Every official Rubik's Cube competition on record: people and timed 3×3 solves. The world-record average got four times faster. The average a first-timer brings to their first competition has barely moved, and half of them never come back.
The world-record average against the average each year's newcomers set in their very first round.
Every line but the blue one falls. The record comes down in steps: Feliks Zemdegs held it for most of the 2010s, then Ruihang Xu, Max Park, Tymon Kolasiński and Yiheng Wang took it under five seconds, and in June 2026 Wang took it to . The median competitor gets faster too, from about 31 seconds to 20, because people who stay keep practising and the people who stay are most of each year's field.
The newcomers don't move. Each year's first-timers bring an average of about 35 to 40 seconds, and they did in 2008 when the record was 11 seconds. The spread is the same too: a quarter of them take longer than 50 seconds. The fast tail has improved — the quickest tenth of newcomers now arrive under , where in 2008 it was 21 — but most people's first competition is still the first time they have solved a cube with a judge watching.
One year breaks the pattern. The few who came to a first competition in 2021, when events restarted after the pandemic, were about seven seconds quicker than usual. Only the keenest turned up.
Of everyone who entered a first competition at least two years before the export, the share who entered another, in any event, within two years.
2019 and 2020 newcomers mostly had no competitions to return to: the pandemic shut them down for up to two years.
Only of first-timers enter a second competition within two years. Speed makes a big difference. Two in three people who arrive averaging under fifteen seconds come back. About a third of those who take one to one and a half minutes do, and a quarter of those slower than that.
The very slowest are the exception. People averaging over three minutes, and people who don't finish an average at all, come back more often than people at a minute and a half. The WCA doesn't publish ages, so the data can't say why. A reasonable guess is that many of them are young children whose parents bring them back.
First averages, grouped by how many competitions each person went on to enter, and how quickly the people who stayed improved.
Bar: middle half. Dot: median. People whose first competition was at least three years before the export, so every group has had time to fill.
People who competed once and never again started at a median of . People who went on to twenty or more competitions started at . The future regulars were about sixteen seconds ahead on their first day.
They do improve, and quickly at first. Among people who entered at least ten 3×3 competitions, the median personal best falls by a fifth between the first and the second, and keeps falling. By their fortieth competition the most committed are under ten seconds. So when a WCA field looks faster every year, part of that is practice and part of it is who stays. The data can't tell whether the people who left would have improved as quickly. It only shows that they started slower.
Type an average (seconds, or m:ss) and see it against everyone who has ever set one.
Times from 3.51 (the world record) to 10 minutes. The page keeps nothing you type.
Densities on a log time axis, built from 400 quantiles of each distribution. Your time's position among personal bests counts everyone with at least one completed average; a DNF-only record is left out.
First averages of newcomers in every country that has had at least 500 of them. Coloured by continent, one sticker each.
Newcomers in Vietnam arrive with a median first average of , and in Brazil . The United States, with the most first-timers of all, sits in the middle. Asian countries take eleven of the thirteen fastest places.
This is more likely to reflect who a country's competitions attract than how talented its cubers are. Where competitions are rare, mostly committed cubers make the trip. Where they are common and easy to enter, people who learned last month enter too. The data can't separate the two. India, Brazil, Bolivia and Mongolia are at the slow end. The return rate in the right-hand column hardly tracks speed at all. Vietnam has the fastest newcomers and one of the lowest return rates, 39%.
Every average-of-five 3×3 round without a cutoff, so all five solves are always attempted. If order didn't matter, each column would read 20%.
The first solve is more likely to be the slowest of the five, and the last is more likely to be the fastest. People warm up during a round. The difference is small, about two rounds in a hundred, but across rounds it isn't chance.
The fifth solve is also the one most likely to go wrong: a DNF rate against about 2.0% for the other four. That could be a time limit running out, but time limits catch slow solvers, and slow solvers show no jump at all. It's the fastest solvers who DNF the last solve more. Among people whose first four solves averaged under ten seconds, the DNF rate goes from on the fourth solve to on the fifth.
By the fifth solve a competitor knows exactly what they need for a personal best or the next round. Fast solvers seem to take more risks on it, and more of those solves fail.