Re: [ietf-smtp] Make username optional in email addresses

[Keith Moore <[e-mail]>]

On 2/17/23 17:26, John C Klensin wrote:

> That would give you something almost completely non-disruptive.

I don't think there's any non-disruptive way to make usernames optional 
in practice.   Even if there are zero changes to the protocol and 
nothing that breaks any existing protocol implementation, what is being 
proposed are significant changes to the function of user agents and also 
of people's habits of interacting with email.

The analogy that comes to my mind is the behavior of web browsers which, 
if a user types in "foo.com" to the address bar, adds "www." to the 
front and probably "https://" in front of that. (note that this 
convention has changed over time and works differently on different 
browsers.)  The problem is that different browsers behaved differently 
in response to typed-in URLs, and also over time, which degraded the 
ability of people to type URLs into different browsers.  And that's 
actually a vital function. (And yes, browsers have tried to be slightly 
more clever about this over time, for example by only prepending "www." 
if the DNS lookup on "foo.com" didn't work by itself, and falling back 
to "http://" if "https://" didn't work - the latter being somewhat 
dangerous if the user expects the connection to be encrypted and the 
browser downgrades the connection without asking.)

What's being proposed in this case is similar - the MUA would have to 
behave differently if "foo.com" is typed into the To: or Cc: field.  And 
some MUAs would implement this and others wouldn't, and users would have 
to learn what worked in their particular MUA(s).

IMO we should consider changes to user interfaces to be as potentially 
disruptive as any protocol changes.  It might not affect software, but 
it still affects the robustness and predictability of the system as a 
whole.   (Or to put it differently, some of the protocol implementation 
is in users' brains.)

And far too many changes to email user interfaces have, IMO, been 
harmful to email.

In conclusion, let's please not lend credibility to this idea.

Keith