Summary
- ICANN's description of its Domain Abuse Activity Reporting project says third-party observations are turned into abuse rates, and the results are categorised by registrar and by TLD — a comparison between named market participants.
- The record of that page retained on 17 November 2026 carries none of the arithmetic behind those rates: no unit counted, no divisor, no reporting window, no named supplier of the observations.
- This desk has published one earlier item touching DNS abuse, a June 2025 report on the ICANN83 advisory committee's communique, and it does not read the measurement programme.
Start with the person the figure is about. A registrar's compliance lead opens a comparative table of abuse rates, finds the company's own name in it, and wants to know whether the number is fair. There is nothing to argue with. A table ranks; it does not show its working.
In the summarised reading of ICANN's Domain Abuse Activity Reporting project page dated 17 November 2026, the project is described as regularly publishing data on domain name abuse across the gTLD namespace. The observations behind it are described as coming from third-party security threat intelligence providers, which identify domains associated with malware, botnets, phishing and spam. Those observations are described as aggregated and analysed to compute abuse rates, and the resulting reports as categorising abuse by registrar and by TLD, with the stated purpose of greater transparency about abuse trends and about the performance of domain industry participants. The same retained record carries no publisher-verbatim sentence, no rate, no count, no stated unit, no stated denominator, no reporting period and no named supplier.
“Trends” and “performance” are doing different work in that description. A trend is a statement about time: more, less, the same. Performance is a statement about a party, set against other parties. The first invites patience. The second invites procurement decisions, compliance attention and reputation — which is exactly why the arithmetic behind it has to be readable by the party it grades.
Every rate has three limbs, and each can move without a single abusive registration changing. Who observes: a detector decides what counts as abuse, and two detectors looking at the same namespace will not agree on the total. What is counted: domains, incidents and attacks are three different units that produce three different rates from one unchanged reality. What it is divided by: the population forming the divisor decides whether a large operator or a small one sits high on the page.
The retained record addresses the first limb only in the vaguest terms, and it does not address the second or the third at all. A rated registrar that wants to contest its position needs the divisor. Without it, the operator cannot tell whether its figure moved because its abuse changed, because the population it is divided by changed, or because a supplier changed its definition of abuse. Only the first of those three events is about the registrar's conduct.
There is a smaller gap that is easier to miss. The observation suppliers are described as third-party security threat intelligence providers — a category, not a list. A detector is not a neutral instrument; it is a commercial product with a taxonomy, a threshold and a customer base. Publishing the category while withholding the names leaves the rated party unable to check what was measured against it, and unable to see whether a supplier's change of method quietly became its own change of standing.
None of that is an accusation. It is a description of what a comparative number can carry and what it cannot. This piece inspected no report, dashboard, table, ranking or per-registrar figure, and it names no registrar, registry or threat-intelligence company as appearing in one. The ICANN Bylaws, the registry agreements and the registrar accreditation agreement were not read for this article, so nothing here says whether a placement carries a contractual or compliance consequence, or none at all.
The material reviewed establishes the point in one direction only: the page describes rates sorted by named participants, and the retained record of that page does not carry the division.
What this desk had already published matters for scale. One search on the phrase “DNS abuse”, limit 25, against our own archive on 17 November 2026 returned a single item: a report of 30 June 2025 on the governmental advisory committee's communique after the ICANN83 meeting. The directory entry for the organisation recorded no prior articles. Neither reads the measurement programme, and that measured delta — not a claim of firstness — is what this briefing tests.
The control that would change this is unglamorous and cheap. Publish the divisor beside the rate: the population it is drawn from, the unit counted, the window the figure covers, and the list of detectors that supplied the numerator. Add one more thing and the instrument stops being merely published and becomes checkable — a route by which a rated operator can see its own division. Until a reader can perform the division, an abuse rate sorted by registrar is an assertion about performance dressed as measurement. The ranking is public. The arithmetic is not.
Member Briefing
Read the full trend briefing
Members read the full briefing with the source notes behind every finding.
Strategic Circle
A daily briefing in your language, built from changes we have verified.
Join for USD 20/monthCancel anytime.
Members also get the full research library, a daily briefing on the networks they follow and the private forum.






