Re: [Nautilus-list] User levels

[John Sullivan <sullivan eazel com>]

on 3/31/01 9:54 PM, Greg Hudson at ghudson mit edu wrote:

> (I've already sent a very similar comment to Maciej, and he
> recommended I bring it up for discussion here.)
> 
> So, in a recent interview with Linuxpower, Maciej recently wrote:
>> Maciej: One thing I'd like to see is to push some of the
>> technologies we've developed for Nautilus into the project as a
>> whole - things like user levels, some of the custom widgets we've
>> developed, our themable icon framework, and so on.  Another thing
>> I'd like to see is to maintain the discipline the GNOME project has
>> had for GNOME 1.4, and set and stick to a schedule.
> 
> The "user levels" part of this comment worried me a bit, since as a
> site integrator, I think user levels are a total botch.  I have two
> big reasons:
> 
> * It presents the first-time user with a choice whose
> ramifications are mostly unknown and which appears
> irreversible (or at least, you'd have to figure out how to
> reverse it).

The text describing the choice does say "you can always change it later", so
I think "appears irreversible" is perhaps an exaggeration here.

> 
> * It magnifies the complexity of documentation and support by
> a factor of three wherever something depends on the user
> level.

I don't think it's the user level that causes this -- it is the preferences,
some of whose values depend on user level. Any preference in any program
multiplies the complexity of documentation and support, since the behavior
varies depending on the value of the preference. The more preferences, the
more complex. This is one of the big arguments against preferences.

It seems a sensible argument that hiding some of the preferences at lower
user levels reduces this complexity. If the user level is Beginner, for
example, a support person doesn't have to ask what the value of many
preferences are, because they are at their invisible predefined Beginner
settings. If all the preferences were available all the time, the complexity
would be considerably worse.

I can easily understand an argument against lots of preferences in general.
But good luck convincing the Linux community that all decisions currently
captured in preferences should be up to the program's author instead.

> Obviously, a large installation can (and probably will) force the
> choice to a specific value in order to mitigate these problems, but
> will still encounter costs related to extra complexity in the
> Eazel-provided documentation and less consistency with the outside
> world.
> 
> A good user interface has to find a way to be usable by novices while
> still presenting powerful options to experts (although, in the case of
> the tasks performed by most GNOME software, expert users will always
> be able to find lots of power via the command line).  "User levels"
> seem like a poor excuse for not doing this design.

To me, user levels is one aspect of this design, not a way to avoid it.

I am not completely sure that user levels are the best design, but you
haven't really convinced me otherwise.

John