Should I stay or leave? What eight lenses see

The question you keep almost asking

You have run the same conversation with three friends this month. Each one told you what you already suspected they would. The one who likes your partner said give it time. The one who never did said you knew months ago. Nobody scored the thing. Everybody just handed your own lean back to you, warmer.

So the question sits where it started: stay, or leave.

The dilemma

What follows is a composite — stitched from the kind of stay-or-leave case we see often in this category, with no real person's details.

Priya and Dev have been together three years. Nothing is on fire. There is no affair, no shouting, no single betrayal to point at. Dev is kind in the ways friends can see. He remembers her sister's birthday. He is easy at dinners.

Here is the ugly detail Priya has not said out loud: she keeps score. Quietly, constantly. She tracks who apologized last, who planned the last trip, who bent on the apartment lease. She is almost always "ahead" in the ledger, and she knows it, and she uses it. When Dev falls short, she doesn't raise it — she banks it. The relationship runs on a debt he can't see and can't pay down, because she never tells him the balance.

Dev thinks they are fine. That is the part that makes her either want to leave or want to confess, and she cannot tell which.

She has framed it to herself as "we've grown apart." That framing is doing a lot of work. It skips the fact that she has been managing him instead of being honest with him. Leaving would spare him a conversation. Staying, as she is currently staying, spares him nothing — it just hides the invoice.

So: stay or leave. Both feel like exits.

The read

We score every dilemma on eight fixed lenses. Four of them bite hard here.

Honesty — the state of the account, not the volume of words

Honesty asks whether your version of events matches what you're doing. Priya's does not. "We grew apart" is a story that arrived after the scorekeeping, to make the scorekeeping unnecessary to mention. She is fluent and warm and completely undisclosed. Dev is agreeing to terms he's never seen. On honesty, the current arrangement scores -3 — not for a lie told, but for a truth withheld as a management tool.

Non-harm — sparing someone pain is not the same as not harming them

Non-harm asks who absorbs the damage. The intuitive read: leaving hurts Dev, so staying is the kinder call. But staying as she is keeps him inside a relationship whose real rules he doesn't get a vote on. That is a slower harm, and a quieter one. Silent departure — leaving without ever naming the ledger — scores -2: it ends the harm but launders it, so he never learns what actually happened to them.

The tension you came here for

This is where non-harm and honesty stop agreeing. The comfortable move — leave gently, spare him the scorekeeping conversation — looks like non-harm. But it buys that comfort by keeping him permanently in the dark, which is honesty at its worst. And the honest move — sit Dev down and show him the balance she's been running — looks cruel in the moment, and scores well on honesty precisely because it hands him back a vote. Two lenses, opposite verdicts, same fork. Your friends resolve this tension by picking the lens that flatters your lean. We don't get to.

Motive — what is the leaving actually for?

Motive asks what's driving the action underneath the reason you'd give. If Priya leaves to avoid the confession, motive scores -2: the exit is a way to never be seen scorekeeping. If she leaves after telling the truth — or stays and dismantles the ledger — motive climbs, because the action stops being about her own concealment.

Restraint — the ledger itself

Restraint asks whether the reactive habit is running the relationship. The scorekeeping is a reflex she feeds daily. It scores -2: not evil, just an ungoverned pattern that has quietly become the operating system.

The verdict: the choice is not stay-or-leave — it's disclose-or-don't, and every ethical lens points at the conversation she's using both stay and leave to avoid.

A general AI chatbot reads your lean in the first two sentences and quietly starts building the case for it — then builds the opposite case just as warmly the moment you push back. That eager agreement is a documented failure mode; the model is tuned to keep you happy, not to hold a line. KarmaLens runs the other way. The eight lenses don't move, the scores add up the same way every time, and the verdict won't soften just because you flinch. It hands you a call that survives your disagreement — the one thing a friend optimizing for your comfort structurally can't do.

The takeaway you can run tonight

You don't need us for the first move. Try the invoice test.

Write down, in one sentence, the thing about your relationship you have never said out loud to the person in it. Not to your friends — to them. If nothing comes, you may genuinely be in a stay-or-leave question. If something comes fast, you are in a disclose-or-don't question wearing a stay-or-leave costume, and the real fork is whether that sentence ever gets said.

Then score just two options — leave-without-saying-it and say-it-then-decide — on honesty and non-harm alone. Watch which cell makes you flinch. The flinch is information. It usually points at the exit you've been calling kindness.

Most stay-or-leave paralysis is not indecision. It is a decision to keep an invoice hidden, dressed up as thinking. When you want the full eight-lens read on your own version, bring it through the relationships door and get a committed call instead of a mirror. You can also read other worked verdicts in the gallery.

What is the sentence you've never said to the person you're deciding about?