The first scoring system I tested for Magnitudle used simple linear distance: how far is your number from the true answer?
It was immediately obvious this was wrong.
Consider a question where the true answer is 10 million. A guess of 9 million is off by 1 million — extremely close. A guess of 100 million is off by 90 million, which sounds catastrophic. But 100 million is only 10 times larger than 10 million. For an estimation question about a real-world quantity, being 10x off is a decent result, not a failure. Linear scoring completely misrepresents how good that guess actually is.
The problem gets worse when questions span very different scales. A question about city populations might have an answer in the millions; a question about global bacteria counts might be in the quintillions. Linear scoring would make the first question trivially easy to score near-perfectly, and the second one essentially impossible to score well regardless of how thoughtful your estimate was.
So I went back to first principles and built the scoring around order of magnitude instead.
What Is an Order of Magnitude?
An order of magnitude is simply a factor of ten.
If something is one order of magnitude larger than something else, it's 10x bigger. Two orders of magnitude is 100x. Three is 1,000x.
This is the natural unit for measuring how "close" two large numbers are. Whether the true answer is 40 million or 70 million, you're essentially in the same ballpark. But 40 million and 400 million are a full order of magnitude apart — a meaningfully different scale.
Magnitudle scores your guess based on how many orders of magnitude separate it from the true answer. The fewer the orders apart, the higher your score.
What Different Scores Mean
As a rough guide:
A score of 90–100 means you're extremely close — within about a factor of 2 or 3. If the answer is 50 million and you guessed 30 million or 80 million, you're in this range. You had the scale right.
A score of 70–80 means you're about one order of magnitude off. You guessed millions when the answer was tens of millions, or billions when the answer was hundreds of millions. Right territory, slightly undershot or overshot.
A score of 50–60 means you're two orders of magnitude off. You guessed thousands when the answer was millions, or billions when the answer was tens of billions. Your scale was in the wrong league.
Below 50 and you're in the territory of a genuinely surprising miss — where your mental model of the quantity was fundamentally different from reality. These are actually often the most interesting results, because they reveal something unexpected about the world.
Try the scoring yourself
Enter a guess and see how Magnitudle would score it
You can write 5B, 5000000000, or 5e9
Enter a guess above to see your score
How to Actually Improve
The most common mistake I see is confusing millions and billions. Intellectually, most people know a billion is a thousand millions. But under the mild pressure of committing to an estimate, both numbers trigger the same vague feeling of "very large," and the distinction collapses.
One useful habit: when you arrive at an estimate, reality-check it against something you already know. If you've estimated that the US produces 50 billion eggs per year, ask yourself — the US has about 330 million people. That works out to about 150 eggs per American per year, or roughly 3 per week. Does that feel plausible? (For the record, US egg production is around 100 billion, so the estimate is in the right order of magnitude, though slightly low.)
When your estimate leads to an implication that feels obviously wrong, recalculate. When it produces a number that feels right, it's a signal you're probably in the right range.
The second pattern worth watching: questions about global quantities tend to be much larger than people expect. Total water on Earth, number of insects alive, global energy consumption — our intuitions about planetary-scale phenomena are essentially useless, because we've never needed to develop them. If a question is asking about something at a truly global scale, your gut instinct is probably an underestimate.
Why This Approach Is Fair
The scoring system is designed so that thoughtful estimation always produces a respectable result, even without specialist knowledge. You don't need to know the answer to score well — you need to reason about the scale.
For questions that span many orders of magnitude, the log-scale approach also keeps difficulty roughly consistent. A question with a true answer in the thousands is equally challenging to a question with a true answer in the quadrillions, because the scoring reflects how many factors-of-ten you are from the true scale — not the raw numerical distance.
For a deeper sense of why large numbers are hard to intuit in the first place, How Big Is a Billion, Really? is worth reading. And for the estimation method that underpins the questions themselves, see What Is a Fermi Problem?