How it works

How the FlagX ranking works

FlagX asks one question, thousands of times a day: of these two flags, which is better? Every answer is a single tap, and every tap feeds a live global ranking. This page explains, honestly and completely, how those taps become a leaderboard, because a ranking you can't inspect is just an opinion with a scoreboard attached.

The duel

Each session is ten rounds. In every round, two flags are drawn so that no flag appears twice in one session, and you tap the one you prefer. There are no criteria imposed: design purists judge geometry, romantics judge history, some people just like green. The ranking doesn't care why you tapped; it only records that you did. After each vote you see what share of the world agreed with you, based on all votes cast on that same pairing during the current window.

The weekly board: Wilson scoring

The main leaderboard covers a rolling seven-day window. For each flag we count its head-to-head wins and total appearances in that window, then rank flags not by raw win rate but by the lower bound of the Wilson score interval at 95% confidence. In plain language: a flag that won 9 of 10 duels does not outrank a flag that won 850 of 1,000, because ten duels is luck and a thousand is evidence. Wilson scoring rewards being both good and tested, which keeps small-sample flukes off the top of the board.

The long game: Elo ratings

Alongside the weekly board, every flag carries a persistent Elo rating, the same system chess uses. All flags start at 1000. When two flags meet, the winner takes rating points from the loser, and the amount depends on expectations: beating a giant pays well, beating a minnow pays almost nothing. We display ratings with one decimal so you can watch them move in near real time. Elo answers a different question than the weekly board: not "who is hot this week" but "who has proven strongest across all history".

The home-bias problem

Everyone loves their own flag. That's the point of flags. But if patriotic votes counted fully, the ranking would simply mirror the population and enthusiasm of each country's internet users. So FlagX applies one deliberate correction: when a voter's own country's flag is in the duel (we infer country from the network connection, approximately and anonymously), that vote counts at 20% weight. Your pride is recorded; it just can't drown out the neutrals. Additionally, your very first session ends with a special matchup featuring your home flag, and that pairing is excluded from ratings entirely: it exists for the drama, not the data.

Keeping the votes honest

A ranking is only as trustworthy as its defenses. FlagX validates every submitted session server-side: each session carries a cryptographically signed token that can be spent exactly once, sessions completed faster than a human could physically tap are rejected, each network address has a daily ceiling, and an invisible bot check challenges automated clients without bothering humans. Votes are stored only as anonymous aggregates, never as personal records; the privacy page covers this in detail.

The weekly champion

Every Monday at 00:00 UTC, the flag on top of the weekly board is crowned champion and enters the permanent hall of champions. During the current calibration phase (see Season 0), the first crowning waits until the arena has collected enough votes for the title to mean something.

Why trust this more than a listicle?

Because nobody wrote this ranking. Every "best flags in the world" article is one editor's taste. FlagX is an experiment in doing it the other way: define fair rules in public, defend them against cheating, and let the answer emerge from whoever shows up to vote. The result changes hourly, argues with itself, and belongs to no one, which is exactly what makes it worth checking tomorrow.