Transparency

How this site is made

The texts and illustrations of this knowledge base are written by an AI system. That is not stated here in the small print, because it belongs to the subject: a site about the fine detail of a communication system should leave no questions open about how it came about itself.

Who does what

The AI system — Claude, by Anthropic — researches the topics, writes the texts, draws the diagrams as SVG and writes the code this site runs on. This paragraph comes from it too.

The human — Ralf Schwab — sets the topics, checks every page before publication, corrects it and decides what appears. Editorial responsibility under § 18(2) of the German Interstate Media Treaty lies with him; it cannot be delegated, and is not meant to be.

What that means for reliability

An AI system can put things convincingly that are not true. That is not a theoretical worry but the normal case, and has to be worked against. Three rules therefore apply to every page here:

• Primary sources instead of retelling. Where an RFC exists, the text is written from the RFC. The source list at the end of every page is not decoration but an invitation to check every claim.
• Uncertainty is named. Where a detail is disputed or a reconstruction, the text says so — as with the date of the first email across a network. A clean number is easier to write than an honest qualification, and that is exactly why the qualification is there.
• Dated rather than timeless. Every page carries the date it was last checked. What is written here about deliverability is a snapshot.

If you spot a mistake anyway: write to us. Corrections are the reason the sources are named.

The legal side

This site is operated from Germany, so the law of the European Union applies. Since 2 August 2026 the transparency obligations of Article 50 of the AI Act have been in force there: anyone publishing AI-generated texts on matters of public interest has to disclose their artificial origin.

That obligation has an exception: it does not apply as a matter of course where the content is editorially reviewed and a person carries editorial responsibility — both of which are the case here. We label it anyway, for two reasons. First, whether an exception applies is of no interest to the reader; they want to know where the text comes from. Second, it would be odd, on a site about the origin of messages of all things, to keep quiet about the origin of its own texts.

Why do it this way at all

This knowledge base is a by-product. It comes about while work goes on towards a successor to email, and writes down what is learned about the existing system along the way. Without AI assistance it would not exist — for a single person alongside the actual work, the research effort would be too great.

That is the honest answer to the question of why a machine writes here: not because it is cheaper, but because the alternative is nothing at all.

As at 16 August 2026