The five-times wall: where size estimates stop working
Two things about the same size are easy to rank. Two things a factor of three apart are still easy. Somewhere around a factor of five the judgement quietly stops being a judgement and becomes a shrug, and the size of the shrug does not grow with the size of the gap.
The observation
Ask someone to rank two things by size and they do well when the ratio is small. Ask them to say how much bigger and the answer degrades in a specific way: it is roughly right up to a factor of about five, then it stops growing with the real ratio.
Beyond that point people reach for the same handful of expressions regardless of the actual gap. Enormously bigger. Nowhere near. Not even close. Those phrases cover a factor of eight and a factor of eight hundred equally, which means they carry almost no information.
Below the wall
The comparisons that work are the ones where the ratio is small enough to be held as a number.
A grizzly bear upright at 2.5 m against a person at 1.71 m is 1.46. That is a ratio anyone can picture: the bear's head is about where you would be if you stood on a low step. A giraffe at 5.3 m against the same person is 3.10, still workable: three people standing on each other's shoulders.
A killer whale at 7.3 m against a person is 4.27, and this is where it starts getting shaky. Four people lying end to end is already an image most people have to construct deliberately rather than see.
The catalogue's own evidence: of the 180 comparison pages on this site, the ones that read as surprising cluster between ratios of 1 and 3. Above about 6 they stop being surprising and start being merely large, which is a much weaker reaction and a much less useful one.
Above the wall
Past a factor of five, the sentence a person produces stops depending on the number.
A sperm whale at 16 m against an African elephant at 3.2 m is exactly 5.0, right at the boundary. The Burj Khalifa at 828 m against a 119.8 m Godzilla is 6.91. The Eiffel Tower at 330 m against the Millennium Falcon at 34.75 m is 9.50. Asked to describe those three gaps without the numbers, most people produce the same three words for all of them.
Push further and nothing changes. A sponge character at 10.16 cm against a plankton character at 1.27 cm is exactly 8.0. Alaska at 3,635.8 km east to west against Vermont at 254.7 km north to south is 14.3. Both get called huge, and one is nearly twice the other as a ratio of ratios.
Why the wall is where it is
Two explanations, and they are probably both true.
The first is counting. Most people can hold four or five items in mind at once without effort and lose track above that. A ratio under five can be checked by imagining that many copies lined up. Above five the check becomes arithmetic, and arithmetic is not something the visual system does.
The second is experience. Almost nothing in ordinary life presents two objects a factor of ten apart in the same field of view at a distance where both are legible. You see a person and a car, a car and a building, a building and a hill. Each step is small. The large ratios are always assembled from chains, never seen, and an assembled ratio does not carry the same conviction.
Working around it
The chain is the fix, and it is worth being explicit about how to build one. Split a large ratio into steps under five and give each step an object.
From a wine bottle at 30 cm to the Burj Khalifa at 828 m is a factor of 2,760. As a chain: bottle to person is 5.70; person to a giraffe is 3.10; giraffe to a sperm whale is 3.02; sperm whale to Elizabeth Tower is 6.0; Elizabeth Tower to the Burj Khalifa is 8.6. Five steps, three of them under five, and the whole thing is now walkable even though the endpoints are not.
The other fix is a shared baseline, which is what every drawing on this site does. Two shapes standing on one ground line convert a ratio into a length comparison on a flat surface, and that is a task the eye is genuinely good at even at ratios of twenty or more. You will not be able to say the number. You will be able to see it, which is different and sometimes better.
What this means for the game
The wall is the reason the pairs in the daily set are drawn from combinations that have been checked rather than picked at random. A round pitting an ant against Saturn would be technically valid and completely pointless, because your guess and everybody else's would be a shrug, and a shrug does not distinguish a good player from a bad one.
Keeping the ratio inside a range where the judgement is real is what makes a score mean anything. The interesting rounds are the ones at ratios between about one and four, where the answer is genuinely uncertain and genuinely knowable. That is a narrow band, and most of the 855 rounds published so far sit inside it.