We assessed 607,349 domains and expected a bell curve. What came back was two mountains with a valley between them, and the shape turns out to say something about how the modern web gets built.
The distribution is bimodal
A population of anything usually clusters around its average. This one does not. The largest group scores above 90. The second largest scores below 10. The middle — the 40 to 60 range where you would expect most of the mass — is the thinnest part of the whole chart.
The average is 62.9 out of 100, and it describes almost nothing. Very few domains actually score near it. It is the arithmetic centre of two populations that have little to do with each other.
You can see the shape on the statistics page, which is recomputed hourly.
Why the middle is empty
The two masses are not two kinds of quality. They are two kinds of intent, and the signals they leave barely overlap.
A domain built to last accumulates evidence almost as a by-product. It gets renewed. It ends up with a real organisation attached because someone had to sign something. It acquires hosting that is not shared with a thousand strangers, links from places that check, an age that grows on its own. None of that is a security effort — it is the residue of existing.
A domain built to be discarded accumulates none of it, because accumulating it costs money and the whole plan is to be gone in six weeks. It is new, anonymously registered, on shared hosting, and unknown to everything that keeps records.
The middle would be a site that has some of this and not the rest. Those exist — they are the genuinely new business, or the old site that has been neglected — but there are far fewer of them than of either extreme.
What that means for reading a score
It makes the number more useful than it would be otherwise. When a population is bimodal, most individuals are not ambiguous: a domain is usually clearly in one group or the other, and the score is telling you which.
It also means the interesting cases are the rare ones. A score of 55 is genuinely worth reading the findings for, in a way that 96 or 4 usually is not. That is the opposite of how most people use a score, and it is why every report shows its reasoning rather than a verdict alone.
The uncomfortable part
112,365 domains — roughly one in five of everything assessed — sit in the bottom band. They are not obscure corners of the web; they were reachable enough to be collected, which usually means somebody was pointing traffic at them.
Nothing about that is inevitable. It is a consequence of registration being cheap, identity being optional, and both of those being deliberate policy choices. The difference between address endings makes that particularly clear: where registration is cheap and unchecked, the bottom band is the majority.