How Findnami works

The short version: you and your partner each judge pairs of names independently, and a chess-rating algorithm turns those judgments into a shared ranking that's honest about how sure it actually is.

The two-pile problem

Most baby-name conversations stall because they're done out loud, in the moment. One of you says "Léna" and the other says "hm" and that's it — Léna's out, and you don't know if it was the name, the meaning, or the way your partner pronounced it that lost it.

Findnami takes that out-loud conversation and breaks it into small, private judgments. Each of you picks between two names at a time, and neither sees the other's picks until the picture is clear enough to be useful.

Glicko-2 in plain language

Glicko-2 is the rating system used in modern chess and many video games. It tracks two things for every "player" — in our case, every name:

• Rating (μ) — how good the name is so far. Every name starts at 1500. Pick it in a comparison and it moves up; the opposing name moves down.
• Rating deviation (φ) — how sure we are. Every name starts at φ = 350, meaning "no idea yet." The more comparisons it appears in, the smaller φ gets.

Once φ is low enough — about 100 in our setup — a name's rating has settled. That's what powers the "stable enough to stop" hint you'll see on the scoring page.

Two ratings, one consensus

We run Glicko-2 separately for each partner — your picks don't influence your partner's ratings and vice versa. Each name ends up with two μ values, one per partner.

The "consensus score" on the dashboard combines them in two steps:

1. Geometric mean of the two μ's. Geometric, not arithmetic — that means a name only ranks high if both of you like it. A name one of you loves and the other hates ends up middling, not great.
2. Disagreement label: we look at the gap between your two ratings, plus how confident each rating is. That produces one of five labels: strong shared favorite, promising — needs more comparisons, loved by one, uncertain for the other, at risk of being narrowed, and still under-tested.

What you don't see during a round

Numerical scores are hidden while you're comparing. This is deliberate: if you can see that a name is "winning," your next judgments anchor toward it. That defeats the point of judging each pair on its own merits.

Once you both stop, the dashboard reveals the labels — but not the raw numbers. The labels are what's actionable; the numbers would invite arguments.

What about ties and "neither"?

You can also pick Love both, Neither, or Talk later on any pair. Those don't move ratings — they're recorded as evidence so the dashboard knows the pair was seen.

You also get a budget of pre-vetoes — names you can permanently rule out before the narrowing phase. The budget is small on purpose: 10 per partner by default. It's a scalpel, not a hatchet.

Privacy

Findnami doesn't sell or share your data, and runs on EU-hosted infrastructure. You can export everything you've added and request account deletion from Settings.