IMP development practices suggestion

Rich Lafferty rich@horde.org
Mon, 22 Jan 2001 14:33:57 -0500


[I composed the following in response to a discussion amongst the core
developers; Chuck suggested that I should post it here as well.]

On Mon, Jan 22, 2001 at 06:12:08PM +0000, Anil Madhavapeddy (anil@recoil.org) wrote:
> 
> Putting a user survey on horde.org about who uses which modules
> might be interesting.  It would take time (a year?) to settle down into
> a meaningful statistic though :-/

I doubt you'd find a lot of users (intentionally -- see below) going
to www.horde.org. Using Concordia as an example -- and while it's
biased one way in that I'm here, it's biased the other way in that the
only mention of "IMP" is in the version string -- no-one knows about
Horde, they know about the IITS Webmail service. The Horde website is
where the software's obtained, and Web applications tend to be
installed by one and run by many.

The message suggesting we find out what users *really* want has to be
taken in context; he wanted frames himself. Further, consider who
makes up the mailing list. There's a *reason* the User FAQ took me
months to put up: the list isn't comprised of users, it's comprised of
programmers and admins. When you do see user bits show up, what are
they? Users that can't figure out the difference between "," and ";",
and even then that's their admin posting on their behalf.  I get half
a dozen messages a week sent to faq@horde.org that say "What's my
email address" or "How can I change my password" or something similar
site-specific.

Consider something as crucial to day-to-day work as email, and
consider not subscribing to the mailing list that discusses the email
program you use. For power users, that'd be silly, but there are still
no users on the list. At our site, people complain that there isn't an
InfoNote on IMP; what they expect to be said about it on a single side
of a single sheet of paper (the InfoNote format) that they can't see
by going to it and trying it, I'm not sure.

These are not advanced users. 

> I guess we're all biased towards how our users use IMP.  I'm not
> a sysadmin anymore, but when I was, I very much sided with Rich, and
> preferred a simple UI.  The power users had a copy of mutt available
> to them ...

I'd be inclined to say that we're biased towards how people use
web-based applications. The browser is an *awful* terminal -- it's got
all the idiosyncracies of the IBM 3270, except that the 3270 is thirty
years old -- and the efficiency costs of using a Web-based program
over an equivalent local one are high. Now, obviously, the big problem
it solves is that of accessing your mail from multiple machines, but
even then, there are much more effective solutions for people that are
comfortable poking around on a Unix machine or a Winframe server or a
desktop box with VNC.
 
> It would be nice to see if a vast majority of our users feel this way
> as well, or do they all want massive power-feature-sets built into it?

It's silly to think that there isn't a vast market of people out there
that are non-experts and want to use webmail -- look at Hotmail. I'd
extend that to suggest that there are a lot of sites out there that
would rather have their users using their own services instead of
Hotmail's.

If we want to be a power-user's webmail program, we're limiting
ourselves to an *awfully* tiny niche. On the other hand, if we try to
offer a general-purpose webmail program that sites can install and
have users mostly understand from their previous experience with
desktop mail programs and Hotmail and similar, then we've got a huge
install base itching to install IMP. We don't need to figure out what
users are doing now, we need to *decide* who we're going to write
for. I think we already have.

I'm always surprised at how many sites run IMP -- it's easy to find a
few hundred with a web search or two. Search for "IMP requires
Javascript", or paste the contents of the new-user's introduction into
a "pages like this" sort of search. Some of my favorites: Yale, CMU,
the US Navy, Tulane, a major Australian and a major Finnish ISP, John
Hopkins U, Purdue, McGill, and the entire Nova Scotia Ministry of
Education. We've got installations in Indonesia, India, Mexico, Peru,
Hong Kong, China, Russia, Brazil, Poland, Ireland, France, Thailand,
Croatia, Australia, and even Florida.

That's not a niche market. If you really wanted to target a certain
bunch of users based on who's installing IMP now, aim at postsecondary
institutions, in and especially outside the USA.

One of the things that I think some contributors are having trouble
with is that IMP is, to an extent, something that's being developed
for other people to use, which is still a bit rare in the open-source
world. Usually, open-source software scratches an itch like "I need
to...", not "the users at my site need to...", which is what IMP seems
to have started out as and, as far as I can see, essentially remains.

  -Rich

-- 
------------------------------ Rich Lafferty ---------------------------
 Sysadmin/Programmer, Instructional and Information Technology Services
   Concordia University, Montreal, QC                 (514) 848-7625
------------------------- rich@alcor.concordia.ca ----------------------