Methodology

A percentage on Optants is arithmetic over votes that real people cast. There are no seeded tallies, no simulated participants, and no placeholder figures anywhere in the product. A new question launches at zero and stays there until someone votes.

This is a constraint in the codebase, not a policy we remember to follow: totals are derived from the vote ledger at read time, so there is no counter for anyone to set by hand.

Voting discloses nothing. You can choose to say how you voted — by weighing in, which publishes your comment with the option you picked — and that option is only ever offered after you have voted and seen the result, so the possibility of disclosure never hangs over the choice itself. An unobserved choice is what keeps a margin honest.

You do not need an account to vote, and you never will. We will never ask who you are as the price of a preference — no name, no email, no signing up, and nothing that ties a ballot back to a person.

We may check that a voter is a distinct person, and that is a different thing from knowing which person. Any such check answers one question — is this one human — returns nothing that can be joined to a ballot, and is described on this page before it ships rather than after.

The sides are dealt evenly, too. Which option sits on the left is fixed for each matchup and the same for everyone. It is set by a coin flip, never by the order we happened to write it in, and in an arena each contender is on the left in about half of its matchups — so neither side inherits the small advantage that comes with being read first.

Optants has no rating system. There is no ELO, no skill estimate, no Bayesian smoothing, no recency decay, no confidence adjustment. Where we show which option is ahead, that is a direct count of head-to-head choices within a defined pool — and it is recalculated from the votes each time it is read, never stored as a score.

An arena’s board works the same way. Each contender scores one for every rival it beats in their own matchup, and that count sets the order. Contenders level on wins are separated by the matchups between them; where those cannot separate them, they are shown as tied. A board appears only once every matchup in the arena has at least 24 votes.

We also never infer a comparison nobody made. If two options have not been put against each other, we do not estimate what would have happened.

The same restraint applies to people. Your record here is a ledger of things you did — questions answered, positions held — never a reputation number, karma, level or rank.

There is no recommendation engine and no profiling of your interests. Everyone’s feed is built by the same rules: editorial prominence — our own judgment about what is worth asking, a single global setting — decides what comes first, and among questions of equal standing the order is shuffled at random.

Three things change that, and only for you: subjects you have asked to see more or less of, arenas you have muted, and what this browser has already shown you, which moves to the back instead of coming round again. All three are kept in your browser. Nothing we infer about you changes what you see.

Where we publish a result broken down by group, every figure comes from information people volunteered about themselves. We never infer who you are from your device, your address or your behavior, every field is optional, and none of it is ever a condition of voting.

A breakdown is published only when every cell within it clears a minimum of 50 votes. Below that the slice is withheld entirely rather than shown thinly — small cells are how an aggregate stops being an aggregate and starts identifying somebody. That floor is enforced inside the database, so no page, query or future feature can quietly lower it.

One vote per question, per voter — enforced in the database rather than by trust. A unique index and a single-pick check reject a second ballot outright; there is no path through the site, or around it, that adds one.

What “a voter” means here, precisely: your account if you are signed in, and otherwise the browser you voted from. Someone determined to vote twice can clear their browser and do it. We would rather say that plainly than claim a guarantee no accountless system can honestly make — because the only way to make it airtight is to demand identity before a preference, and that is the trade Optants exists to refuse. You do not have to tell us who you are to tell us what you think.

Changing your answer replaces your earlier one — it does not add a second. The tally always reflects where people currently stand.

Some questions close on a date we set. After that a question takes no new votes and no changes, and its final result stays up for everyone to read.

If we find votes that were not cast by real people, we remove them — and when a removal moves a published number, we say so: what went, when, and why. Taking out a ballot nobody cast is not an adjustment to the count; it is the count. We would rather correct in the open than defend a total we know is wrong, and we will never remove a vote for being the answer we did not want.

Optants measures the people who chose to answer. It is not a representative poll of any country or population, it is not weighted to match census demographics, and no result here should be read as a projection of what everyone thinks. It is an honest count of a self-selected group — which is exactly what it says it is, and nothing more.