Stay vs. leave, side by side: two options, eight lenses

You have made a list. Two columns. Pros on the left, cons on the right.

By the third rewrite the columns balance perfectly, which is exactly what a pro-con list is built to do. The tie is the trap. What your list can't produce is the one row that matters: the difference between the two options, lens by lens, where they actually diverge.

The dilemma

This is a composite of the kind of stay-or-leave case we see constantly in this category — invented, but assembled from the real shape of them.

Priya has run a four-person product team inside a mid-size company for three years. A rival firm has offered her a bigger title, a 30% bump, and a team she'd build from scratch. The offer letter expires Friday.

Here is the ugly detail she doesn't say out loud. Two of her four reports were hired on her promise that she'd "protect this team no matter what." One of them turned down another job because of that promise, eight weeks ago. Her current company is stable but slow; her boss has told her, vaguely, that a reorg is "probably coming" in Q3. She doesn't know if her team survives it. She suspects it doesn't.

So the two options aren't clean. Leave: take the offer, tell her team the day she signs, let them navigate the maybe-reorg without her. Stay: decline the raise, keep the promise, and hope the reorg either doesn't come or that she can steer it. She wants to leave. She also can't stop hearing her own sentence about protecting them. She's told three friends and gotten three different answers, each one confirming whatever she'd said first.

The read

We score both options on all eight lenses. Here the four that actually bite — and the two that break the tie.

Duty (does the role's obligation get met?)

Leave: +1. Her duty to her own career and to a team she'd build is real, not selfish. But she leaves a stated promise mid-air. Stay: +3. Keeping a promise she made to specific named people is dharma in its plainest form. The score gap here: +2 toward staying.

Honesty (is anyone kept uninformed to make the move easier?)

Leave: +2 — she plans to tell the team the day she signs, which is clean, if late. Stay: -1. This is the surprise. Staying looks honest, but it hides something: she'd be selling her reports a protection she privately believes she can't deliver through a reorg she thinks is coming. The comforting option keeps them uninformed of her own suspicion. The gap flips: +3 toward leaving.

Non-harm (who absorbs the damage?)

Leave: -1 — real disruption to people who reorganized their lives around her. Stay: 0. Staying feels harmless, but a hollow promise doesn't prevent harm; it delays and disguises it. If the reorg lands in Q3 and she couldn't stop it, they get hurt anyway — later, and by then they've stopped looking. Roughly even, tilting toward leave.

Motive (what is actually driving the leading choice?)

Leave: -1 — some of the pull is title and money, and the read is cleaner when that's named rather than dressed up. Stay: -2. Staying isn't pure loyalty; part of it is the wish to keep seeing herself as the manager who never abandons anyone. That's reputation wearing the costume of duty.

The tension, named: Duty says stay. Honesty says the staying is built on a promise she no longer believes. Those two lenses are pointing in opposite directions, and the pro-con list was never going to show you that — it just kept adding rows until the weight felt equal.

The verdict: The clean move is neither silent leaving nor silent staying — it's telling the team what she actually suspects about the reorg first, then choosing, because every version of the decision that only works if they never hear her real assessment fails the honesty lens hard.

Now watch what a general AI chatbot does with the same case: lean toward leaving and it will build you a moving speech about seizing your growth; lean toward staying and it will praise your rare loyalty an hour later. It mirrors the tilt you walked in with — sycophancy is a documented behavior in these systems, not a personal failing of yours. KarmaLens runs the opposite way. The eight lenses are fixed, the scores aggregate deterministically into one committed call, and the verdict does not renegotiate itself when you push back or restate your preferred option. It also cites the source verse verbatim rather than paraphrasing it into agreement. You don't get the answer that flatters your Friday-afternoon leaning. You get the delta row — and the delta row doesn't care which column you were already living in.

The takeaway: the two-lens delta test

You can run the core of this tonight, on paper, without us.

1. Write your two options as verbs, not moods. "Take the offer and tell them Friday." "Decline and keep the promise." Named actions, not "leave" and "stay."
2. Score each option on just two lenses: honesty (is anyone kept uninformed to make this easier?) and non-harm (does this prevent harm, or just delay and hide it?). Use -5 to +5.
3. Now subtract. Write the difference for each lens as a single row: "Honesty: +3 toward leaving. Non-harm: +1 toward leaving."

That delta row is the artifact your gut cannot produce, because your gut scores the whole option at once and calls it a tie. When the two lenses disagree — one favors staying, one favors leaving — that disagreement is the finding. The option that scores worse on honesty is almost always the one that felt kinder, because it kept someone from knowing something. That's the flinch. Chase it, and you usually find a third action that stops the trade-off — the disclose-first move that neither original column contained.

When you subtract your two options lens by lens, which row are you hoping doesn't say what you already suspect it says? Put your own stay-or-leave through KarmaLens and see the full eight-lens delta.