Why Deadlinewatch has no formula chain.

I built a version of this tool that ranked my deadlines, used it for a few weeks, and then took the ranking out. This is the argument for why. It’s the closest thing I have to a core belief about how a tool like this should behave.

Every deadline tool I’ve looked at closely ranks for you. Some use machine learning, which in 2026 usually means a model whose reasons you can’t see at all. Some use weighted scoring. Some, surprisingly often, use a number-soup algorithm someone wrote in a single afternoon and then never revisited. You open the tool and see a list pre-sorted in an order you didn’t choose.

Deadlinewatch doesn’t do this. The list sorts by date. That’s the whole sort. Mark a deadline a priority to lift it above the date order. Snooze it to set it aside. Otherwise it runs in order by when it’s due.

This is deliberate. Four reasons.

One. The math has to be visible

You have to be able to read why the list is ordered the way it is. “Tax filing is at the top because it’s due Friday” is a read-and-trust statement. “Tax filing is at the top because a weighted score of 8.3” is not, even if the score is thoughtfully constructed. You can’t verify the score from the surface. You have to trust the tool, blind.

Trust delivered blindly is fragile. The first time the score is wrong, you notice. You start second-guessing. You open the tool less often. The tool earns its place in the morning routine by being legible, by letting you read why it’s showing what it’s showing, and a hidden formula chain breaks legibility.

An early internal prototype of Deadlinewatch had a formula chain. It scored each deadline from a handful of factors and pre-sorted my day. I used it myself for a few weeks. I liked it for about one. Then I noticed I had stopped trusting it. By the third week the pre-sorted order felt intrusive, an answer I didn’t agree with, so I ignored it and re-sorted in my head anyway. I never shipped it. I cut the formula chain in an early simplification pass, and the math has been days-until-due ever since. That’s all that was ever underneath it.

Two. An algorithmic rank implies expertise the tool doesn’t have

Deadlinewatch doesn’t know which of your deadlines matter most. It only knows the dates. Anything beyond that is a guess dressed up as a recommendation. When a guess proves wrong, you blame the tool, and the trust it needs to earn its place every morning evaporates.

This goes wrong most often in finance-adjacent territory. A tool that scores deadlines and surfaces them in a sorted order looks like it’s offering professional advice. It isn’t, but it reads like it is. If you act on the recommendation and miss something else, you can reasonably feel the tool let you down. That’s a bad position to put you in.

The honest move is to render the shape and let the human decide. The human knows their context. The tool does not.

Three. The experienced practitioner doesn’t want a ranking

They want a rendering. They’ll do the ranking themselves, because that’s the part of the work they’re actually good at. Pattern recognition is the senior skill that this whole publication keeps coming back to, and that skill produces the ranking automatically as a byproduct of reading the shape. A tool that does the ranking for them removes the part of the work they want to keep doing. It also removes the daily practice that keeps the skill sharp.

My target user, someone who carries many submission dates and feels the cognitive load all day, is almost always a senior practitioner of something. They already have a ranking. They just want help holding the shape that produces it.

Four. The math has to be honest

The math you see in Deadlinewatch is days until due. Priority items appear at the top. Overdue items appear in their own section. Snoozed items appear as ghost rows. That’s the entire ordering logic. It fits in a sentence. It can be checked at a glance. It doesn’t require trust.

The methodology page lays out the choices behind all of this, the things I deliberately don’t do and why. Not ranking your work is one of them. It belongs to the same family as not telling you what to do next. Both are design choices that keep you in charge of your own work. The same principle runs through the four axes the mechanics are built on. Each one renders a fact and lets you act on it, rather than acting for you.

Most of these are easy to add and hard to take back. I chose not to add them.

If you remember one thing, remember that a tool that ranks your work is making a claim it can’t back up. It doesn’t know your context. You do. The honest job for the software is to show you the dates clearly and then stay out of your way, which is the whole reason the only math you’ll ever see here is days until due.