WebTumble is a web catalog. It has published under that description since 2011, and the description is meant literally: the site accumulates data related to websites, and it points at where that data lives rather than pretending to hold it. Everything on the site falls into one of two halves, and they are organised on opposite principles.
The catalog half
The catalog is indexed by the question somebody asked. That is an unusual way to organise anything and it produces the range you can see in the index: 8 full entries running from a 2009 mobile game to academic appendix formatting, and well over a hundred further terms recorded. Nothing about those subjects belongs together. They are there because they were looked up.
A full entry means the sources have been read rather than collected. Each one is named, described accurately rather than promotionally, marked with what class of source it is, and dated at the point it was last reached. The distinction between an authority, a contemporary account, an aggregator and an archive matters more than any of them individually, because it decides how much weight a reader should give what they find.
The remaining terms are listed rather than deleted. An index that quietly drops the parts that no longer have an answer is a less honest index, and the closures are frequently the interesting part.
The reference half
The website data section is 11 pages on what can actually be known about a website, organised by subject because this half has one. It works outwards from the record: what a registry publishes, how a name resolves, and then the measurement layer where nothing is published and everything is estimated.
That last part is why the section reads the way it does. A great deal of what was once freely checkable about a website is no longer checkable at all, and any guide written as though it were is wrong. The report on closed measurement tools documents nine that closed between 2011 and 2023, with dates, because a reader following an old citation deserves to know that the endpoint stopped answering rather than that they are looking in the wrong place.
The 3 decoders exist for the same reason. They do not query anything, because the services they would have queried are gone. What they do is name the fields in a record you already have, in your own browser, which is the part of the old lookup experience that is still possible and still useful.
How it is written
By reading the sources. That sounds like a low bar and it is the reason there are eight full entries rather than eight hundred: reading a source properly, describing what it holds, noting where it will mislead and recording when it was reached takes time that summarising somebody else's summary does not.
Three rules govern the writing. A source is named with its host, so a reader can see whose claim they are about to read. A figure is quoted with where it came from, or not quoted. And a service that has closed is described as closed, with a year, rather than linked to as though it still answered. The third rule is the one that generated most of the reference section.
Corrections
Where an error is confirmed, the page changes and its review date moves with it. If a source has changed rather than gone, the annotation is rewritten to match what is there now; if it has closed, the entry says so and keeps the record. Anything factual can be reported through the contact page, and it helps enormously to name the page and quote the sentence.
Publication details are on the imprint, and what the site does and does not collect is set out on the privacy page. The short version of the second one is that there is no analytics, no advertising and no tracking of any kind, and the decoders work without sending anything anywhere.