A Week With GNOME As My Linux Desktop: What They Get Right & Wrong

[moilami]

I think I am happy that no "exhaustive testing" was made for Gnome 3, else I bet lowest common denominator would had pressured Gnome to remain as W95 ripoff

[liam]

I'm curious, since this is brought up constantly regarding the GNOME design team: How does this work in the KDE camp? Ubuntu? Cinnamon?
Are they doing user testing on every design decision?

Ubuntu might be, but almost certainly not Cinnamon and KDE. At one point I know KDE had at least one ux professional associated with them but currently I'm not sure what their status is.
To be clear, not EVERY decision needs testing. Also, testing isn't easy. It's not necessarily expensive but it requires expertise to avoid biased results (something that the gnome team loves bringing up are anecdotal "I 'tested' this with my family/friends and they loved it".
What really annoys me about this whole thing is that there was even a public plan, a pretty detailed one at that, to test gnome (put together by an actual ux person). I think you can guess what became of that


bkor

TL&DR: I agree with the sentiment you expressed.

They basically should consistently testing everything within GNOME, but that's not done much (or at all). Any changes should be (eventually) tested as well. This would sort out a lot of problems. What's great about the current situation is that everything is very consistent. But if it is consistently unusable, then that's not really helpful. It's a slow process to change this. It took a while to have a design team. Also took quite some time before they really finished/polished things enough (e.g. non-discoverability of the app menu, notifications, etc).

I think having the team is great. They went from being totally overwhelmed to being on top of things. Now it would be great if it was expanded to add user testing in everything that they do. It's not unfamiliar btw, lots of GNOME people usually talk about how we lack doing this. There's every GUADEC a one-off talk about someone doing one test or maybe a few. IIRC (not sure.. might be someone else) Novell used to have a usability testing lab.

What's pretty clear is that Red Hat now has the majority of the manpower in GNOME. Which is a bit unfortunate, because although I think they're great, it might be a bit off putting for any other company as well as volunteers (any paid person can spend way more time than I can). Maybe just a matter of bringing user testing up with whomever decides on how many people are dedicated to GNOME and that it is worth hiring a few people (e.g. in Czech).

I'm glad that you, as a seeming supporter of GNOME, feel this way. As I've said many times before, I really liked the original design document for G3, but feel they've fallen victim to group think and outsider antipathy.
You don't need to test every decision. They haven't gone their own way at every point, and many of the paradigms they use are simply part of the "traditional" hci models. Others, like no always visible window list, app menu, or overview, would certainly have benefited from some testing. Some of it comes down to a, somewhat, qualitative question: is this change worth the distraction? App menu seems, to me, to pretty clearly not solve a widespread problem, and, arguably, doesn't improve workflow. Overview has a stronger case, but it feels only partially implemented to me.
As you say, consistency is very, very important, but there's no reason to sacrifice ease of use for it. They have their usage patterns, but how did they decide upon them? THOSE are things you test. Test often and early. Don't prematurely optimize
The GNOME folks exist in their own little bubble. It's just super political like you wouldn't believe. That's what's stopped others, who actually know better, from getting involved and trying to change things. Even to the point of people being discouraged from helping b/c of...things. This is disgusting, and something that just won't get talked about in the open. I'm telling you b/c you seem to be genuinely open, and might be unaware of some of the backstory. There are very good reasons why GNOME is the way it is, and has a certain reputation. It's well earned, and no amount of "come, join the X team and see that we're open to everyone" will fix things until they are willing to consider alternate views.
This won't change until GNOME is willing to accept serious criticism (serious in the form of detailed, data driven critiques) without burying them. Thus far, they just aren't open to it.

Best/Liam

[mdias]

What a shame not including Nautilus vs Dolphin in a Gnome vs KDE comparison...

[bkor]

In what way is the gnome design team qualified to make ux decisions?

TL&DR: I agree with the sentiment you expressed.

They basically should consistently testing everything within GNOME, but that's not done much (or at all). Any changes should be (eventually) tested as well. This would sort out a lot of problems. What's great about the current situation is that everything is very consistent. But if it is consistently unusable, then that's not really helpful. It's a slow process to change this. It took a while to have a design team. Also took quite some time before they really finished/polished things enough (e.g. non-discoverability of the app menu, notifications, etc).

I think having the team is great. They went from being totally overwhelmed to being on top of things. Now it would be great if it was expanded to add user testing in everything that they do. It's not unfamiliar btw, lots of GNOME people usually talk about how we lack doing this. There's every GUADEC a one-off talk about someone doing one test or maybe a few. IIRC (not sure.. might be someone else) Novell used to have a usability testing lab.

What's pretty clear is that Red Hat now has the majority of the manpower in GNOME. Which is a bit unfortunate, because although I think they're great, it might be a bit off putting for any other company as well as volunteers (any paid person can spend way more time than I can). Maybe just a matter of bringing user testing up with whomever decides on how many people are dedicated to GNOME and that it is worth hiring a few people (e.g. in Czech).

[kigurai]

In what way is the gnome design team qualified to make ux decisions? They do a lot of things well, but they are also very, very, very stubborn and don't do what is essential for ux: extensive user testing (and yes, it CAN be done for very little if there is will to do so).

I'm curious, since this is brought up constantly regarding the GNOME design team: How does this work in the KDE camp? Ubuntu? Cinnamon?
Are they doing user testing on every design decision?

[liam]

..

You're assuming that they needed to dictate. When a few people decided on starting a design team they didn't have much influence, despite 2 people being employed by Red Hat. They had an idea and convinced others. Quickly there was so much worked pushed to them that they became completely overwhelmed. There were also blog posts from developers who questioned / criticized the need for a design team (on planet gnome). One tried proving the uselessness of it by working with them, then completely changed his mind (resulting in a talk at GUADEC). Eventually this led to Red Hat hiring Allan Day.

So there is no "core difference", KDE could do the same. What is likely very different is the amount of time they have available in a given week.

This speaking as someone who pushed for way more involvement from the design team (basically said everything should have a design). We changed the way proposals should be done to have it checked by the design team (way before they were ready to handle this amount of work).

In what way is the gnome design team qualified to make ux decisions? They do a lot of things well, but they are also very, very, very stubborn and don't do what is essential for ux: extensive user testing (and yes, it CAN be done for very little if there is will to do so). Some of their decisions are just worthy of a good ole head scratch. Never would've gotten past the brainstorming stage for a real product.
I know for a fact they are...not very receptive to professional ux reviews.
Gnome just gets a pass to, largely, do what they want in rh (they got the hammer to create the absolute min with Classic Gnome, but that was very unusual), but if it were an actual product that made money you can be sure the folks leading the design team would be doing something else.
I'm sure they're great to work with if you largely agree with their decisions, but if you have serious criticisms it just won't work.

[profoundWHALE]

Now I guess you'll have to write a reverse article like every other reviewer out there.

Friday: "10 things that Mac OS X does to make you want to shoot yourself."
Monday: "10 things that Mac OS X does to make your life a breeze."

[cutterjohn]

osx hackintosh v. 'real' mac

sorry i still find osx overall as stagnant on a 'real' mac as i do on a hackintosh.

sure there are a few trivial items that don't work as well but can be generally trivially made to work on a desktop. on notebooks the only stumbling block is optimus/hybrid gfx.

at the end of the day my hackintoshes are more current and better specced than all but the top end mac pro.

as to the article, ive gone through wms, des with a variety of reasons everything from bloat, speed, to ram usage. i just dont have afavorite any longer as far as des go as theyre all bloating up, even xfce and lxde. i might be going back to base wm and making it up as i go again, esp given that the next project is an old msigt725 w/p8600 and 4gb ram.

[Passso]

Now, I'm using openSUSE and not Fedora, so there are some differences to point out. So I'll go in order.
....

I agree with you Emerald, most parts of the article are about "Fedora's implementation of KDE". The Opensuse experience is much better.

[Citan]

Hi all, hi OP!

Thanks for starting such a discussion, it's always interesting to compare interface ideas and actual user feedback.
With that said, there are many things in your points that puzzle me...

So, before reviewing each point (with which I don't agree) in detail, a general remark about your feedback: most of your problems either come from using Fedora flavor or using the "still-in-polishing" KDE5 environment. Because things such as "requiring root user for printers" never occured to me since the last three years for example (presently on a 14.10 Kubuntu).

So...

1. Login: you said it yourself, Breeze is much better. Agree with all your points.

2. Desktop and GPG: never encountered your problems, maybe because I didn't encrypt my home and always type my passwords, so I dismissed Kwallet one time then it didn't bother anymore. With that said, I fully agree with all your points: security is complex, so if an environment proposes it, it must provide sensible defaults in a full "clickodrome" way.
Apart from that, you didn't say anything about the desktop customization potent, I suppose it's because you were testing "as a plain user".

3. Applications: this is where you really have been way too partial to my taste...
a) I globally agree with you on Amarok. For Dragon though, well sure they should use Baloo, and THEY WILL (as Amarok did with Nepomuk and will again with Baloo). Just give it some time, it's a fresh new technology.
b) Muon: while I preferred the ancien package manager, I didn't find any blocking gripe with it. I especially don't see the "white space loss" you're talking about, Muon (Discovery) is nearly exactly the same presentation as with your screenshot from Gnome. You also get a three-pane detailed presentation, Launch/Update buttons and all. Really, you're borderline trolling on this one, or using a specific flavor of "Muon-after-Fedora-devs-waste"...
c) You didnt speak of the many goodnesses Dolphin has over Nautilus (which I feel a real pain in the *** to use in real work use-cases) : advanced breadcrumb (quickly change directories from any level thanks to the arrows), integrated terminal, displaying both full tree and bookmarks, quick split button, option to put directories apart or not etc...
d) You didn't speak neither of the Ksnapshot beauty, the top screenshot application of all which I tested (Shutter is nearly as good but not installed by default).
e) Nor digiKam which is one of the best pictures manager (fully integrated with Nepomuk at the time so will be with Baloo) in terms of tagging and other features...
I don't use the other KDE-related applications (apart from Krusader, only file browser which gives "enqueued" copy but it's a bit too stern for a lambda user) so cannot say anything about them though.

4. System configuration:
a) Printers: well, as I said before, there is clearly a problem with your distributions. I never had any prompt to check current printers, no more to add one (just tried now). I'm surprised however you don't fire a rocket at a general "over-the-top" richness of the options available in the main system configuration (or maybe they resolved this in KDE5, I know this is an old ongoing workshop). Although it's also thanks to that that you can configure anything through an UI.
b) Audio: did you notice that KDE is THE ONLY ONE (AFAIK) allowing on-the-fly change of output for any audio source, straight from the "volume" icon on desktop (to switch game/discussion on headphones when wife comes in the room for example)? Did you also note that the same icon provides also full mixer by input/output or by application? Never saw something as slick and efficient elsewhere. Basically unless you have a configuration problem, you never have to use the "audio" section of the system configuration.
c) Network: again, you're basically looking too far away. For user convenience, most of the daily configuration options are directly accessible from the network icon. So you won't have to go trifle in system configuration unless advanced use-cases.

Final thoughts: reading some of your final thoughts "KDE doesn't feel like cohesive experience." "This is about a mentality of apathy, this is about developers apparently not thinking things through when they make the UI for their applications. " This is nearly insult to the developers. Because, as yourself say in the same section, they DO care. As they DO have a general direction to strive towards. In fact, KDE is contrarily the only one desktop having a true objective in terms of features and ergonomy. It's just that, they are far more ambitious than they should be considering their available resources. So they make progress on all fronts, slowly but steadily. Sometimes they even have to drop whole concepts because too difficult to gain the critical mass necessary for it to really shine (goodbye Nepomuk). But they progress nonetheless, precisely towards a fully cohesive experience.
Furthermore, you tried KDE5 on Fedora, which is "raw environment" on "bleeding-edge technos". It's not a fair comparison: KDE devs still have much work to do to make KDE5 as good as KDE4 and then enhancing further. While Gnome 3 didn't suffer big changes in interface/technical paradigms so it's obviously more stable and polished.

My advice: try again with latest KDE4 (Kubuntu 14.10), see how many of your gripes are still here. KDE5 will be really up-to-par in a few months.
Sorry for the very long post, have a nice day.