> Email is not part of the web
That's "email", not "emails". I understood that to mean email the concept / technology, not "emails" as in individual email resources.
By that reading, yes, email is on the web. You can link to a mailbox with the mailto: URI scheme.
If you read it as "Individual emails are not on the web" instead, then webmail seems to cover that.
If I reside in "urn:oid:2.16.840" and I read "https://news.ycombinator.com", is that what they call the inner-platform effect?
The web is an abstract information space that can contain arbitrary links between objects. The web isn't something you load in a browser, it's a way of organising links between things. So, yes, you can link to books, movies, etc. Anything with a URI.
You might be interested in this early document about the World-Wide Web:
https://www.w3.org/People/Berners-Lee/1996/ppf.html
It describes what I'm saying in more detail, including explicitly mentioning books as something that can be addressed in URI space.
More seriously, I think that -- if pressed -- I'd say that the identifiers are on the web, but the things are not. Other than that, I believe we would agree???
That said, the origin of this discussion is in relation to Google's launching of AMP for e-mail. The justification was that e-mails are on the web, so it's OK to use AMP to push them forward. A better viewpoint, though one I disagree with, relates AMP to being a better mode of HTML. (As opposed to one of the Web.)
Serving public URIs for messages must be done by separate web servers like a webmail client, or a mail archiving tool such as Pipermail.
Note that there's no requirement for something to be an HTTP URI specifically. Any URI will do.