The eight lenses: how the Gita scores a decision

You already know which answer you want. You are scrolling for a reason to keep it. Most advice obliges — it finds the one angle where your plan looks noble and hands it back to you gift-wrapped. The Gita does something less comfortable: it makes you look through eight windows at once, and refuses to let you close seven.

The eight lenses, in plain sentences

Every dilemma submitted to KarmaLens gets scored on the same eight lenses. Not the ones that fit. All of them. Each runs from -5 (this action badly violates the lens) to +5 (this action strongly honors it), with 0 meaning neutral or not in play.

• Duty — the obligation your role actually carries, not the one you wish it did. Example: you promised the junior hire you'd review her code before the release; skipping it to hit your own deadline scores low here.
• Honesty — whether the action keeps your words and reality pointing the same way. Example: "I never got the email" when you got the email and archived it.
• Non-harm — the damage the action does to living people, including the quiet, deferred kind. Example: a layoff announced by group message the night before a holiday.
• Detachment — whether you can act without the outcome owning you. Example: sending the honest feedback and not refreshing your inbox for the reply.
• Motive — the real engine under the action, not the caption you'd give it. Example: "for the team" when it's for the credit.
• Restraint — whether desire or anger is driving instead of riding along. Example: the text you want to send at 1am about who was right.
• Wider welfare — everyone downstream who never gets a vote. Example: the customers, the two teammates, the person who inherits your shortcut.
• Discernment — whether you've actually thought through consequences and your own capacity, or are calling a reflex a decision. Example: quitting on Monday's frustration and calling it courage.

Each lens gets its own entry in the eight lenses. None of them can be voted off the panel.

Why the lenses are fixed

Here is the whole trick, and it is not a trick. If you choose the lens, you choose the verdict. Someone breaking a promise reaches for wider welfare ("it's better for everyone"). Someone hoarding credit reaches for duty ("I did the work"). Fixed lenses take the steering wheel out of your hands. You must be scored on motive even when you'd rather only be scored on duty — and that is exactly the moment the score gets useful.

The read: watch two lenses disagree

The dilemma (a composite of the kind we see in this category, not a real person): Priya manages a five-person team. One report, Sam, is slipping — missing standups, shipping late, coasting on goodwill he earned two years ago. The company is quietly trimming headcount and has asked managers to "flag underperformers by Friday." Sam confided last month that a family situation is eating his time and he's barely holding on. Priya can flag him honestly and protect the other four's workload — or stay silent, absorb his slack herself for a quarter, and bet the situation resolves. Nobody would know she softened the report.

Watch four lenses bite.

Honesty (score: -3 if she softens the report)

Honesty asks whether her words to leadership match what she knows. Reporting Sam as "on track" is a clean falsehood with her signature on it. That is a real hit — not fatal, but not zero.

Non-harm (score: +2 for softening)

Non-harm sees the person about to lose income during a family crisis. Silence buys Sam a quarter. Concrete, immediate care for a real human being — this lens rewards the soft report.

Duty (score: -2 for softening)

Duty sees her obligation to the other four, who are already covering Sam's gaps, and to a company that asked for straight information. Absorbing his work herself for three months is generous, but it quietly taxes people who never agreed to pay.

Discernment (score: -2 for softening)

Discernment asks whether "stay silent and hope" is a plan or a flinch. "I'll absorb it and bet it resolves" has no checkpoint, no honest conversation with Sam, no capacity math. It's kindness improvised to avoid a hard sentence.

Now the tension is out in the open. Non-harm (+2) and honesty (-3) are pulling in opposite directions, and that conflict is the entire dilemma — not a bug to smooth over. The soft report protects one person by deceiving several. The interesting move isn't picking a side; it's the third option both lenses would sign: flag Sam honestly and attach the context and a support plan, so the report is true and the harm is named instead of hidden.

Verdict: the honest flag with full context scores higher than either silence or a bare flag — truth and non-harm stop fighting the moment she stops choosing between them.

This is where KarmaLens parts ways with asking a general AI chatbot the same question. Lead a chatbot toward loyalty and it tends to build you a case for silence; lead it toward integrity and it tends to build you a case for the flag — matching the user's lean is a behavior people observe in these systems again and again. KarmaLens runs the same eight lenses in the same direction every time and aggregates them deterministically into one committed call, so the verdict doesn't renegotiate itself to match the way you framed the question, and it cites the source verbatim from the Gita rather than paraphrasing whatever tone you brought. You don't get eight windows so you can open your favorite; you get eight windows so the one you were avoiding is already open.

The takeaway: run the two-lens flinch test tonight

You don't need us for the smaller version. Take your live decision and score it on just two lenses out of the eight: honesty and non-harm. Give each a rough number from -5 to +5.

1. Write your leading option in one sentence — the thing you're already leaning toward.
2. Score it on honesty. Does your action keep your words and reality aligned? Flinch = negative.
3. Score it on non-harm. Who absorbs the cost, including the quiet, later kind? Flinch = negative.

The reveal is in the gap. If both scores are positive, your instinct is probably clean — proceed. If they point in opposite directions, you've found the actual dilemma, and the useful question becomes: is there a third option that stops making me trade one for the other? Priya's honest-report-with-context was exactly that third option, hiding behind a false binary. Most of the decisions that keep you up have one — you just haven't scored yourself into finding it yet.

Curious how all eight land on your specific situation instead of two? Walk through the full read on your own dilemma, browse worked examples in the gallery, or keep reading the eight lenses series. Which lens were you hoping we wouldn't score?