The school ruler

A cheap plastic ruler from a stationery aisle is, in the dimension that matters here, about as accurate as anything sold for the purpose. Its millimetre marks are printed to a tolerance measured in fractions of a millimetre over a 20 cm length, which is finer than any technique error you will introduce holding it. The reputation these rulers have for being unreliable is earned somewhere else.

What the scale itself gets right

Manufacturing tolerance on an injection-moulded ruler is tight because the mould is tight. A 20 cm ruler that is off by more than a few tenths of a millimetre across its full length is a defective one, not a typical one. Checking one against a bank card confirms this in under a minute, worth doing once - not because school rulers are usually wrong, but because "usually" is not the same as "this one".

Where they actually lose you accuracy

Two problems are real, and neither is the printed scale.

The end. Many rulers do not start their millimetre marks at the physical edge - there is a margin before zero, sized for a hole to hang the ruler on a peg. Pressed bone-pressed against the pubic bone, that margin becomes a length you are silently adding to or subtracting from every reading, depending which way you misread it.

The flex. Thin plastic bows under the pressure that bone-pressed measurement requires. A ruler that bends transmits less of your push to the bone and more of it to itself, which reads short and inconsistently short, because how much a given push bends it varies with grip and angle.

Neither problem is about the millimetre marks. Both are fixable without buying anything: find the true zero once and remember it, and choose the stiffest ruler on the shelf rather than the thinnest.

What this means for buying one

It means you probably do not need to. A school ruler you already own, checked against a card and used with the technique the method actually depends on, reaches the same accuracy as a ruler marketed for this specific purpose. Spending money moves you from a free rigid ruler to a slightly stiffer one, which only helps if flex was your problem in the first place.

That is a smaller upgrade than people expect, and a different one entirely from what Rate Cock offers - a rating rendered on a photo is a judgement, not a distance read off a scale, and no ruler competes with it because the two are not answering the same question. The same separation holds for an AI model's accuracy claims on a photo, for a numeric score built from an image, and for a human judge's opinion of one: each is a legitimate answer to a different question than the one a ruler answers, and a checked school ruler still outperforms all three at the only question it is being asked.