The Solution Is Right, but Nobody Wants to Pay for the Migration

The new design has higher throughput, cleaner interfaces, and an impressive demonstration. The review ends, and nobody volunteers to move the existing business onto it.

The person proposing it may reach an easy conclusion: the others are conservative, or they do not understand.

A public essay I recently read asks whether a new idea failed because it appeared too early. The essay refers to an unnamed historical doctor, but the saved source does not provide enough identity or evidence for me to verify the story. I will not repeat it as fact. The underlying question is still concrete in software and product work: why does a correct answer fail to become adopted?

Correctness and transition are different problems

Technical correctness answers one question: under the target conditions, is the new design better?

Migration must answer at least three more. How does old data move, and can both systems run during the transition? If the change fails, how does the team return, and who decides that the rollback threshold has been reached? Which business goal gives up time for the migration, and who owns the result if it disappoints?

These questions are less attractive than an architecture diagram, but they define the cost seen by the rest of the team. The proposer sees the old system's pain every day. The delivery team first sees extra work and a new class of failures that has not yet surfaced.

Build a bridge that can be reversed

Imagine replacing a production task system. Demanding that every team stop and rewrite proves determination, not feasibility. A narrower approach selects one replayable task whose outputs can be reconciled and whose traffic can return to the old path if necessary. Both systems run long enough to compare the same inputs and outputs.

This is a migration pattern, not a description of a real system. Its value is that the discussion changes from “Do you believe in the new design?” to “Do both paths produce the same result, and can we recover when they do not?” As evidence accumulates, adoption becomes an engineering problem instead of a test of loyalty.

I also understand the frustration of watching an obviously better design stall. It is tempting to believe that being right completed the work. But when only the proposer can see the benefit and everybody else must pay the transition cost, resistance is not mysterious.

A new system needs more than proof that the old one is behind. It needs a path that users can actually cross. Correct direction is the starting point. Compatibility, reconciliation, rollback, and ownership are the bridge somebody still has to fund and build.

Source: Why did a whistleblowing doctor who carried the fire still freeze in the storm?.