Survex (interface expectation)

[Mark Shinwell]

On Tue, Jul 16, 2002 at 10:44:06PM +0100, Robert Smallshire wrote:
> The current, rather misguided attitudes surfaces on the Survex mailing list
> distress me rather.  They are all too symptomatic of the attitude of too
> many software engineers - namely that the programmer is right and all users
> are clueless idiots.

If you're accusing myself of that then I'm quite saddened, given that
I've spent a very large amount of time over the last two years
developing Aven, which was designed primarily for increasing the
accessibility of Survex.

In general, though (not referring to the Survex list) I have to agree
with what you say...

> As somebody who is both a software engineer and a domain-expert (in my case
> a structural geologist) I am in an ideal position to create the
> 'best-of-class' application for the benefit of my users. This involves
> spending a lot of time on 'usability' - if I want people to use the tool, it
> has to be usable at _their_ level, not mine.  This doesn't mean that I am
> clever and they are stupid - it just means that they are not software
> engineers - they are geologists.

That is very true.  However, to be honest I think "real" testing in that
regard comes from those users who actually don't know anything about the
software side -- it is very easy to be tainted by that (especially if
you are also one of those who has written the software).  With respect
to Aven, this is why I highly value feedback from the users on this
mailing list.  (Also see the point below about issues to do with how
Survex has evolved so far.)

> Likewise, many of the members of this list will have a dual interest, in
> both the problem domain - cave surveying - and the means to its solution,
> namely Survex software development.  These people too are in an ideal
> position to create a 'best-of-class' application usable by cave surveyors.
> I'm sure there are many opinions on how this is best achieved, but I can
> guarentee that alienating your less technically literate users

I should hope that pledging to produce a fully graphical cave surveying
software package, which is likely to be one of the best in its field
worldwide is not alienating users...

> and potential users by being abusive to them will result in failure
> and obscurity for the project - and deservedly so.

Speaking from my point of view, of course I'm not trying to alienate
users.  That would be stupid.  I think the issue is that given how
Survex has evolved so far, without a "full" GUI (and I would add here
that I wasn't involved in any serious development on Survex before I
spotted the need for a good graphical cave viewer) it would be wrong to
simply assume that one can be added simply and easily on top.  It
requires somewhat more work than that...

> I have always suspected that the reason many software engineers shy away
> from graphical user interface design for scientific problems is not because
> the command line and a text file necessarily offer more power or
> flexibility

Certainly in my view that is certainly not the case.  The command line
_does_ often offer more power and flexibility; on the contrary, also
graphical interfaces often do too.  It depends on the application.

By the way, one should note that my previous comments about Word were
nothing to do with the fact that it was graphical.

> but because designing an effective graphical user interface for
> such an application is actually a very difficult thing to do well.

This bit I do agree with.  As stated in my previous mail:

> I take the point that we don't have a fully-fledged GUI yet for
> Survex, but that is on the way, and is unfortunately very much more
> complicated to design than just saying "we need New, Open, etc".
> (There is a serious point here about what we _do_ actually need
> though.)
>
> For example, it's taken nearly 2 years' of work to get Aven's GUI to
> the standard it is: and that is very much more simple than the sort of
> system which would be required to support processing of complicated
> survey projects.  Apart from the design time, the sheer
> coding/debugging/testing time needed is enormous.  

which is why I don't think that rushing into this is the correct way: it
is going to take a lot of time to design it correctly, etc.  It's very
easy to over-simplify things in terms of the "what's required for a GUI"
department: especially when in this case we have a lot of possibly
inter-related things interacting with each other.

I would hope at least with the current advances in Survex on the GUI
side, notably the advent of Aven, have improved matters dramatically for
the average user.

> I have seen graphical user interfaces which offer power and
> flexibility well beyond what it would be possible to achieve on the
> command line.

...and in the same way I've seen exactly the opposite.  It depends on
what you're trying to achieve, really -- which is why I think it's such
a difficult task to produce a _good_ GUI for the idiom which needs to be
brought across for a good cave surveying package.

Mark

-- 
Mark Shinwell -- http://mrs30.quns.cam.ac.uk/ -- Mark.Shinwell@cl.cam.ac.uk
Theory and Semantics Group, University of Cambridge Computer Laboratory