"Switch User" considered harmful

[Carlos Ribeiro]

On 2/27/06, Tristan Wibberley <maihem at maihem.org> wrote:
>
> Hi,
>
> Now that my trollish subject has got your attention :) I've watched my
> father on his Windows XP machine trying to grapple with the switch user
> concept. When finishing with the machine at a time when somebody else
> wants to use the computer, he will want to press switch user every time,
> no matter how much I explain the general rule "If you've finished with
> the computer use log off, only use switch user if you expect to go back
> on shortly". I wonder if something other than "Switch User" should be
> used. I also think he gets confused with clicking log off only to be
> offered something else instead - he might be trying to select the other
> option because he reckons the computer thinks he may be choosing the
> wrong thing.


Good point. There are cases when switch user makes sense, and there are
cases when logoff makes sense. I guess most people, given the option, would
always choose "switch user". We tend to think, "I may want to resume
working, so let's keep it ready". Logoff implicitly means "restart", in the
sense that the user will get a new session when loggin in later. So in this
sense, "switch user" is the best option (at least, it is what people want
most of the time).

Now, if you forgive me, a short rant...

The issue here is persistency. Current desktop systems are not really
designed to support persistent sessions. I guess this is more a consequence
of historical technical limitations than a well thought decision. Regardless
of reason, we still do not know which is the correct model to apply to the
design of a persistent system.

Let's assume a perfect world, one where hibernation really works, 100% of
the time, and where RAM usage is not an issue. Sessions could then be made
persistent by default, and it would be much more convenient to suspend and
resume sessions instead of loggin on & off. I guess this is more intuitive
-- if you leave the computer, you could always resume working where you
left. Of course, things are not that easy because there's the issue of
sharing (and locking) resources such as files. In a fully persistent system
a user could (potentially) leave resources locked in such a way that no
other user could actually use.

Going further, I'm really not sure what's the correct behavior in a
persistent system; do you keep files locked for the user that is editing
them, or do you allow other users to edit it? What do you do when a user
resumes working with a file that was modified by someone else? It's about
the same as leaving you office and finding next morning that someone wrote
over the sheets of paper that you left over your desk the previous day. In
the real world this is not a big issue for purely physical reasons -- most
people do not mess with a coworker's desk, and if they do it, they will
usually leave a message telling what they did. Even if they do not leave a
message, most people would promply realize that someone messed their
desktops. A virtual environment is not well suited to this model.

Implementing seamless persistency poses interesting questions. It goes
beyond our physical experience, in the sense that objects can be shared or
locked in ways a physical object can't. In this sense, things like a wiki
seem to point to another way of doing things, with finer grained control
over resources and a fairly volatile concept of 'locking'. Now that
technical limitations against persistency are finally being overcomed, it
will still take a lot of time to get the model right in usability terms.

--
Carlos Ribeiro
Consultoria em Projetos
blog: http://rascunhosrotos.blogspot.com
blog: http://pythonnotes.blogspot.com
mail: carribeiro at gmail.com
mail: carribeiro at yahoo.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.ubuntu.com/archives/ubuntu-devel/attachments/20060227/e3958930/attachment.htm