How does Message ID in Usenet / NNTP works?

Question

According to RFC 3977, the Message ID of an article should be globally unique in the whole Usenet network.

I'm curious of how an NNTP server can ensure the global uniqueness of a Message ID when it allocates one.

I think my question can be further broken down into:

1. Who is responsible for creating the Message ID? The client posing the article or the NNTP server receiving the article from the posting client?

2. How does the program creating the Message ID ensure its global uniqueness across all Usenet servers? Is a Message ID conflict avoidable?

3. If a Message ID conflict is possible, how does the article propagation between two NNTP servers handle the situation? Does it mean two NNTP servers can have partitioned views of the article of the same Message ID but different body?

Answer 1

Who is responsible for creating the Message ID? The client posing the article or the NNTP server receiving the article from the posting client?

The news server of the where the message was originally received creates the Message ID.

How does the program creating the Message ID ensure its global uniqueness across all Usenet servers? Is a Message ID conflict avoidable?

It doesn't. Usenet is a simple trust-based system. When it was founded, only universities and big tech companies could afford the hardware capable of being Usenet servers. If your server sent a message, then it was assumed to be be legit by virtue of being a relatively small number in existence. If one server was misbehaving, then it could be kicked off the network.

If a Message ID conflict is possible, how does the article propagation between two NNTP servers handle the situation? Does it mean two NNTP servers can have partitioned views of the article of the same Message ID but different body?

There is nothing stopping this from happening except still not that many people run servers. A news server admin could send new articles with spoofed ID's and nobody could stop them at first. Of course, it is still true that if they become a nuisance, they can loose all of their network peers and be kicked off of the network. Technically though, there's not much stopping it from happening.

Also, read about the KREMVAX hoax, the first time a spoofed message was sent. This is long before the internet.