Re: Email address conventions

In article <1184574691.974475.296600@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
dasgorb <dasgorb@xxxxxxxxx> wrote:

I have some questions about conventions used to create
the local part of an email address. I know there is no
standard, I am searching for the better conventions and
why they are used or why they were chosen.

My company uses a non-standard local address. We have a
director that _insists_ we need to use a standard like
firstnam.lastname. Claiming it is the standard

He's a liar :)

and
is better than our current convention.

That may or may not be true. Likely not.

I would like tho hear comments on conventions used
and any pro's and con's.

There are 4 things to look for in a local part convention in order of
importance in the modern world:

1. Uniqueness
2. Spammer guess resistance
3. Speakability
4. Memorability

The firstname.lastname convention only meets #3 and #4, which may be the
most important things for many people. Sadly, some people think it fits
#1. See http://www.sendmail.org/faq/section3.html#3.5 for the perfect
argument against it on uniqueness grounds. The collision cases may seem
ignorable, but they really are not when they actually hit, which is
pretty random.

The further from a 'real name' the local part is, the less troublesome
collision avoidance is for #1, and the better #2 is met. You also have a
problem with firstname.lastname when people get picky about it. For
example, I've had a user at a firstname.lastname site whose surname had
an apostrophe in it. He insisted (against technical advice) that his
email address carry that apostrophe, which is formally acceptable but as
a pragmatic issue is a very bad idea because many sites have noticed
that spammers have a fondness for such addresses, and either reject or
drop mail with such addresses. It took him 6 months and scores of lost
mail to be convinced that his somewhat reasonable insistence that his
name be represented correctly was inconsistent with the real world of
email. It is better that local parts NOT be allowed to intersect with
egos.

My favorite approach is first initial, first 6 letters of surname, and a
sequence number. So, if I were the first William Cole in a domain, I'd
be wcole1. This allows for good uniqueness and strikes a balance between
guess resistance and speakability/memorability. If you want to further
purge ego from address (an admirable goal, although it risks alienating
users) you might prefer combining initials with some sort of memorable
ID number (preferably one with only internal meaning like part of an
employee number.)

--
Now where did I hide that website...
.

Relevant Pages

  • Re: On writing negative zero - with or without sign
    ... But we're talking about the Fortran standard here, ... What's to prevent some new generation of computer and compiler to come ... Suppose the convention was changed such that an extra bit was ... answer is *exact* zero. ...
    (comp.lang.fortran)
  • Re: bad-idea-p?
    ... pointing out an obvious ambiguity in the convention -- that it really ... And then when someone decided to test for ideaness they would be expected to parallel bad-idea-p with idea-p? ... Intriguing, not what I was thinking, and given the standard lack of interest in consistency people keep quoting to me I would expect ideap to fit right in even with a bad-idea-p toehold. ... Another thing being missed is folks not reading the whole thread, where they would discover I pointed all this out to the first bozone molecule that did quote me the convention. ...
    (comp.lang.lisp)
  • Re: relative-to-source file names
    ... A simple convention could help. ... HELPME at the prompt if the loading process goes "the way of the ... The standard could be kept at 1K for embedded systems. ... technically adept and will be used to having to solve some problems. ...
    (comp.lang.forth)
  • Re: The Big Bang & Variable Light Speed
    ... in standard units: ... "There is no such term as "full metric" and no standard convention nor ... on what dimensions its components are to have, ... the hell you're talking about; you have no background in anything ...
    (sci.physics.relativity)
  • Re: IBM error messages getting worse?
    ... The "IBM convention as to importance " does not apply to ... For console messages I believe W is for a Wait State message, ... applies to z/OS, z/VSE and z/VM. ... The standard describes what a message ID should look like ...
    (bit.listserv.ibm-main)