09/02/2026
Three things went wrong on a project weโre proud of and each one reinforced why responsible speed requires more than good technology.
๐๐ ๐ฉ๐ซ๐๐ฌ๐๐ซ๐ฏ๐๐ ๐ญ๐ก๐ ๐ฐ๐ซ๐จ๐ง๐ ๐๐ฌ๐ฌ๐ฎ๐ฆ๐ฉ๐ญ๐ข๐จ๐ง.
Logic from the legacy system was copied, assumed to be correct, and reinforced by a test written around it. The test passed, but the behavior was still wrong.
Tests can tell you whether a behavior has been preserved. They cannot tell you whether that behavior deserves to survive. A test user caught what the test suite could not.
๐๐ก๐ ๐ซ๐ฎ๐ฅ๐๐ฌ ๐ฐ๐๐ซ๐ ๐ซ๐ข๐ ๐ก๐ญ, ๐๐ฎ๐ญ ๐ญ๐ก๐ ๐ฉ๐ซ๐๐ฌ๐๐ง๐ญ๐๐ญ๐ข๐จ๐ง ๐ฐ๐๐ฌ ๐ฐ๐ซ๐จ๐ง๐ .
The logic worked as intended, but our assumptions about how the results should be displayed did not match how users actually interpreted them.
๐๐จ๐ง๐ฏ๐๐ซ๐ฌ๐๐ญ๐ข๐จ๐ง ๐๐๐๐ญ ๐๐ฎ๐ญ๐จ๐ฆ๐๐ญ๐ข๐จ๐ง.
For one feature, the feedback mechanisms built into the prototype limited responses instead of encouraging useful input. A fifteen-minute conversation uncovered what email exchanges and automated feedback could not.
These issues were not resolved because we had all the answers. They surfaced because working software was placed in front of real people earlyโand because the team kept testing, listening, and asking questions.
That is what responsible speed looks like: a strong test suite, real users working in the system, and the willingness to pick up the phone when technology alone is not enough.
Read the full story, โ๐น๐๐๐๐
๐น๐๐๐๐๐๐๐๐๐๐ ๐พ๐๐๐๐๐๐ ๐๐๐ ๐ฉ๐๐-๐ฉ๐๐๐ ๐ฉ๐๐
๐๐๐,โ by Todd Miller, in JMA Field Notes.
๐https://www.jmares.com/post/rapid-refactoring-without-the-big-bang-budget
Stop throwing money at big-bang software rewrites. Learn how a small team rejected traditional, multi-million-dollar budgets to replace a failing, public-facing search tool in just weeks. By building continuous feedback loops into the product, working in the open, and using AI as a disciplined force...