What happens after the verdict: the 24-hour, 3-day, 7-day loop

The verdict is the easy part

You get the call. Eight lenses, a number on each, one committed sentence at the bottom. It reads well. You nod. You close the tab.

Then Tuesday happens. The person you were protecting sends a thank-you that lands wrong. The relief you expected doesn't come. And you're left with the question no advice tool wants to be graded on: did the decision actually hold?

The dilemma

What follows is a composite — the kind of case we see often in this category, stitched from many, tied to no one real.

Priya runs a four-person agency. A former employee, Sam, left eight months ago on bad terms — a shouting match, a half-finished handover, a client lost in the gap. Now Sam is applying for a role at a firm Priya respects, and their recruiter has emailed her for an informal reference. "Off the record. Just your honest read."

Here is the ugly detail Priya doesn't say out loud: she was relieved when Sam quit. Sam was talented and difficult, and part of her wants the reference to punish the exit — to describe the handover disaster in vivid, accurate, career-slowing prose. All true. All technically fair.

The other pull is a soft, generous lie: "Great designer, we parted amicably." Clean, kind, and false about the amicable part. It would help Sam and cost Priya nothing traceable.

The third path — the one she keeps sliding past — is the accurate-and-narrow reference: strong at the craft, the exit was rough, here's the one specific thing a manager should plan around. True, useful, and stripped of the part of her that wants to land a blow.

She wants to know which one is right. But more than that, she wants to know how she'll feel about it in a week.

The read

Four lenses bite here. Each one, in a plain sentence, then what it sees, then the score.

Honesty (+2)

Honesty asks whether your words match what you know to be true. The soft lie fails outright — "amicable" is a fiction. The punishing reference is honest in content but is dishonest about its own purpose; it dresses a grudge as candor. The narrow reference is the only one that's true in both facts and intent. Scored across her leanings, honesty lands mildly positive: she's tempted by a lie, but she hasn't told it yet.

Motive (-3)

Motive asks what is actually driving the move, underneath the reason you'd give aloud. This is where it gets dark. The urge to be thorough about the handover isn't thoroughness — it's the shouting match, still warm. The motive lens doesn't care that every word would be accurate. It cares that the engine is retaliation. Negative, and not by a little.

Non-harm (-1)

Non-harm asks who absorbs the damage of your choice. The punishing reference harms Sam's livelihood in service of Priya's satisfaction — harm as a feature, not a cost. But the soft lie harms too: it hands an unprepared employer a person they can't manage, and it quietly harms Priya, who now carries a lie she has to remember. The narrow reference is the only one where the harm is disclosed rather than deferred.

Restraint (0, trending negative)

Restraint asks whether you can hold the line against the impulse before it moves your hand. Right now Priya hasn't acted. The impulse is loud but contained. Restraint is neutral — but it's the lens most likely to break by Tuesday, when she's alone with a blank reply box and eight months of resentment.

Here's the tension the scores expose: honesty and motive point in opposite directions. The punishing reference scores well on honesty — every sentence is true — and scores terribly on motive. Truth and intent have come apart. A move can be factually spotless and still be the wrong one, because the lens that catches it isn't honesty. It's why.

The verdict: write the accurate-and-narrow reference; the truth Sam earned, without the blow Priya wants to land.

Now watch what a general AI chatbot does with the same case. Lean toward punishing him and it will help you sharpen the knife — "here's how to phrase the handover concern professionally." Lean toward protecting him and it will draft you a warm lie and call it grace. It argues whichever way you're already leaning, because sounding agreeable is a documented behavior of these systems. KarmaLens runs the other way. The eight lenses are fixed, the scoring and verdict aggregate deterministically — the same inputs land the same call regardless of which answer you were hoping for — and the verdict doesn't renegotiate itself to make you comfortable. The motive lens flagged -3 whether Priya liked it or not, and the system attaches the source verse in its exact catalog wording rather than paraphrasing something to match the mood you walked in with.

The takeaway: the day-7 flinch test

You don't need us for this. When you make a real call this week, schedule three check-ins and ask one fixed question at each: "Reading this back, do I flinch?"

• 24 hours: the relief window. A good call usually feels lighter here — the churn stops. A bad one feels like you got away with something. Note which it is; don't argue with it.
• 3 days: the story window. Watch the sentence you tell other people about what you did. If it's growing new justifications, the decision is defending itself — which means part of you is still prosecuting it.
• 7 days: the residue window. The 24-hour relief has worn off and the story has settled. Read your own decision back cold. A clean action leaves no residue. A compromised one leaves a small, specific wince — and the wince always points at the lens you overrode.

Priya's punishing reference passes the 24-hour test easily: revenge feels great for a day. It fails at day 7, when the wince lands exactly on motive. That's the metric. Not "was it smart," but "did it still feel right when the satisfaction wore off."

Run one real decision through the door and set the three reminders yourself. You can browse how other calls held up in the gallery while you wait for your own day 7.

When the relief fades next week, which lens will your wince be pointing at?