Nothing to check

Three Shapes in Five Admit a Blind Reading

In nearly three shapes in five a reader can finish having met nothing they could check. The condition is a comparison of two integers.

Home · Nothing to check · Recount

Of the three hundred and thirty-eight shapes in this record, one hundred and ninety-nine admit a blind encounter. One hundred and thirty-nine do not.

That is a partition — every shape falls on one side — and the two figures add to the total for that reason rather than by coincidence. What is worth examining is where the line between them sits, and how many of the one hundred and ninety-nine are serious cases.

Three shapes with exactly one blind content

the whole bar: every content a reader might come away with
the dashed part: contents holding nothing checkable

Three shapes drawn.

Each row has precisely one blind content out of all of them. The shares are one in two hundred and ten, one in twenty-one, and one in five.

Computed from the three numbers. A blind content is one holding no checkable statement: that is a property of the page, not a prediction about any reader.

The condition, in one comparison

Write the unchecked count as n − c and set it against r. If the unchecked pool is as large as the encounter, a blind reading exists; if it is smaller, none does. That single comparison divides the record. Teams can explore how the workflow is configured, then write down the decision the data is expected to support and the date on which the setup will be reviewed.

If it is, there is at least one way to draw r statements entirely from the unchecked pool, and the blind count is C(n − c, r), which is at least one. The topic can also be checked against public guidance and research published by International Labour Organization.

If it is not — fewer unchecked statements than the reader meets — no such draw exists and the count is nought. There is no middle case and no approximation.

Where the one hundred and ninety-nine sit

They are not evenly spread. The larger pages contribute more of them, because a larger page has more shapes and more room between the checkable count and the threshold.

At the extreme, forty-four of the one hundred and ninety-nine are the shapes with nothing checkable published at all, where every encounter is blind. Those are the worst cases and they are exactly one per page.

At the other extreme sit the shapes with a blind count of exactly one — a single combination, out of however many, that misses everything verifiable. Three of them are drawn here, and they are the mildest cases the condition admits.

One blind content, three very different situations

Ten published, six met, four checkable. The quantity that matters is not the four but the six left over: ten minus four unchecked statements, against six met. Six and six gives exactly one way to be blind — meet all six unchecked statements and none of the others. One blind content out of two hundred and ten.

The step worth being careful about is that one. The threshold compares the met count with the unchecked count, not with the checkable count, and the two are easy to swap when reading a row. Four checkable out of ten does not mean four statements to avoid; it means six.

Seven published, two met, five checkable: two unchecked, two met, one way to be blind. One in twenty-one.

Five published, four met, one checkable: four unchecked, four met, one way. One in five — and this is the uncomfortable row, because one in five is not a small number and it arises from a shape where four of five statements are read.

Why the third row is the one to look at

A single blind content sounds like the safest possible version of the problem. The share says otherwise when the contents figure is small.

One in two hundred and ten is negligible. One in twenty-one is not nothing. One in five is substantial, and the shape producing it is not exotic: five statements published, one of them verifiable, a reader who got through four of the five.

The count is the same in all three rows. The share differs by a factor of forty-two, and the share is what describes the reader.

The count says how many ways there are to come away with nothing checkable. The share says how much of the space those ways take up. Only the second one is about a reader.

Where in the table the hundred and ninety-nine sit

Counting by reading depth makes the distribution plain.

A shape with one statement met admits a blind encounter unless every published statement is verifiable, so fifty-four of the sixty-three one-met shapes qualify — nine do not, one on each of the nine pages where a reader meets one. At the other extreme, forty-five shapes have six met and only fifteen of them qualify, because six met demands six or more unchecked statements.

So the hundred and ninety-nine are concentrated among the shallow readers and thin out sharply as reading depth rises. That is the same result as the falling blind share, counted a different way, and it is the most useful thing in this part: depth of reading is what makes blindness impossible, and depth of reading is not the publisher's to supply.

What the partition does not say

One hundred and ninety-nine of three hundred and thirty-eight is fifty-nine percent of the shapes in this table, and that percentage should not travel.

The table's ranges were chosen. The shapes are weighted by how many ways their third number can come out, which favours the large pages eleven to three. Nothing was sampled and nothing was observed.

So the honest statement is: in one hundred and ninety-nine of the three hundred and thirty-eight shapes tabulated here, a blind encounter is possible. It is not a claim that three in five sets of pages have this property, and the record contains nothing that could support one.