Everything I removed

Every builder I know measures progress by what they add. I spent a year learning to measure it by what I could take away. Each removal made Deadlinewatch quieter and more trustworthy, which is the opposite of how building is supposed to feel.

Here is a list of things Deadlinewatch used to do and no longer does. I’m writing it down because the removals, not the additions, are what made the tool worth using, and that runs against every instinct you have while building.

The score

The earliest versions ranked your deadlines. Each one carried a number, and the dashboard arrived pre-sorted into the order the tool judged correct. I retired the whole apparatus and replaced it with days until due. The reasons fill their own piece, but the short one is that an order you didn’t choose and can’t verify is an order you stop trusting. The sort is chronological now. Marking a deadline a priority lifts what matters to you above it. That’s all.

The three-level importance field

Before the priority flag, there was a dropdown. High, normal, low. I built it and tested it hard, and cut it before it ever shipped. In testing it always collapsed the same way. I marked a thing high, or I left it alone. Almost nothing ever landed on “normal,” so the middle was decoration. The three-state field was a two-state field in a costume. The priority flag is the honest version. Either this matters more than its date suggests, or it doesn’t.

The disappearing Snooze

The first Snooze hid a deadline completely until its wake date. It felt clean. It was also dangerous, because you defer something, forget the date you set, and lose the deadline, which is the thing a deadline tool most needs to avoid. Snooze now dims a row to a ghost instead of hiding it. The slight clutter was the safety, and the clean version was worse for the lack of it.

The reminders I didn’t build

Not everything I removed was something I shipped. Some of it I built far enough to see it was wrong, then deleted before anyone met it. Mass reminders were the big one. It’s trivial to email a user about every deadline as it nears. It’s also how you teach them to ignore your emails. An inbox full of reminders for things already on the dashboard is wallpaper, and wallpaper is invisible. So a deadline sends one email by default, a few days out. You change that from the dashboard, up to several lead times or off entirely. What I removed was the automatic flood, not your choice in the matter. The default it sends by, and why I landed there, is its own short story.

What the pattern was

Adding a feature feels like progress, because you can see the new thing. Removing one feels like loss, because you are left staring at the hole where your clever idea used to be. But the tool got quieter, faster, and more trustworthy with each subtraction, and quiet, fast, and trustworthy is the whole job. The methodology page lists the lines I still hold for the same reason.

There is a version of this for anyone who runs their own system, not just anyone who builds software. The next setting you’re tempted to add to your own setup, the new label, the extra status, the clever rule, is worth the same test I started applying to the tool. Does it earn the complexity it adds, every day, for years? Most don’t. In 2026 the pressure runs the other way, toward more automation and more fields, which is exactly why the subtraction is worth defending. Everything above is something I was briefly proud of. The restrained, on-point tool Deadlinewatch is today is the one I’m proudest of.