It’s time to fix the window controls (ubuntu 10.04) | Hacker News Reader

It’s time to fix the window controls (ubuntu 10.04)

yokozar.org

138 pointsdotcoma89 comments

Whether they had good reason for changing a central UI element or not, it comes across as if they decided on a whim. This change seems so random. It's strange that they didn't anticipate the uproar in the community and at least try to answer the concerns and justify their decision.

It's a weakness of all centralized software construction mechanisms that change chosen by a handful of people can end up thrust unwillingly on large crowds. It's not like casual users can fork Ubuntu simply because they dislike the window layout.

But there are some pathologies that seem endemic to open source. A revealing comment by one of the bug shepherds on the relevant bug[1] says "Change is always good , if there is a good reason to change". I think this is plainly false: change doesn't just need good reasons, it needs excellent reasons, because change by itself has such high costs.

Open source is often too open to change for its own sake, and often lacks consideration for the maintenance costs of new functionality. To take the on-topic example of OS changes, it was viable for XP users to completely skip Vista, and continue to use the old OS, over a period of nearly 10 years, without issues of incompatibility. Try doing the same with a typical Linux distro, and you'll find your apt-get (or equivalent) will pull in all sorts of dependencies that you may or may not desire if you want to use the latest software.

[1] https://bugs.edge.launchpad.net/bugs/532633

Isn't this really a weakness of this system's abstraction? This appearance decision is just a skin layer. Rather than thrust it upon people, it could just be the default skin/view/appearance/theme, and there would be ways for contributors to submit additional themes that smoothly plug into the system. Don't like the buttons on the left? Use this other theme instead, which is mostly the same but with this exception.

Yes, there's no problem that can't be solved with another layer of indirection, etc. But making things configurable has its own costs, which are not trivial.

It's a weakness in the user interface. The preference to control button placement isn't exposed in the Appearance applet, so the only way to change it is to set a gconf preference.

To take the on-topic example of OS changes, it was viable for XP users to completely skip Vista, and continue to use the old OS, over a period of nearly 10 years, without issues of incompatibility. Try doing the same with a typical Linux distro, and you'll find your apt-get (or equivalent) will pull in all sorts of dependencies that you may or may not desire if you want to use the latest software.

You've made a mistake in your Linux distro selection, if lifecycle in a priority (and for me, it usually is). RHEL (and thus, CentOS) has a seven year update cycle. During that period they will never "pull in all sorts of dependencies you may or may not desire". It is a stable platform and it remains the same throughout its lifecycle. If you prefer apt-based distros, Ubuntu LTS has a five year lifecycle. Debian tends to be about 3-4 years, but is based on the development of the next revision rather than a stated lifecycle, as far as I know (so if Debian develops faster, you have to upgrade faster).

That said, I suppose one could argue that on older stable Linux distros, one might have to add some dependencies from outside of the core OS in order to run super new stuff...though I rarely find things that won't build against the libraries available on CentOS 5 or Ubuntu 8.04LTS. Some developer tools move really fast, as do some of the new-fangled Gnome or KDE user applications, but mostly I never notice that the OS is a couple years old.

So, yes, Open Source does change faster, but there are very easy ways to opt out of the rapid release cycle for systems that need stability (like servers). I can't tell you the number of folks I've met over the years who've started out with Fedora or Ubuntu, only to figure out that upgrading every 12-18 months is stressful and unproductive, and moved to a more predictable distro, even though it has older packages.

In short, you have a choice about how long you can use a Linux distro safely and productively. You can prioritize for the latest and greatest or you can prioritize stability and predictability.

slaaaaaaaaaaackwaaaaaaaaaaaare ;)

No package management? No thanks.

One of the most confusing parts about this switch, to me, was that it immediately precedes an LTS release. Why would you spring experimental changes right before the release you're calling 'stable' and committing to for 5 years? I was under the impression that LTS was supposed to be the distillation of all the new features that were embraced or rejected during the wild and crazy intermediate releases.

The actual change isn't that big of a deal. It's pretty easy to fix, and it would have been fine in a minor release with plenty of lead time before it became 'official', just to try it out. The big deal is that this reveals a broken management and release process, and comes across as a serious trust breaker.

Ubuntu has a long history of making dumb little changes without warning and then committing to them against all reason. I don't know if they've fixed it yet, or not, but my favorite is when they broke the Apache configuration file by adding shell variables to it (thus making it it unparse-able by everything other than Ubuntu scripts; the Apache standard tools can't parse the Ubuntu configuration file without pre-processing, for an example of how absurd this decision was). All this breakage was caused just because someone was too lazy to parse out a few different types of variables from the existing configuration file syntax (I looked it up, and the comment about this in the changelog was that parsing it wasn't possible; despite the fact that numerous, probably hundreds, of applications and scripts do parse it correctly). The Ubuntu developer culture of superiority and self-importance is quite trying at times.

An interesting contrast: One of the Windows Experience Principles introduced for Windows 7 is "Change is bad, unless it's great."

If they want to make the window controls like the window controls on Mac OS X, I'm fine with that. (I use Mac OS X a lot anyway, and am already comfortable with controls on the left side of the window.) They should at least get it right, however, with the close button on the far left of the title bar, and the minimize and expand buttons to the right of that button.

It looks to me like they just moved the right hand centric window controls to the left, which is terrible ergonomics and design.

OS X doesn't suffer from the File Menu/Close Button proximity problem either since the menu bar isn't attached to the window. That proximity alone should be a disqualifier for this new layout.

As a long time Mac OS X user I recall the general feeling was that the Mac OS X window controls where a big step back from Mac OS 9. Most felt that they put all 3 controls on the same side (rather than put the dangerous close button on the opposite side) for purely aesthetic reasons i.e. so they could use the traffic light metaphor. Which is a bit cheesy when you think about it.

Bruce “Tog” Tognazzini covered this complaint in 2000, in a quite excellently detailed look at the usability of the new OS X controls.

http://www.asktog.com/columns/034OSX-FirstLook.html

search for "gumdrops"

> People with experience with Windows know how easy it is to hit the wrong control and accidentally close down the whole application when setting it aside was the goal. Apple has actually provided some pretty decent spacing in these controls, in addition to enlarging their size. I wouldn't expect the kind of error rates from this that people have been predicting, although the original scheme of having "dangerous" on one side of the title bar and "benign" on the other made a lot more sense.

I think the key complaint at the time was:

> The button sequence is also a problem, in that it clashes badly with both the current layout and with Windows. This will also be a problem.

Plus ça change....

Might be a little bit off topic, but the detached menu bar gradually pisses me off as screen gets bigger. It's really a long-distance travel for the mouse to click the menu on a 1920x1200 screen. Now Apple is rolling out 27' iMac with an even bigger screen ...

The situation gets even worse when I use multiple monitors as menu bar only shows on one of them.

I agree. Fitt's law falls apart at high resolution

I'm not convinced it falls apart. It seems more like a failure to take account for the distance you must travel on your return from the menu bar.

Interesting I have no problem with it on my 30" monitor. But then again I don't use the menus very often.

I don't use menus often, either. I guess this design really forces app developers to carefully choose buttons and other widgets on the UI to avoid menu as much as possible.

But on the occasions when menu is required, the experience sucks big on my dual-monitor setting. And there is no keyboard shortcut to ease the pain.

Use control+F2 to focus on the menu bar. :)

Yeah I know that trick ... but it's not quite the same as ALT+(underlined character of the menu item) thing on other systems. For fast typists, it's pretty awkward to hit the  first and then use arrow keys to move around. I wanna go directly to "File" menu by hitting SUPER+F, for example :|

Well you can hit control+f2 and then hit F for file. :)

Yes I could do that ... but I left the F1-12 keys on my MacBook to its default, which controls dim the screen, volume and stuff like that. So to press CTRL+F2 I actually have to press FN+CTRL+F2. Try this combination on a MacBook (with only the left CTRL) and see how awkward it feels ... It completely screws the experience of "shortcuts".

The proximity of toolbars in OSX can be a problem in relation to the window controls. In Safari for example the back button is extremely close to the close button in the top left of the window. Unless we want to waste a ton of screen space on isolating window controls I'm not sure there are any better options out there. You can't assume the top right of the window will be devoid of other UI controls so it's not totally safe either. Somewhere in the past I remember reading about the idea of making UI controls "sticky" by slowing or snapping the cursor to a control when you got near it along with using some haptic feedback built into the mouse. I think these are both better ideas than trying to shuffle around pixels to solve the problem. No matter where you put it there will always be a chance of a miscommunication between your eyes and clicking finger.

Someone mentioned multiple monitors/high resolution -- personally I think none of the major desktop operating systems have scaled to handle multiple monitors or high resolution displays very well. We're still stuck with this idea everything useful has to be on the edges of the screen.

The 5th location by Fitt's law is the current position of the pointer, and Openbox (Linux) has the ability to open the 'start menu' there.

The title should also be centered if they're going that route. I wish I could explain why, but it just looks funny with the title right next to the buttons.

Amen.

I use both OS X and Ubuntu and I don't mind remapping my behaviours depending on the context. But if you make things similar but not quite the same... Irritating!

I'm not sure that Ubuntu should be copying OSX but if they are going to do so, then bite the bullet and do it.

When using Gnome I've always configured them to be on the left anyway. But using the same order as in OS X, close on the far left.

People still have the choice to move them to the right if they want. Viva la choice. Windows and OS X both suck in that regard (though I understand them wanting the consistency).

Not only is this a violation of Fitt's law- it's practically the canonical example of how to violate it. That is, when an application is maximized the exact upper corners of the screen become incredibly valuable because a user can get there without aiming but just by flicking the mouse in the general direction. Flick it to the upper left and click for all window controls, flick it to the upper right and click to close.

I remember when Microsoft made it so that you had to go to the corner but then pull it back several pixels, thus destroying forcing you to aim the mouse cursor. It may sound like a silly little thing, but I thought at the time that it was a good metaphor for how they could get something to _look_ similar (to, say, a mac) and yet still get it completely wrong. When I switched to linux around the same time I thought it was so refreshing that the UIs of the time (minus some skins people had made) didn't violate this example of Fitt's law, and also that if anyone were dumb enough to change that in the future, we could all just fix it.

I usually disregard comments about linux desktops copying window's mistakes, but this is just very irritating to me, mostly because it's elective- it's like you actually have to _try_ to get it wrong.

I'd love to see some modern testing results on that, honestly. The idea of "flicking" the mouse to the corner of the screen made sense on a 512x348 Mac Plus desktop. It doesn't seem as appropriate on a 1920x1080 monster, where generally nothing is maximized anyway. And with a touchpad you can't do it at all.

You're neglecting pointer acceleration, which applies for touchpads too.

If yours is anything like mine, try this: position your mouse to the far left of the screen; quickly swipe across the touch pad from left to right. Your pointer should reach the far right hand side of the screen. Now swipe ever so slowly from right to left, your pointer might get about a quarter of the way across the screen.

Once you consider acceleration (eg, the further you try to move, the faster you are moving when you get there), Fitt's law is more, not less, important on a big screen.

i disabled acceleration, and i suspect many others do.

i tried several times to get used to it. it's just not predictable enough for me. hitting or missing a target just because i was too fast or slow, sounds useless to me.

The acceleration curve on macs is horribly horribly broken. There's a bunch of utilities to fix it or turn it off because strangely you can't do it yourself.

I may have confused the issue. It's not about the flick of the mouse- it's about not having to aim (e.g., can you perform this action using your peripheral vision?). It sounds like you've become accustomed to aiming at everything and therefore don't see what the big deal is. Obviously the designers in question feel as you do, so it's certainly widespread.

As for the research, just to further clarify- they rigged a mac so that there was a second monitor above the first monitor (and above the main mac menu) that the mouse pointer could move to. When mac users sat down to use it, even knowing exactly what was going on, they consistently overshot the menu by an average of half a screen. I'd be highly surprised if resolution was even minimally relevant.

You appear to be making assumptions based on the Windows OS which don't apply here, which confuses me as you seem to be saying you use Linux.

Both Ubuntu and Mac OS X have a top "bar" which has functions in both the top-left and top-right corners, easily accessed according to Fitt's Law.

In Ubuntu top-left and right are Applications and Logout/Sleep etc. respectively and in Mac OS X it's the "Apple" system menu and Spotlight search. Maximised windows in both these OS's don't expand to the top corners and in Mac OS X windows don't even "maximise" to the full size of the screen, rather optimising for the correct size which will often not be full screen.

Whoops, assumptions about linux actually. When using ubuntu I always turn off the top bar first thing. Most of the people I know that use ubuntu do the same, but I have no idea if it's representative. I turn it off specifically because the quick window actions are far more valuable to me. In any case I run crunch-bang, which is ubuntu based but doesn't have the top bar, so I forgot about it when posting.

It's not like it ever happened before, is it? Oh wait:

https://bugs.launchpad.net/ubuntu/+source/update-notifier/+b... ("update notifications change" outrage with 20+ duplicates, 100+ reporters, 400+ comments) - this is what happens when you implement an official pop-under to do system upgrades.

Timeline: Feature freeze 19 Feb 2009, change done 23 Feb 2009, UI freeze 5 Mar 2009?

Ahh crap. You just turned one of my minor annoyances into "who the thought this was a good idea?" I thought I was misclicking somewhere or accidentally hitting a keycombo. I have come to accept people's mistakes, but when they put some thought into it before pulling the trigger--arrgh!

I mean wow, just wow, I would love to hear the rationale for such a boneheaded decision.

Thank god whoever makes this decision doesn't work for a car company, or we'd see next year's model with reversed gas and brake pedals.

FTA: http://www.ivankamajic.com/?p=281 "Many of you have been asking for some correspondence regarding the button position in the window manager.

Here it is..."

In that link they really don't actually go into the 'why' they just say they thought about it and show us a snapshot of a white board that says nothing as far as rationale.

The most telling part of the article parent linked to is: "Is it better or worse? It is quite hard to tell." If that's what they think, then why change it?

Additionally, the author says: " I have had it running on my machine with the buttons in this order since before the Portland sprint (first week of February?) and I am quite used to it". That sounds like it conferred no benefit whatsoever.

Sorry. I didn't mean to suggest that the article was entirely complete. Only that you might have missed it, and it might go further in explaining things you seemed interested in. =)

Sorry if I sounded offended, I actually did miss that link, and I up voted your comment when I read it.

Though I have to agree with his final thoughts:

  Personally, I would have the max and min on the
  left and close on the right.

Why they chose against this idea (with the title centered) is not explained. Personally, I think that his blog post poses more questions than it answers.

So why don't they leave it on the right as default (old way) and just a gui way to switch it to the other site for left handed/osx people. Then both sides are happy. Problem solved-ish.

I was going to say exactly the same thing -- if it was a THEME, then it seems like it would be just as easy to theme it the other way.

I guess the really tricky bit revolves around which one is default, and how obvious it would be to switch it to the other way. Knowing Gnome menus (with dated knowledge, naturally) it would typically be 6-7 clicks into somewhere and nobody would ever find it.

Sorry, but themes sucks. They make it very hard to make excellent GUI applications, since you have no fixed foundation to build them on.

To build a custom control you need to be able to integrate it in the surrounding context, i.e. the theme. If you do not know what that context is, you have to spend needless energy to make is fit anywhere. This is energy which could be spent improving the GUI instead.

And whatever you do, don't go around changing the order of buttons and such in the theme. This makes it even harder to make an application which fits in its context.

Not having a theme also allows you to communicate you vision about how the platform should look and work without having the message diluted. A platform needs this vision, and of course this vision needs to be good, for it to work as a breeding ground for excellent applications.

I'm not saying that Ubuntu's new look is the right look, I'm just saying that the solution is not themes.

When presented with two interface options, we should resist the urge to implement both and let the user decide.

Why not just go with the one that is more useful to more people as the default and bury the option to switch deep in some 'advanced' config setting? You still have to make a design decision in choosing the default correctly since that is most crucial.

Sure you can configure it. http://fosswire.com/post/2008/02/customize-your-titlebar/

There's a big difference between letting the user decide and choosing for the user but making it possible for the user to switch.

It's already been fixed!

http://www.omgubuntu.co.uk/2010/03/worried-ubuntu-1004-will-...

so link to a blog post that just links to another blog post?

To be fair, I couldn't be bothered to click through. Just wanted to pass along the information regarding this topic (which is now obsolete).

EDIT:

http://blog.dustinkirkland.com/2010/03/ubuntus-new-themes-an...

Now they just need to make the buttons bigger like this http://blogs.pcworld.com/staffblog/archives/Google-Chrome.jp...

Oh and make the top right pixel click-able too (for Fitts's law).

Seems it's time to fix the database, too: "Error establishing a database connection"

Edit: Bing cache has it (Google doesn't!)

http://cc.bingj.com/cache.aspx?d=802200748186&w=92581924...

Here's a brilliant idea for any Google engineers out there: Dedicate a few bots to caching everything that hits the front page of digg/reddit/hn before the sites go down.

The Ubuntu folks need to get a good UI designer to deal with this stuff. I'm not sure a design-by-committe approach works for producing a good UI with sensible defaults, good UI organization, good colors, consistency, etc. It seems to me Ubuntu 10 is taking a huge step backwards in all these areas.

Except that this wasn't a design-by-committee approach... This appeared fully done from the internal design department of Canonical a day before UI freeze, which just reeks of everything FOSS shouldn't be :(

They need to change the entire GUI. Desktop metaphors are done.

EDIT: For some reason I can't reply to anything anymore...Just redirects me to http://news.ycombinator.com/r with a blank page.

Anyways: cookie: Oh man I have a ton of ideas! A lot of them flow well into the touch interface world, but that means they don't work well with the keyboard interface.

For instance, one of my ideas was the infinite vertical, limited horizontal screen. Check out Shoes's STACK & FLOW positioning for some hints.

Radial menus instead of file menus. Also, menus open-able from from anywhere on the application ui.

If I had a decent design station and PS4 I bet I could make some nifty examples.

What do you suggest?

KDE wanted to change the whole game up with KDE 4. They solicited the community for ideas for months. They still ended up at roughly the same interface and concepts that have always existed for the overall interface metaphor.

I think this is just how people expect computers to work now. Of all the proposals out there, I haven't heard many feasible replacements for the desktop idea.

"They solicited the community for ideas for months." Maybe soliciting for ideas visionary UX guru's would have worked out better for them.

UI design is a technical art-form not a popularity contest.

Right, but the thing is that nobody could present to KDE an idea that was likely to work out better than the traditional desktop/windowing model.

If you're right handed and start with the pointer in the centre of the screen, the base of your hand resting on the desk and palm/fingers covering the mouse, it is more awkward trying to move the mouse to the right than to the left. Movement to the upper right corner seems to require arm muscles.

Maybe I'm on my own here, but I like the look of it on them on the left. I'm used to that with OS X.

I wish we had some kind of cache service or something that generates multiple cache links for a website thats slashdotted/reddited/digged/ycombinated/etc. Every day someone posts a link and everyone else shares it with everybody, and before you know it the site gets DDoSed and nobody can reach it. Unless we're lucky enough the website was cached by google/bing.

Coral Cache was invented for just that.

http://www.coralcdn.org/

Can someone copy/paste the article from their browser cache? The site is down.

I'm running ubuntu 10.04 right now (upgraded this morning), it's just not that annoying. I'd like it to be fixed (I think the chromium close/redize/minimize buttons look much better), but I think the flurry about this is over statued.

People still use windows controls?

Who closes windows with the mouse these days anyway?

The same people that still use mouses.

It is a little quaint, but old habits die hard, I guess.

The vast majority of people using computers.

Wow... if only the window manager was a separate program or something... then you could just run the one that you like instead of the default one.

Oh, you mean that's true? Oh...

Good defaults are crucial to the success of a platform. Ubuntu is going for mainstream adoption. The mainstream wants their computers to "just work". They don't want to have to change themes or window managers, they just want the buttons to live where they're expected.

Even if the users were willing to change the WM for the button position, almost no user knows how to change the WM, and it probably necessitates editing .desktop files or .xinitrc or something (unless you use something like fusion-icon to control it, which might be reasonable but is still extra configuration).

Furthermore, this is custom to that theme. Any other Metacity theme has the buttons on the right, so they'd just have to change the theme, not the WM, to get the buttons where desired.

> They don't want to have to change themes or window managers, they just want the buttons to live where they're expected.

By this reasoning no one on Windows would ever switch to OSX.

At least OSX and Windows both put the close button in a corner, this theme puts it between the maximise button and the window title. I've never seen another OS put it in such a strange position.

How does that negate my statement? Normal people don't look for things like, "Is the close button on the outside of the window?" Are you really going to tell me that the only reason that people tolerate a switch from Windows to MacOS X is because the close-window button is still towards the outside of the window even though the buttons are on the opposite side?

Sometimes frustration with one system can cause one to tolerate the learning curve on a new one.

There are some people who don't mind learning things about their computers but in my experience they are pretty rare as a segment of the whole population.

That's fine unless you're billing yourself as a user friendly Linux. Personally I think the issue of mis-clicking while going for the file menu would be infuriating.

Most users care more about looks than functionality, and not even much about that. People still use Windows XP and are perfectly happy, right?

I also like how people invent mysterious "other people" and "average users" to use as arguments when they don't like something. The average user also thinks that technique is suspect. Why not just say, "I wish they didn't change this" instead of saying, "the average user is too clumsy to use this".

My biggest concern here would be going for the "Edit" menu and hitting "Close" by accident. I know I miss the item I'm aiming for from time to time.

Otherwise I could careless, people learned to use Hotmail, Facebook, and eBay just fine.

I wish I could see a graph of the number of votes on this post over time. I've looked at it a few times today; at one point it was 5, now it's -3.

Everyone who is complaining is smart enough to easily change the theme to something they are more used to. Less technical people are either: 1. Not using Ubuntu, probably never will regardless of the position of window controls. 2. Perfectly happy to learn to click wherever required, as is evident when they purchase Macs and use those.

It's not about preference - it's about usability (if we leave out strange release management out of it):

- proximity to "File" menu

- proximity to top "Applications" menu

- inconsistency (windows close on the left, tabs and system on the right)

Not to mention that, according to TFA, no existing themes work with the new structure: changing to another theme can't help.

It's not that no themes work, if I understand correctly, it's that any other theme would still have the buttons on the right. Someone please correct me if this is wrong.

IIRC, the theme itself has control over where the buttons on the window titlebar go. I know for sure that this is the case in Emerald, and I know that other window managers allow various themes to move around the position of the buttons at will. If they changed the underlying code to force all themes to use the same button position regardless of what they have it set to... that's a lot bigger of an issue than just changing the button position in the default theme.

Maybe someone running 10.04 Alpha 3 can confirm that this is being forced upon other themes. Another thought is that maybe most themes don't specify the button locations because they are used to the defaults being 'buttons on the right,' and this Ubuntu change messes with the global defaults rather than just applying the changes locally to the new default theme.

Sorry, I completely retract any statements I made lauding the intelligence of people loudly complaining about the default theme change. They are definitely _not_ smart enough to change the theme (note confusion in comments surrounding this issue). It's Gnome, people.

Reply on news.ycombinator.com