minimal disruption
  • Only your team is affected (no external commitments)
  • You can contain the blast radius (feature flag, one service, one sprint)
  • Cost of trying + reverting is cheaper than cost of analysis

Examples:

  • Trying a new meeting structure
  • Choosing a code formatter
  • Setting alerting thresholds
  • Assigning a project to person A vs. person B
  • Picking a frontend component library

How I treat these:

  • Decide today. Not next week.
  • Delegate downward. The person closest to the work decides.
  • Don't require consensus. One person's judgment is enough.
  • Set a review date (2-4 weeks). Try it, then evaluate.
  • Don't write an ADR. Just move.

Timeline: minutes to hours.

The Gray Zone

The hard part. Many decisions CAN technically be reversed, but the reversal carries real cost (rework, lost months, team whiplash, context-switching tax).

Examples:

  • Selecting a service mesh (reversible in theory, 6 months to migrate in practice)
  • Restructuring team responsibilities (reversible, but trust erodes if you whipsaw)
  • Choosing a CI/CD platform (reversible, but pipeline migration is painful)

What I do in the gray zone:

  • Time-box the decision. "We decide by Friday."
  • Limit blast radius. "Try this on one service before going org-wide."
  • Set an explicit reversal trigger. "If metric X doesn't improve 20% in 8 weeks, we revert. No debate."

Your Job as the Leader

Not to make every decision. To classify them correctly.

Most leaders treat too many things as Type 1. The team gets slow, political, permission-seeking. People stop deciding without asking you. Everything escalates.

When someone brings you a decision, ask: "What happens if we get this wrong?" If the answer is "we fix it next sprint," it's Type 2. Send it back down.

The Mistakes I've Made

Treating everything as Type 1 (my first year managing). The team stopped deciding anything without me. Velocity dropped. I thought I was being "rigorous." I was being a bottleneck.

The "we can always change it later" excuse. Reversible doesn't mean thoughtless. Type 2 decisions still deserve 5 minutes of thinking. They just don't deserve 5 weeks.

Not recognizing gray-zone decisions until too late. We adopted a tool, went all-in across 4 teams, and realized it didn't fit 8 months later. That was a gray-zone decision I treated as Type 2. Should have piloted with one team first.

When This Doesn't Work

Regulated environments where even "reversible" decisions have compliance implications. In healthcare or financial services, the regulatory cost of reversal can make technically reversible things effectively irreversible. Audit trails, change control boards, and regulatory review cycles change the calculus.

Teams with very low trust. If every decision gets second-guessed regardless of type, the issue is culture, not framework.