In message , at 14:51:10 on Wed, 19
Dec 2012, Jethro_uk remarked:
I'm trying to explore what people think "reliable" means.

In this context it's knowledge that once the message is sent, it will get
to it's destination within a set period of time - enough to be useful.
Which is of course impossible.

Depends on how you define destination. I can provide people knowledge
that an email to one of my domains has reached the in-house server three
feet from me.

The next best thing, is the knowledge that if the message could not be
delivered (for whatever reason) the sender will be notified within a set
period of time - enough to be able to either resend the message, or
arrange an alternative route.

And anyone sending an email direct to that server will be informed
immediately if it's undeliverable. Although if they go through an
intermediary (such as their ISP) to send the email then there will be a
process of silent retries for maybe a day or two, followed by noisy
retries for anything up to a week.

The old world analogy of course, would be with the postal system. It's an
inherently unreliable one, but can be enhanced with recorded delivery.

Recorded delivery just says it got to the premises. Not to the person,
nor that they opened (let alone read) the letter.

My scheme above is at least as good as Recorded Delivery as far as those
aspects are concerned.
--
Roland Perry