Re: portable e-mail, now Trying to do too much (was Re: the introduction problem, etc.)

[Phillip Hallam-Baker <[e-mail]>]

On Sun, May 22, 2022 at 10:57 PM Keith Moore <[e-mail]>
wrote:

> On 5/22/22 15:14, Phillip Hallam-Baker wrote:
>
> The architecture of mailing lists sucks. It has always sucked. There is no
> way to implement a push messaging protocol that is not going to suck and a
> push messaging protocol without ubiquitous authentication is going to suck
> really bad.
>
> This says nothing about why, or how, it sucks.  I can guess, but it would
> probably help to know specifically what you mean.
>
The problem with a push messaging protocol is that anyone can gunk up the
mailboxes of the recipients with large amounts of data. So the receiving
server has to decide whether the user is going to want to accept or not and
that must inevitably be a guess at some level.

Authenticating the senders reduces the guessing somewhat but you still end
up having to limit message size. I can't email you a 400GB video file. But
I might easily want someone to edit a file of that size for me.

I am aware someone has an IMAP service somewhere for IETF lists and there
> have been NNTP services. But neither of those work with my mail clients.
> And configuring my clients to be able to post while accessing the mail that
> way sucks even worse.
>
> Perhaps, but unless your mail client is very odd, getting it to work with
> IETF's public IMAP server, or switching to a mail client that does work
> with that IMAP server, still seems like it should be easier than, say,
> getting your mesh system widely deployed.
>
I was not being precise about my problem which is actually the amount of
complaining that we get about these issues and the refusal to look at the
fundamental problem which is that there is an abuse issue and the only
solutions we have involve heuristics which are at root guesses.

I have much higher acceptance criteria for the user experience than most
folk. It is not laziness as I will spend a month coding to avoid the need
for my user to perform a single mouse click.


Keith
>
> p.s. Of course, mailing lists predated by 15-20 years anything resembling
> a fully-connected global network, even if you count on-demand dialup access
> as part of that network.  So if the architecture of mailing lists has
> really always sucked, at one time it at least sucked less than an
> alternative that would have required an interactive connection to either
> the sender's host or a central distribution server
>

I have a vintage computing collection too. I just keep it in museum cases
where it belongs.