[Owncloud] Re: ownCloud design evolution

Tue Jul 26 14:01:46 UTC 2011

On Mon, July 25, 2011 10:18, François K. wrote:
> I can agree with that. What's currently existing is mostly based on my
work, which was a WIP when someone decided to integrate it. So yep, it
wasn't finished at all and was clearly needing improvements.

Of course. I know that and wasn’t criticizing. No interface is ever
finished as people and interactions evolve.

>> 2) I used ownCloud on a netbook last week.
>
> What I understand behind this is that the current CSS doesn't fit small
screen sizes.
> Well, we have mediaqueries to handle this, there is no need to change
everything.

I know. I didn’t propose to change _everything_.

> We could apply the same strategy for smartphones : imho, there's no need
for an Android app or a WebOs app. We just need to work out a CSS.

Seriouly? Everyone uses apps, websites take way too long to load on mobile
still. Bartek and Aaron do awesome work.
Feel free to work on CSS for a mobile site. As said before, mobile is on
my plan next but first I want the desktop experience to be cool as there
already are mobile apps which are really needed for a project that
essentially is about files and synchronization of devices.

>> 4) I did some usability testing over the last weeks.
>
> What kind of tests ? How ? What did you test ? What are the results ?

As I said before I write my thesis on usability testing in FLOSS projects
with ownCloud (and Diaspora) as examples. You can check the methods I used
/ my current thesis draft at http://jancborchardt.titanpad.com/thesis
Any suggestions or just doing usability tests are appreciated, I hope it
is helpful.

I did a few interviews about how people handle files, use Dropbox, general
sharing etc. – everything concerning ownCloud
functionality really. Also some user testing which among other things
resulted in a leaner upload dialog (reasoning in my first mail, on the
bottom) & multiupload functionality implemented by Robin. The items on
hover are a result of the dropdown being not obvious enough, a list of
shared files is expected as a subentry of files, etc. etc. I paraphrased
most of the results in my mails to this list, the mockup, on
http://owncloud.org/index.php/Tasks or directly on IRC or in the git
commit messages.
Feel free to check out the interface branch where the mockup is currently
being implemented.

I will of course do subsequent tests to assess the design further. As I
said, it’s a continuous evolution and I don’t claim to be 100% right. I
know that this is not true, hence I don’t operate based on my opinion but
on actual usage & tests.

> The top bar sets ownCloud identity. I never thought about it as a
"controls" place.
> I never understood why the search bar is here. Never understood why the
buttons have no text either.

It only sets the logo. The whole design, the concept is the identity. Look
at Diaspora, Facebook, Google, all the apps people use. The search bar is
global – or at least will/should be – searching files, music, contacts,
calendar entries etc. because people expect that. The icons don’t have
text because after clicking them once (or even before that) their function
is self-evident. Icons are to convey functions and not text decoration.

>> Files control bar (breadcrumbs etc.)
>
> This would be an error, IMHO. You are putting navigation and controls in
the same zone.

That’s simply not how people think & interact. No one expect we developers
categorize interfaces like that. Making sense is above allegedly mixing up
different interaction elements.

As I said, I don’t claim to be right but I user test and will surely
continue to test the design evolution thoroughly.

>> The modified date will be displayed relatively (last week, yesterday, 2
>> hours ago) and in similar brightness as the file size - the older, the
lighter.
>
> I think date format should be an option. I personnaly hate to deal with
relative dates.
> I don't understand that brightness thing. I should probably give the
latest version a try to see what it is about.

That’s why the absolute date will be shown on mouseover. Design is
certainly not about the personal opinion of us developers because we have
to cater to a lot more people. I would highly advise against having any
options which are not completely needed. It’s about sensible & good design
in the first place so that people can use the program, not waste time
setting preferences.

In the tests it was pretty clear that people care very little for the
sizes and dates – the Dropbox web interface for example doesn’t have them
at all. While I don’t like comparing us to them they certainly do way more
testing and are the most used program in that regard after all.

Regarding the brightness, you should most certainly try the latest version
and also the interface branch. I personally like activity in git more than
on the mailing list as it’s much more useful.

> I don't think relying on "hovering things" is a good idea. As you said
it,
> everything tactile doesn't provide hover facilities.
> I also think the dropdown thing isn't that great and should be replaced.
But it clearly needs to be more efficient. What do others projects
provide?

I think it’s a great idea. Twitter does it, Identi.ca does it, Diaspora
does it, and the list goes on. As said before, most people use apps when
it comes to mobile.

At the moment, we don’t have the first world problem of tablet users (at
least not that much) that’s why I focus on a kick-ass desktop experience
which will certainly be the biggest user base.

I will do further testing and benchmarking to see what other projects have
and what fares best. It’s never finished, always interative design. Just
saying both current solutions aren’t that great doesn’t help much to be
honest.

>> And I finally solved how to properly do multiselect (no mockup for that):
>
> To me, it doesn't seem to be a good idea. We could just add a little
zone
> to display that. We shouldn't change the table headers.

Why? Please provide reasoning other than that we just shouldn’t and that
it isn’t a good idea in your opinion. As far as I know this hasn’t been
done before ever and we should totally do it and test it out. Of course
this isn’t a thing that has come up in the usability tests but was a
design idea of mine. One which I revised multiple times and talked through
with other interaction designers. Having a live version of it to test it
properly is what it takes over all talk about it, and I can not thank
Robin enough for actually doing the hard work.

We can especially improve file management which isn’t done properly by
_any_ file manager out there at the moment. And we’re doing pretty great
already, just check the interface branch.