Our Score Bands Claimed a Rarity Our Own Catalog Does Not Support

Two of our machine-readable documents describe how to read a SleekNova score. They list the four bands a published score can fall in, and beside each band one of them added a rarity: this band is the top one percent, that band is the top five percent, and so on down.

Three of those four rarities were wrong, and one of them was wrong by a wide margin.

What the numbers actually are

We do not publish a product that scores below eight. That floor is the whole point of the Lab, and it means the products on this site are not a cross-section of anything. They are the survivors of a filter.

Counted across every published product carrying a score: none sit in the top band. Seven sit in the band below it. Ninety-three sit in the band below that. Two hundred and eighty-four sit in the lowest published band. So the band we described as the top quarter is, in fact, three quarters of everything we publish, and the band we called the top fifteen percent begins at the twenty-sixth percentile.

How a true sentence became a false one

The rarities were written when the catalog was small and they were a reasonable description of it then. Nobody edited them afterwards, and nobody needed to. Every product we added moved the distribution underneath a sentence that stood still.

This is the second correction we have published this month with that exact shape, and the repetition is the useful part. A number that is computed fresh on every page load cannot go stale. A number typed into a document by a person is correct on the day it is typed and drifts silently from then on, and nothing about the document looks different afterwards.

What we changed

We removed the rarity claims. We did not replace them with computed ones, because a percentile of a deliberately filtered catalog tells a reader very little that the band itself does not already tell them, and we would rather publish nothing than publish a figure whose meaning we cannot defend.

In the same change, the four bands themselves stopped being written down in three separate places. They now come from one definition in our code, and every document that states them reads that definition. The published wording did not change by a single character, which is how we know the move was faithful. What changed is that there is no longer a copy that can drift.

What a reader should take from this

A score of 8.2 on this site means what our methodology page says it means: a weighted total across four dimensions, above the floor we publish at. It does not mean the product beat some percentage of a field. We should not have implied otherwise to a machine reading our documents, and we no longer do.