What a Judgment Support Tool Actually Looks Like
I've spent five essays describing a problem. Now I want to describe a solution — or at least, the shape of one.
This isn't a product pitch. It's a design philosophy. I want to establish what a judgment support tool should and shouldn't be, because the category is so new that it's easy to accidentally build the wrong thing.
The core job
A judgment support tool has one job: make your current intention the easiest thing to act on.
Not the most important thing. Not the thing the algorithm thinks you should do. Not the thing you said you'd do yesterday. Your current intention — right now, in this session, at this moment.
Everything else is secondary.
What it shouldn't be
It shouldn't ask you questions.
The moment a tool requires you to articulate your intention explicitly, you've added cognitive overhead. Good judgment support is inferential or trivially low-friction. It picks up on context, or it lets you express intent with a single gesture, and then it works silently.
If you need to configure it, it has failed. If it needs you to train it, it has failed. If it shows you a setup wizard the first time you open it, it has failed.
It shouldn't enforce.
Judgment support is not a warden. It doesn't lock you out of things. It doesn't create friction for choices you might legitimately want to make. It shifts the environment slightly toward your intention, leaving everything else accessible at one additional step.
The key word is shift, not block. You should never feel trapped by a judgment support tool.
It shouldn't moralize.
No scores. No streaks. No "you spent 3 hours on social media this week" graphs designed to make you feel bad. Judgment support is forward-looking. It's about making the next action easier, not reviewing the last hour with a red pen.
What it should be
Ambient. It works by changing the shape of the environment, not by adding a new thing to manage. The best version is one you stop noticing because it's just become the way things work.
Reversible. At any moment, you should be able to turn it off and get back to exactly where you were. Zero-risk interaction builds trust. And trust is what lets users actually rely on the tool instead of working around it.
Immediate. The effect should be visible in seconds. If you have to wait to see whether it's working, the feedback loop is broken. Good judgment support gives you the result before your attention moves on.
Invisible by default. The tool should be quiet 99% of the time. It does its job in the background. You only notice it when something is happening — a page getting quieter, a commitment being gently surfaced, a signal getting cleaner.