On Sat, Jan 1, 2022 at 11:20 AM John C Klensin <john-ietf@xxxxxxx> wrote:
--On Friday, December 31, 2021 19:03 -0500 Phillip Hallam-Baker
<phill@xxxxxxxxxxxxxxx> wrote:
My main point was something else. Your assertion that these
things cannot be done with SMTP and supplemental tools is simply
false. An assertion that your method is better, or would be
much better if we were starting from a clean slate, might be
true, but that is different from saying "can't be done".
I am not saying we can't fix SMTP, on the contrary, I am saying that I have fixed it. I have a solution to every one of the issues that I originally set out to fix and I don't require a lot of mechanism to do that.
The Mesh is very compact, it is based on a very small number of components which have been stripped down to the bare essentials. It is essentially a LISP approach only instead of 'everything is a list', 'everything is a list represented as an encrypted append only log authenticated by a Merkle tree."
The problem is the Internet fax problem. Sure, you can send faxes over the Internet. My last boss made a small fortune out of a company that did exactly that. Fax-to-email.
The problem is that fax-by-email is a transitional technology that cannot possibly survive for very long because every user of fax-to-email has the ability to send and receive email attachments eliminating the fax machine altogether.
(1) If we were starting from a completely clean slate today,
trying to build a communications system for messages longer than
typical 19th century telegrams, would we base it on something
resembling the postal services and/or SMTP models or on
something closer to your model? In thinking about that,
remember that --the present situation with a small number of
dominant providers notwithstanding-- the SMTP design is for an
extremely distributed system with no central authority while
yours, as I understand it, requires a centralized mechanism for
(at least) callsign registration and handling no matter how
benign, profit-free, and robust that mechanism might be.
It can make use of but does not require the callsign registry.
SMTP does have a central authority. It is called ICANN. I am proposing a system that offers users a choice of the ICANN authority or the new authority.
(2) Given that we probably cannot start over and that SMTP-based
systems are very widely deployed and used (the rapid growth of
systems optimized for very short and quick communications
notwithstanding), is your system sufficiently better to take
over despite the considerable inertia implied by the installed
base? Of course, if "take over" is not your goal but, instead,
you would be satisfied to help out a small and sophisticated set
of users who would be happy with a new system to communicate
with each other while continuing to use traditional email for
communications with others, the threshold of "sufficiently
better" might be much lower.
The Web was originally designed to allow particle physicists to exchange research electronically. It succeeded on a larger scale in part because it was designed to scale but mostly because the scheme field in the URL allowed a single client to access information via Gopher, NNTP, WAIS, FTP, etc. etc. in addition to HTTP.
The Mesh contact book is designed to provide users with a central hub from which they can effect a secure communication via any modality supported by the device, not just mail, but chat, voice, video, Office collaboration, discord, VR, etc. and via any mutually supported protocol, not just XMPP, but IRC, Telegram, Signal, etc. etc.
I know that the providers of those protocols will resist interoperability but folk who try to build walled gardens can expect to find folk knocking holes in their walls.
(3) Given how poorly we have done with deployment and active use
of S/MIME and [Open]PGP (and scaling the PKI required for the
former and the web of trust for the latter) to the
non-specialist public and the almost equally poor track record
of new and incompatible systems taking over from ones that are
widely used --at least in the absence of the "new" version being
completely free and/or having huge amounts of bundled marketing
muscle and money behind them-- would your system be more likely
to be accepted and deployed if, instead of saying, "SMTP is dead
or irretrievably broken and needs to be replaced", you were
looking at hybrids and transition strategies?
At present, the only Mesh messaging applications are the confirmation interaction which is an upgrade on 2FA, the contact exchange interaction and the device connection interaction. I am not doing a mail replacement yet.
Back in 1992 when Tim B-L persuaded me to work on the Web, he asked me to help fix the protocol before the user base grew to the point it became impossible. Well Mosaic bjorked that idea. It took five years to implement about half the fixes I had hoped we would get to and many of those fixes are a lot kludgier than I hoped.
Since then, I have made a point of designing for the success-disaster case because once you pull the deployment lever, thats it, there is no recall button.
The Mesh has been running in the lab for almost 9 months now.
Again, I don't know the answers but, no matter how much better
your ideas might be (at least in your opinion), the installed
base (or at least the portion of it that has not abandoned
conversations like this for << 500 character messages) so far
does not agree with your opinion about the non-usability and
non-adaptivity of contemporary Internet email models. And every
time someone looks at your messages or specs and makes comments
about telephone numbering, X.500 or other universal directory
systems, the DNS as it has evolved, or transnational use of
national identity or certification systems, it is a challenge to
you, not just to show that you can do better but that your
system won't ultimately succumb to the problems for which those
systems have been criticized.
There is actually another option that you missed: I take the other person's idea and add it to my own.
I don't often take someone's entire idea but I have frequently changed course when someone has made a comment that has led to a better idea.
