Re: The path of least blame

[António Fernandes <antoniojpfernandes gmail com>]

2011/8/4 Akshay Dua <dakshay gmail com>

I hate that Gnome 3 gets a bad rap for little issues like starting a new terminal instance (http://digitizor.com/2011/08/04/linus-torvalds-ditches-gnome-for-xfce/). I personally love the interface, so seeing everyone get angry over very fixable little faults makes me disappointed. Then, it got me thinking: how can one avoid Gnome 3 users from getting angry?

Thought: If users can't blame Gnome 3, then they won't be frustrated by it. 

Case in point: assume an instance of terminal is already open and the user has forgotten about it. Also, assume that clicking the terminal icon always opens a new instance. Now when the user clicks the terminal icon and a new window pops up, she might see the old instance and go, "oops! I totally forgot I had a terminal already open. silly me!". Furthermore, users that wanted to start a new instance get the expected result. This way, Gnome doesn't get the blame. 

Is it fair to blame the user for forgetting that a given application was already open? I don't think so. We all have many things to think about in our lifes, why do we need to keep track of that kind of thing? Software is suposed to save us the trouble and just let us work.

In the current scenario, the set of users that forgot about the existing terminal are happy, but the rest are very unhappy. Raising the existing terminal is a completely unexpected result for those that want to start a new instance. So, they blame Gnome for that result...because they can.

So, my thesis is that in a situation where the common case is not known, or the outcome is hard to predict, software should choose the outcome of least blame to avoid user frustration and eventual alienation.

Your thesis bases itself in the assumption that "the expected result" of "clicking the terminal icon" is "start a new instance", but you don't provide evidence to found that assumption.

Sure enough, that would be the expected result if one was clicking on a laucher on Gnome Panel. But the Dash is not Gnome Panel and doesn't even look like it, so why would one expect it to behave like it? Dash is much more similar to Docky or DockBarX than a set of launchers on Gnome Panel.

Maybe it's users' expectation that is wrong, because they are used to Gnome 2. Why is Gnome 2 way treated as "the expected result" and Gnome 3 way "a completely unexpected result"? Maybe it is the other way arround. What is an "expected result", by the way?

If an action always gives the same result (from the user perspective), and that behaviour is reasonable, I would say it gives an expected result. In Gnome Shell, clicking on any application icon on the Dash always has the same logical result: it takes me to that application. I don't need to care if the application is already open or not - that is software business. If it was already open, it takes me to the previously used instance, letting me continue my work with that application where I left it. Otherwise, it opens a new instance for me. From the implementation detail perspective, those are 2 different behaviours. From the user perspective, it is one and the same: it takes me to the applica

Just my thoughts,

- Akshay
http://web.cecs.pdx.edu/~akshay/

Footnote: Users might want the feature to raise existing windows, but they will be understanding of the fact that Gnome cannot possibly know if they wanted to raise the existing window or create a new one.

The same can be said to users that want the proposed behaviour.

 

Besides, the activities overview already shows what's open on the desktop so raising the existing window is a patronizing move on Gnome's part.

_______________________________________________
gnome-shell-list mailing list
gnome-shell-list gnome org
http://mail.gnome.org/mailman/listinfo/gnome-shell-list