One of the things I enjoy most about Magnitudle is seeing how people approach questions where they don't necessarily know the answer. Sometimes you're surprisingly close; other times you're wildly off. That's part of the fun, but the score you receive should always feel fair.
Recently, I've been looking at how people actually play across Magnitudle, Size It Up and Weigh It Up. After analysing millions of rounds, I found a few things that weren't quite working as I'd intended, so I've made some changes to how scoring works.
How the updated scoring curves compare
The graph below shows the score each game awards as a guess moves further from the correct answer. A factor of 1× is exact, 2× means twice as high or half as low, and 10× means one tenth or ten times the answer.
Explore scores by guessing error
Hover or tap the graph, or move the control below, to compare scores at the same relative error.
Daily Question: making accuracy matter more
The original Daily Question scoring was a little too forgiving. You could guess ten times the correct answer and still receive 75 points; even being a hundred times out earned 50. That's a decent score for a guess that, realistically, wasn't very close.
Under the updated scoring:
- A guess that's twice the correct answer earns 80 points.
- Ten times away earns 42.
- A hundred times away earns just 5.
You'll still be rewarded for getting into the right ballpark, but really accurate guesses will stand out more. Whether your guess is too high or too low, the scoring treats equivalent errors equally.
Size It Up: fixing an unfair advantage
This was probably the most important change. Previously, Size It Up calculated accuracy using percentage error. That sounds sensible, but it created an unintended imbalance.
Imagine you're estimating something that's actually 10 metres long. A guess of 5 metres would score 45 points, while a guess of 20 metres would score only about 21. Both guesses are equally far away in relative terms - one is half the correct size and the other is double - yet the underestimate was rewarded much more generously.
That's now fixed: both guesses would earn 33 points under the new system. Whether you overshoot or undershoot, you're judged by the same standard across every Size It Up mode.
Weigh It Up: making a tricky game more rewarding
Weigh It Up is deliberately challenging. Working out how many of one object would balance another isn't something most of us do every day, and the data confirmed what some players were probably already feeling: the scoring was a bit too harsh.
The updated scoring gives you more credit for finding the right general scale, even when you don't get especially close. Being twice the correct answer now earns 85 points, for example. There's still a big difference between a good estimate and a wildly inaccurate one, but getting reasonably close should feel more rewarding.
One consistent approach in Weekly 1K
I've also brought the scoring closer together across Weekly 1K. Previously, its numerical questions used a slightly different approach from Daily Question. That was intentional: Weekly 1K questions were designed to be a little more fact-led, while Daily Question leaned more towards Fermi estimation.
With this update, I'm bringing those two approaches closer together. Question rounds, Size It Up rounds and Weigh It Up rounds now use the same scoring rules as their standalone equivalents, so the same accuracy earns the same points wherever you play that type of challenge.
A better way to see how you did
Alongside the scoring changes, result screens now put your rank in context. The familiar grouped score graph remains, and your rank is shown alongside the total number of players who completed that puzzle. For example: "Rank #230 of 2,300 players".
Survival now includes the same rank context. Its rank is based on the number of questions completed in that weekly challenge, and players with the same total share the same rank. Remaining life is not used as a tie-breaker.
It's a small change, but it makes the rank much easier to understand. Getting a score is one thing - knowing how that compares with everyone else's guesses is another.
What happens to your old scores?
Nothing changes about the scores you've already earned. Your previous results, personal bests and statistics are preserved; the updated scoring applies only to new games after the change takes effect.
Archived puzzles use the scoring rules that were active when they were first published. This keeps historical results consistent and ensures your previous scores remain unchanged.
The goal hasn't changed
Magnitudle has always been about having a go, thinking about the world in a different way, and occasionally surprising yourself with how close you can get. These changes aren't intended to make every game harder or easier; they're intended to make scoring more consistent, fair and satisfying.
Thanks for playing, and as always, I'd love to hear what you think of the changes.
Teej
One more thing: what about the Archive?
Archive replays currently keep the scoring rules that were active when each puzzle was first published. That preserves comparability with historical results, but I'm interested in what you would prefer going forward: should Archive replays stay as they are, or use the latest scoring rules instead?