"It greatly reduces human error, and on a personal level, helps save innocent lives. From the bottom of my heart, and I don’t say that lightly, it helps me sleep at night.”

When a user said those words to me in early 2018, I was hooked.

I knew my designs mattered, and the work I was doing was make the world a better place in a tangible way.

So why did I advocate for disabling that same app in production, three years later?

You Have To Know When To Kill Your Darlings.

The product I helped build, Raven, was incredibly valuable to a very specific user group, at a very specific time, in a very specific conflict. The application helped its users create accurate, clear records from actionable intelligence that resulted in high-confidence that the Air Force was making the right choices in life-or-death situations and creating the best plans we could.

For the next two years, I staunchly defended the scrappy little application when my organization's leadership argued that we should stop supporting it. They claimed we should ignore requests from the users for updates and new features, in favor of devoting all of our efforts to an unproven new undertaking, which was languishing in development hell.

They insisted Raven was unused and unloved, but when we dug into the usage data in 2020, we found that the app was still being utilized, so I worked to rekindle Kessel Run support for it and worked diligently to update the application and train new users.

A year later, I returned to the facility where we had first fielded Raven.

I requested the same usage data again, curious to see how the app had fared while I had worked on other projects.

There had been a sharp decline in usage.

No user had touched the app in months.

I went to talk to the user group in question, and found out that the nature of the conflict in the region had changed dramatically, and so had the planning processes for Raven's users.

The research I did confirmed what the data already showed: the team I had built the application for was no longer responsible for the same outputs, and Raven no longer met the organization's needs.

I'm proud to say that I wrote the support ticket to disable the application's last operational instance myself.

I'll build more software that saves lives in the future, and it's important to know when to kill your darlings.