Majestic put an experimental domain above Microsoft after forgetting a reporting filter
A November 16 update resolves Jack’s backlink mystery and gives reputation-list users a concrete reason to check what their rankings measure.

An odd-looking domain sat directly below Apple and above Microsoft in a web-popularity ranking. Jack clicked through and got a DNS failure. The impressive position had arrived without a working destination, which is a fairly committed way to demonstrate that a ranking and a website are different things.
The first-person investigation began with a backlink from a site organizing popular pages by domain ending. Looking at .com exposed w2rt6byl5nf1opmc-a.com at number eight. Jack followed that ranking to its underlying Majestic data and found a domain that appeared spectacularly well connected while refusing to resolve.
Majestic’s display counted 241,085,378 nofollow backlinks; Jack also found the domain at number 38 on Tranco. Those are the numbers he documented, rather than a fresh measurement performed here. A list intended to help describe the web had acquired something that looked more like an unanswered riddle.
The important turn arrived in the post’s November 16 update. Steve Pitchford, identified there as working for Majestic, attributed the anomaly to “Some internal experimentation.” His response said an esoteric domain had been used for experiments and a filter had been forgotten, allowing it to appear in reporting. That changes the story from an unexplained ranking puzzle into an admitted reporting problem.
Jack’s account also says somebody else bought the domain. He didn’t register it, and the post doesn’t document a takeover operation. Turning this into an attacker story would add a villain the evidence hasn’t supplied. An experiment and a missing filter already provide plenty to explain without casting additional characters.
For someone consuming these lists, the useful question is what a rank authorizes downstream. Using popularity to choose an interesting page and using it to decide a page is safe are different decisions. A high position cannot supply evidence about a destination’s behavior when the number itself came from a reporting experiment.
An administrator could review where a list enters an allow rule, a scoring system or a purchasing claim. The sensible check is specific: which dataset, which snapshot and which decision depends on it? That gives a correction somewhere to travel when the producer fixes its feed, rather than leaving the same old number living forever in a spreadsheet.
Experiments will keep needing production-shaped data, and public reports will keep needing ways to exclude those experiments. A promise to sort out the filter still leaves the corrected output to be checked by anyone whose decision depends on it.
Number eight needed a filter.