Very interesting. Seeing this I can't help but feel how much more tangible and solid the old Windows interface was looking. I miss buttons that look like buttons and Window edges that are so hard-lined they're easy to grab and visually separate from underlying windows.
You're very much right and I agree that modern designs like making stuff flat is a huge step backwards. If you want to remove 3d widgets, then at least implement mouse over highlighting so that one can see what can be interacted with. Another regression is making stuff larger and occupying screen space with big title bars or empty white spaces.
Something similar happens in web pages where you're supposed to click a link but the styling doesn't it allow to be a conventional link, so you don't know where to click. Interface designers have a responsibility to not break usability if they diverge from well known widget styles.
"Another regression is making stuff larger and occupying screen space with big title bars or empty white spaces."
Or ... making some stuff larger but not others.
In OSX, I can use a slider to make stuff 'bigger' in Finder, but... it only increases the icon size, not the text size, which makes using the UI from a distance impossible, even though they've got "make stuff bigger" built in.
Yes! The current flat design trend makes it difficult to distinguish clickable from non-clickable elements due to the fact that they are all flat. Android "Material Design" is even worse because of the dreadful pastel color scheme which makes things even worse.
That's interesting you find Material Design to be a further regression compared to flat design. Material Design's primary goal is re-establishing depth and dimension to interfaces with implied shadows and volumes across multiple depth planes, a move designed to reclaim what flat design has taken away.
Thank you! I knew there had to be somebody else who can't stand flat design!
IMO, pseudo-3D appearances like Windows 95, NeXTStep, Mac OS 8, and early KDE/GNOME were the pinnacle of aesthetic GUI design (and Windows 95 in particular is the pinnacle of UX as a whole), and everything since then has been a step backward.
I think about the only GUI aesthetic I can stand that's come out of the 21st century has been the use of transparent glass on some elements, and that fad has already died off, sadly. I have my KDE configured to look like a Windows 95-ish design (i.e. pseudo-3D and gray) for most things, but render a small handful of other things with glass, and it works pretty well for me.
While I do hope that this is "just a phase", I don't think it will pass too quickly. The one thing that flat design excels at is quickly and cheaply creating "compliant" UIs and associated elements.
A former colleague of mine sketched the situation bitterly over a few beers; ten years ago or so, he'd spend days, sometimes even a week on an OS X application's icon. It was exhausting and it was hard to convince clients to pay him a week's worth of hourly fee for a damn icon. Nowadays (well, nowadays was about two years ago, I think), he could have two or three mockups ready in an evening and the hip entrepreneurs he worked for on the side would fall off their chairs gasping at how "clean" and "intrinsically meaningful" those blobs of colour were.
We joked around that a lot of the activity could be automated by just writing clever software that takes a 3D picture or a good enough photo of something and makes a "flat" version of it, requiring only a little intervention on a vectorized version to get something final-ish. Voila, icon "design" as a service.
I got to boot a Pentium III desktop (Dell Dimension XPS 1450) few monthes ago. I missed nothing about today's UX. It was no brainer yet not restricting for 90% of your needs. Extremely light and fast (the whole OS runs in less memory than an average webpage). I almost tried to build emacs and a scheme and live in heaven forever.
There was a classical canonical sweet spot in this era design mindset. Kinda like going into a country side cabin. No fancy stuffy, yet most things are just like they need to.
ps: It's hard to remove nostalgia / habit bias; but I consider myself far from prejudiced, having gone into almost all UX fads since I had a keyboard and a mouse. I accept qualities in all systems be it Apple, Linux, Amiga, SGI, HP handeld calcs and all kinds of different ergonomics (CAD, Compositing, Video, Audio, code, shell, IDEs, Smalltalk) ... And even after seeing all of this, Old Windows (until 2K) had a nice sweet spot feel.
I definitely agree with you! The first computer I could call my own was a hand-me-down eMachines eTower 733i Mini-tower desktop with a 733MHz Celeron, upgraded to 256MB of RAM, and a 20GB HDD I was gifted around 2004. I installed a copy of Windows 2000 on it and used it daily for another 2-3 years. I miss the minimal, no-frills UI that caused little-to-no performance hiccups. And I definately feel similar about the Classic MacOS "Platinum" interface (Mac OS 8 thru 9). Despite both sharing what many percieve as a "boring" gray color scheme, the 3D look of the buttons and UI elements made things easy to see and understand their function. I currently work at a company where I find myself often using monitors connected to KVM switches, often with improperly adjusted contrast and brightness. Modern flat web application GUIs in this setting are barely usable, forcing the user to struggle to determine where a button or text area's boundaries are. Though I have to hope this environment is not the norm for 90% of users, I wish modern interface design encouraged such "GUI robustness" and functional minimalism whereby the UI remains functional despite the display's configuration. A similar poor-contrast situation could arise when viewing a display from an improper angle or in a very bright environment.
Not only that, but buttons usually had hotkeys (underlined letters) that made the OS very easy to use without a mouse. Until we moved away from Windows 2000, colleagues used to be amazed at how fast I navigated. With the precedent set by XP, many UI's became HTML-like, and instead of buttons, we'd have links. Mouseless navigation became impossible with more and more apps. And now, I've lost that habit.
I don't use Windows these days, but Linux GUI programs are not as well-designed as the old Windows programs, so I find myself reaching for the mouse every now and then.
Linux GUIs follow different standards for different toolkits. Some try to use a relatively consistent core (GNOME, KDE, and xfce4 to an extent, I think), others can diverge widely. If you're old enough, or like collecting old software, you'll know several wildly inconsistent scrollbar metaphors and controls....
I find it slightly useful -- I can usually visually guestimate a tool's graphics toolkit and predict its behavior, but novice users will almost certainly get confused.
Like you I find environments I cannot keyboard control are hugely unproductive.
Note that this was violated pretty soon thereafter, as one of the first Office versions made toolbar buttons flat. Heck, basically every Office version introduce some new UI stuff, often nicer looking yet conceptually impure.
Seeing the windows 3.1 filesystem reminds me every time of the "Load DB from Backup" in SQL Management Studio. Even 2008r2 had the same restricted win 3.1 equivalent dialog.
It's amazing how such a dialog held out for so long resisting a change to a modern modern file selector.
Which brings me to windows 10, which has different dialogs for the same dialogs depending how you launch them. There are two different "Display Settings" dialogs depending on if you get there via the "control panel" desktop app or "Settings".
At least windows 95 was reasonably consistent, even if it feels restrictive after using windows XP and then windows 7 especially.
I find there's almost no effort made for UI on the enterprise level of things. Which is understandable nowadays with powershell, but previously the GUI was usually the only way to get things done and seeing the old simpler interfaces was the norm. I guess this is what happens when you have engineers in charge and product managers not caring about the experience. This should lead to wonderful things, but in the MS world at least, it leads to old broken dialogs like pressing "open" to create a new file, not remembering previous folders and being thrust into c:\windows\system32 everytime, lack of handy quick access and shortcut buttons, and lots of F5 mashing because nothing refreshes itself for some reason.
Everytime I deal with MS enterprise products it just blows me over how anything of theirs actually works. There seems to be a strong empahsis to dogfood everything (okay to install this you need a .net version you don't have, a bunch of random libraries, some framework thing MS is kinda sorta pushing, a random KB or two, maybe silverlight, and then install process itself requires multiple reboots and usually troubleshooting). Then you have to install the application itself and its many hotfixes and service packs. This will definitely result in troubleshooting cryptic error codes as none of this stuff really works out of the box.
That said, while 10 and 2012 are perhaps uglier and less refined beasts, there does seem to be an emphasis on a single UI style with a real effort to hide the legacy stuff, as well as a much stronger emphasis on getting away from the GUI and just using powershell. No idea how that will ultimately pan out, but I do miss the old start menus. The 10 menu seems like it would be amazing, like you should be able to stretch it and fill it with tiles and shortcuts trivially. Instead it seems like something a little hostile to you and something MS uses to promote whatever it feels lime promoting at the time (xbox, bing, cortana, etc).
SSMS' backup dialog has to show the server filesystem, exposed over a TDS protocol. This means the implementers cannot use any off-the-shelf OS/Shell file selector dialog (comdlg) and had to design their own. Any 'advancement' in the OS/Shell dialogs would have to be 'matched', manually, in the SSMS code base. I'm not trying to defend the visual appearance of the said dialog, I'm just trying to give context why that particular file selection dialog has nothing to do with the OS/Shell file selection dialogs. Think at it more like a query result browsing dialog.
I consider the different dialogs in windows 10 a benefit of a well decoupled interface from the underlying application instead of frowning upon it.
Let's be real, they cannot change the whole interface in a single release so they are rolling things out partially, and good application design is what allows them to do so.
Maybe it's a bit inconsistent but it's the more pragmatic approach.
I still maintain that Windows 95 was the pinnacle of desktop UX. Nobody else, before, since, or at the same time, has released anything better.
Every Windows version since has been incrementally worse. Hell, Microsoft started ruining it with Windows 98, by merging Windows Explorer and IE into something really terrible. I still miss Windows 95 Explorer. And then there was Microsoft's obsession with trying to turn the desktop into a web page... they ruined a lot of things, like the Find utility, and several Control Panel applets with that.
Been a very, very long time since I had this argument, but I do think System 7 was better. Absolutely not technically, but definitely in terms of a coherent and consistent desktop metaphor, especially across shell and apps.
W95 broke its own metaphors all the time (hell, it supported both MDI and SDI!) and wasn't consistent. Sure, you can drag and drop this document icon to this window, but try to drag a document icon on top of a window minimized to taskbar and you get a popup that tells you "we know what you want to do and why, but we're not going to let you. Do something else", which is a cardinal UX sin.
Even the look, which I really liked at the time, is a shameless lift of NextSTEP (another competitor for "pinnacle of desktop UX") with other bits pinched from OS/2.
What made me pick Win95 over System 7 is that Win95 absolutely nailed window management, while System 7's lack of good window management drove me up the wall.
The taskbar, and more importantly the ability to minimize windows to the taskbar, was a godsend. WindowShade didn't compare (and it didn't come with the OS until 7.5), and the oddball options in System 7's multitasking menu were frustrating to deal with. System 7 was a singletasking OS hacked to support multitasking, and it really shows.
About the only thing System 7 did better than Win95 with regards to window management was that they made sure to put the close box on the opposite side of the titlebar from the zoom box and the shade box, to prevent a misclick from closing your window (I've since incorporated this into my KDE configuration).
Oddly enough, it's the technical aspects of the classic Mac OS that fascinate the hell out of me. The whole OS was made of clever hacks. Resource forks were a stroke of genius, and I wish modern OSes had them. I spent so much time hacking around with ResEdit on my Mac as a kid, and later on I learned to appreciate the subtler things about resource forks. Like how the System 7.1 (and above) Finder lets you drag and drop FONT resources between files as if they were folders (they called this a "suitcase" instead of a "folder", but the UI was identical). And then there was the brilliant -- and rather scary -- hackery of how desk accessories were implemented (something that became unnecessary with System 7, but Apple still had the sense to hack the System 7 Finder to treat DA suitcases as programs, which itself was pretty clever).
Or, for that matter, System 7's extensions were the single most flexible method of modifying the behavior of the OS I've ever seen. Mind you, it wasn't exactly stable (untangling extension conflicts was a nightmare, even after 7.5 added a utility to help with it), and it was certainly insecure, but I really miss that flexibility and hackability in modern OSes. The only thing I've seen that comes close is Android's Xposed Framework.
Honestly, if there was one thing I missed about the '90s, it was the UX design across the board. Just thinking that Windows 95, System 7, and NeXTStep were all competing with each other at the same time makes me realized how blessed that period of time was, and how things have really gone downhill since.
Windows 95 is the best of Windows — no doubt about it. I think NeXTSTEP was the pinnacle of all UX and it is pretty clear to me that Windows 95 copied NeXTSTEP, albeit rather badly. If only Apple would return to a refined NeXTSTEP like interface, I'd be much happier than with this disastrous flat-transparent-color-bleed-blurry-crap interface. The way Apple's been going feels like Windows 98 era of Windows. The webification of Windows ruined it for me and whatever you call what Apple is currently doing, is ruining macOS for me now.
I think Apple's contemporary OSes (System 7 and System 8) did a much better job, personally.
Windows 95 has the advantage that tons of people used it day-to-day, where Apple's systems were less popular. But as a person who used both, I definitely preferred Apple's design, and their dedication to designing everything about the experience.
Even the keyboard layout. Macintosh was the system that logically and sensibly separated the function of "Enter" from "Return". Two functions should have two keys, dammit. Every time I have to use shift-Enter to put a newline in a IM window instead of just hitting return, I die a little inside.
One thing I'd like to see explained is the location of "Programs" at the top of the Start menu, rather than right at the bottom where it would be closer and less mouse movement. I do think the Start menu is a much better location for the program list than their earlier idea, a Programs folder on the "Desktop," and better still than the Windows 3.0 Program Manager. But I wonder what the thinking behind making the user mouse up past Run, Find, and Settings to get to Programs, when their research indicates that "opening a program" was the primary objective of a user. I know it's an old joke that "you have to click Start to Shut Down," but it does seem far from the way the Apple menu worked at the time for effectively the same thing.
"Launching Programs: Start Menu. Although we abandoned the idea of a separate shell for beginners, we salvaged its most useful features: single-click access, high visibility, and menu-based interaction. We mocked up a number of representations in Visual Basic and tested them with users of all experience levels, not just beginners, because we knew that the design solution would need to work well for users of varying experience levels. Figure 5 shows the final Start Menu, with the Programs sub-menu open. The final Start Menu integrated functions other than starting programs, to give users a single-button home base in the UI."
I think it's because people read top-to-bottom. If they had located the task bar on the top of the screen, they could have had programs listed on top and it would work for reading order and mouse movement.
Well by that logic, why's the Start menu at the bottom? (Rhetorical question: I assume it's because then it would have been TOO much like the Apple menu).
> One thing I'd like to see explained is the location of "Programs" at the top of the Start menu, rather than right at the bottom where it would be closer and less mouse movement.
If you move the taskbar to the left or top, this problem is dealt with. (As is the problem that later versions of windows -- before they got rid of it altogether -- moved "Programs" below the dynamically-added links to things that Windows thinks you are more likely to use, which keeps those more-likely things farther from the start button if you keep the taskbar at the bottom.)
I wonder if the taskbar was originally planned for one of those two (top/left) locations, but got moved to the bottom later.
We're in an industry where absolutely nobody learns anything from the past.
People going to school to learn software development don't spend a week on DOS, a week on Novell 4, then a week on Macintosh 6.0.8, then a week on Windows 98, a week on BeOS, etc. There's no history education at all. No learning from past ideas. Not even any conception of history, really.
Compare it to, for example, a film class. Students learning how to create films watch films. They watch films from the 1910s, the 1930s, the 1950s, etc. They learn how their industry changed and matured over time. They get a solid sense of history, and an appreciation for their forebears. Software developers get none of that.
Anyway, this shouldn't come as a surprise, is what I'm getting at.
It's an unfortunate aspect of the human condition that every generation, in their youthful defiance, ends up discarding some of their forebears' painstakingly accumulated wisdom, only to rediscover it later (of course crediting themselves for the discovery). To paraphrase Santanaya, those who cannot remember the past are condemned to reinvent it.
Though I think I really have to write some user CSS or a bookmarklet or something for pages like this, namely:
body {
width: 50em;
margin: auto;
line-height: 1.6;
}
The Firefox Reader View is supposed to do that but it's way too narrow (30em I think) and way too often messes up some formatting essential to understanding the content.
<title> tags were rendered in Internet Explorer as a tooltip. Other browsers would only render the alt tag as a tooltip, but there was pressure on them to also render the <title> because IE was the dominant browser at the time.
My main question here is whether this kind of rigorous user lab testing - at a time when things could still be changed - was also applied to Windows 8. That thing seemed such a top down dictated hot mess to me.
IMO the issue of productivity and discoverability has still not been solved on touch based UI, even with our post-iPhone interfaces. Shame that Microsoft didn't go back to these roots and innovated on exactly these issues.
I don't think anybody does that type of user testing anymore. The last project I've seen a good write-up about was Office 2007, and that's a decade old at this point. Which is a shame.
Just using most web-based UIs shows that there's almost zero usability testing going on in that field, and since more and more apps are web-based, well.
You'd think that MS would want to replicate their success with Win95. The task bar and start menu is easily the most iconic UI to ever come out from them and made them jump ahead of MacOS in terms of usability - and apparently it can all be traced back to this lab. The people responsible should all have been promoted to A level where their lab becomes the equivalent of a Steve Jobs product curator.
But it seems like MS at some point got on the high horse and started thinking that they can just shove anything down user's throat anyways. When they lost the entire segment of the fastest growing computing market, a competent CEO would have gone back and study why their 95 UI was so successful that everyone wanted to buy their product instead of being forced to.
Very interesting. Seeing this I can't help but feel how much more tangible and solid the old Windows interface was looking. I miss buttons that look like buttons and Window edges that are so hard-lined they're easy to grab and visually separate from underlying windows.
You're very much right and I agree that modern designs like making stuff flat is a huge step backwards. If you want to remove 3d widgets, then at least implement mouse over highlighting so that one can see what can be interacted with. Another regression is making stuff larger and occupying screen space with big title bars or empty white spaces.
Something similar happens in web pages where you're supposed to click a link but the styling doesn't it allow to be a conventional link, so you don't know where to click. Interface designers have a responsibility to not break usability if they diverge from well known widget styles.
"Another regression is making stuff larger and occupying screen space with big title bars or empty white spaces."
Or ... making some stuff larger but not others.
In OSX, I can use a slider to make stuff 'bigger' in Finder, but... it only increases the icon size, not the text size, which makes using the UI from a distance impossible, even though they've got "make stuff bigger" built in.
9 replies →
Yes! The current flat design trend makes it difficult to distinguish clickable from non-clickable elements due to the fact that they are all flat. Android "Material Design" is even worse because of the dreadful pastel color scheme which makes things even worse.
That's interesting you find Material Design to be a further regression compared to flat design. Material Design's primary goal is re-establishing depth and dimension to interfaces with implied shadows and volumes across multiple depth planes, a move designed to reclaim what flat design has taken away.
2 replies →
Pretty much everything is clickable though. At some point the UI chrome starts outweighing the content.
Even in Windows 95, each program in the programs list didn't look like a 3D button. That inconsistency can be confusing too.
2 replies →
Thank you! I knew there had to be somebody else who can't stand flat design!
IMO, pseudo-3D appearances like Windows 95, NeXTStep, Mac OS 8, and early KDE/GNOME were the pinnacle of aesthetic GUI design (and Windows 95 in particular is the pinnacle of UX as a whole), and everything since then has been a step backward.
I think about the only GUI aesthetic I can stand that's come out of the 21st century has been the use of transparent glass on some elements, and that fad has already died off, sadly. I have my KDE configured to look like a Windows 95-ish design (i.e. pseudo-3D and gray) for most things, but render a small handful of other things with glass, and it works pretty well for me.
There are plenty of us :-).
While I do hope that this is "just a phase", I don't think it will pass too quickly. The one thing that flat design excels at is quickly and cheaply creating "compliant" UIs and associated elements.
A former colleague of mine sketched the situation bitterly over a few beers; ten years ago or so, he'd spend days, sometimes even a week on an OS X application's icon. It was exhausting and it was hard to convince clients to pay him a week's worth of hourly fee for a damn icon. Nowadays (well, nowadays was about two years ago, I think), he could have two or three mockups ready in an evening and the hip entrepreneurs he worked for on the side would fall off their chairs gasping at how "clean" and "intrinsically meaningful" those blobs of colour were.
We joked around that a lot of the activity could be automated by just writing clever software that takes a 3D picture or a good enough photo of something and makes a "flat" version of it, requiring only a little intervention on a vectorized version to get something final-ish. Voila, icon "design" as a service.
I got to boot a Pentium III desktop (Dell Dimension XPS 1450) few monthes ago. I missed nothing about today's UX. It was no brainer yet not restricting for 90% of your needs. Extremely light and fast (the whole OS runs in less memory than an average webpage). I almost tried to build emacs and a scheme and live in heaven forever.
There was a classical canonical sweet spot in this era design mindset. Kinda like going into a country side cabin. No fancy stuffy, yet most things are just like they need to.
ps: It's hard to remove nostalgia / habit bias; but I consider myself far from prejudiced, having gone into almost all UX fads since I had a keyboard and a mouse. I accept qualities in all systems be it Apple, Linux, Amiga, SGI, HP handeld calcs and all kinds of different ergonomics (CAD, Compositing, Video, Audio, code, shell, IDEs, Smalltalk) ... And even after seeing all of this, Old Windows (until 2K) had a nice sweet spot feel.
I definitely agree with you! The first computer I could call my own was a hand-me-down eMachines eTower 733i Mini-tower desktop with a 733MHz Celeron, upgraded to 256MB of RAM, and a 20GB HDD I was gifted around 2004. I installed a copy of Windows 2000 on it and used it daily for another 2-3 years. I miss the minimal, no-frills UI that caused little-to-no performance hiccups. And I definately feel similar about the Classic MacOS "Platinum" interface (Mac OS 8 thru 9). Despite both sharing what many percieve as a "boring" gray color scheme, the 3D look of the buttons and UI elements made things easy to see and understand their function. I currently work at a company where I find myself often using monitors connected to KVM switches, often with improperly adjusted contrast and brightness. Modern flat web application GUIs in this setting are barely usable, forcing the user to struggle to determine where a button or text area's boundaries are. Though I have to hope this environment is not the norm for 90% of users, I wish modern interface design encouraged such "GUI robustness" and functional minimalism whereby the UI remains functional despite the display's configuration. A similar poor-contrast situation could arise when viewing a display from an improper angle or in a very bright environment.
Not only that, but buttons usually had hotkeys (underlined letters) that made the OS very easy to use without a mouse. Until we moved away from Windows 2000, colleagues used to be amazed at how fast I navigated. With the precedent set by XP, many UI's became HTML-like, and instead of buttons, we'd have links. Mouseless navigation became impossible with more and more apps. And now, I've lost that habit.
I don't use Windows these days, but Linux GUI programs are not as well-designed as the old Windows programs, so I find myself reaching for the mouse every now and then.
Linux GUIs follow different standards for different toolkits. Some try to use a relatively consistent core (GNOME, KDE, and xfce4 to an extent, I think), others can diverge widely. If you're old enough, or like collecting old software, you'll know several wildly inconsistent scrollbar metaphors and controls....
I find it slightly useful -- I can usually visually guestimate a tool's graphics toolkit and predict its behavior, but novice users will almost certainly get confused.
Like you I find environments I cannot keyboard control are hugely unproductive.
Note that this was violated pretty soon thereafter, as one of the first Office versions made toolbar buttons flat. Heck, basically every Office version introduce some new UI stuff, often nicer looking yet conceptually impure.
Plus ca change...
Seeing the windows 3.1 filesystem reminds me every time of the "Load DB from Backup" in SQL Management Studio. Even 2008r2 had the same restricted win 3.1 equivalent dialog.
It's amazing how such a dialog held out for so long resisting a change to a modern modern file selector.
Which brings me to windows 10, which has different dialogs for the same dialogs depending how you launch them. There are two different "Display Settings" dialogs depending on if you get there via the "control panel" desktop app or "Settings".
At least windows 95 was reasonably consistent, even if it feels restrictive after using windows XP and then windows 7 especially.
I find there's almost no effort made for UI on the enterprise level of things. Which is understandable nowadays with powershell, but previously the GUI was usually the only way to get things done and seeing the old simpler interfaces was the norm. I guess this is what happens when you have engineers in charge and product managers not caring about the experience. This should lead to wonderful things, but in the MS world at least, it leads to old broken dialogs like pressing "open" to create a new file, not remembering previous folders and being thrust into c:\windows\system32 everytime, lack of handy quick access and shortcut buttons, and lots of F5 mashing because nothing refreshes itself for some reason.
Everytime I deal with MS enterprise products it just blows me over how anything of theirs actually works. There seems to be a strong empahsis to dogfood everything (okay to install this you need a .net version you don't have, a bunch of random libraries, some framework thing MS is kinda sorta pushing, a random KB or two, maybe silverlight, and then install process itself requires multiple reboots and usually troubleshooting). Then you have to install the application itself and its many hotfixes and service packs. This will definitely result in troubleshooting cryptic error codes as none of this stuff really works out of the box.
That said, while 10 and 2012 are perhaps uglier and less refined beasts, there does seem to be an emphasis on a single UI style with a real effort to hide the legacy stuff, as well as a much stronger emphasis on getting away from the GUI and just using powershell. No idea how that will ultimately pan out, but I do miss the old start menus. The 10 menu seems like it would be amazing, like you should be able to stretch it and fill it with tiles and shortcuts trivially. Instead it seems like something a little hostile to you and something MS uses to promote whatever it feels lime promoting at the time (xbox, bing, cortana, etc).
I don't think that "held out" is what happened with these dialogs. I think it's more like "don't fix what isn't broken" or "we forgot about that".
Very true. For the longest time, the dialog box to install new fonts on Windows was a hold-over from Win 3.1.
1 reply →
SSMS' backup dialog has to show the server filesystem, exposed over a TDS protocol. This means the implementers cannot use any off-the-shelf OS/Shell file selector dialog (comdlg) and had to design their own. Any 'advancement' in the OS/Shell dialogs would have to be 'matched', manually, in the SSMS code base. I'm not trying to defend the visual appearance of the said dialog, I'm just trying to give context why that particular file selection dialog has nothing to do with the OS/Shell file selection dialogs. Think at it more like a query result browsing dialog.
I consider the different dialogs in windows 10 a benefit of a well decoupled interface from the underlying application instead of frowning upon it.
Let's be real, they cannot change the whole interface in a single release so they are rolling things out partially, and good application design is what allows them to do so.
Maybe it's a bit inconsistent but it's the more pragmatic approach.
I believe there's a font selection dialog lib that is still present in every windows system folder since 3.1.
I still maintain that Windows 95 was the pinnacle of desktop UX. Nobody else, before, since, or at the same time, has released anything better.
Every Windows version since has been incrementally worse. Hell, Microsoft started ruining it with Windows 98, by merging Windows Explorer and IE into something really terrible. I still miss Windows 95 Explorer. And then there was Microsoft's obsession with trying to turn the desktop into a web page... they ruined a lot of things, like the Find utility, and several Control Panel applets with that.
Been a very, very long time since I had this argument, but I do think System 7 was better. Absolutely not technically, but definitely in terms of a coherent and consistent desktop metaphor, especially across shell and apps.
W95 broke its own metaphors all the time (hell, it supported both MDI and SDI!) and wasn't consistent. Sure, you can drag and drop this document icon to this window, but try to drag a document icon on top of a window minimized to taskbar and you get a popup that tells you "we know what you want to do and why, but we're not going to let you. Do something else", which is a cardinal UX sin.
Even the look, which I really liked at the time, is a shameless lift of NextSTEP (another competitor for "pinnacle of desktop UX") with other bits pinched from OS/2.
What made me pick Win95 over System 7 is that Win95 absolutely nailed window management, while System 7's lack of good window management drove me up the wall.
The taskbar, and more importantly the ability to minimize windows to the taskbar, was a godsend. WindowShade didn't compare (and it didn't come with the OS until 7.5), and the oddball options in System 7's multitasking menu were frustrating to deal with. System 7 was a singletasking OS hacked to support multitasking, and it really shows.
About the only thing System 7 did better than Win95 with regards to window management was that they made sure to put the close box on the opposite side of the titlebar from the zoom box and the shade box, to prevent a misclick from closing your window (I've since incorporated this into my KDE configuration).
Oddly enough, it's the technical aspects of the classic Mac OS that fascinate the hell out of me. The whole OS was made of clever hacks. Resource forks were a stroke of genius, and I wish modern OSes had them. I spent so much time hacking around with ResEdit on my Mac as a kid, and later on I learned to appreciate the subtler things about resource forks. Like how the System 7.1 (and above) Finder lets you drag and drop FONT resources between files as if they were folders (they called this a "suitcase" instead of a "folder", but the UI was identical). And then there was the brilliant -- and rather scary -- hackery of how desk accessories were implemented (something that became unnecessary with System 7, but Apple still had the sense to hack the System 7 Finder to treat DA suitcases as programs, which itself was pretty clever).
Or, for that matter, System 7's extensions were the single most flexible method of modifying the behavior of the OS I've ever seen. Mind you, it wasn't exactly stable (untangling extension conflicts was a nightmare, even after 7.5 added a utility to help with it), and it was certainly insecure, but I really miss that flexibility and hackability in modern OSes. The only thing I've seen that comes close is Android's Xposed Framework.
Honestly, if there was one thing I missed about the '90s, it was the UX design across the board. Just thinking that Windows 95, System 7, and NeXTStep were all competing with each other at the same time makes me realized how blessed that period of time was, and how things have really gone downhill since.
1 reply →
Windows 95 is the best of Windows — no doubt about it. I think NeXTSTEP was the pinnacle of all UX and it is pretty clear to me that Windows 95 copied NeXTSTEP, albeit rather badly. If only Apple would return to a refined NeXTSTEP like interface, I'd be much happier than with this disastrous flat-transparent-color-bleed-blurry-crap interface. The way Apple's been going feels like Windows 98 era of Windows. The webification of Windows ruined it for me and whatever you call what Apple is currently doing, is ruining macOS for me now.
I think Apple's contemporary OSes (System 7 and System 8) did a much better job, personally.
Windows 95 has the advantage that tons of people used it day-to-day, where Apple's systems were less popular. But as a person who used both, I definitely preferred Apple's design, and their dedication to designing everything about the experience.
Even the keyboard layout. Macintosh was the system that logically and sensibly separated the function of "Enter" from "Return". Two functions should have two keys, dammit. Every time I have to use shift-Enter to put a newline in a IM window instead of just hitting return, I die a little inside.
From the article:
> We also collected data from product support about users' top twenty problems with Windows 3.1.
As someone with a long term interest in Shell design, I'd love to know what these were ?
One thing I'd like to see explained is the location of "Programs" at the top of the Start menu, rather than right at the bottom where it would be closer and less mouse movement. I do think the Start menu is a much better location for the program list than their earlier idea, a Programs folder on the "Desktop," and better still than the Windows 3.0 Program Manager. But I wonder what the thinking behind making the user mouse up past Run, Find, and Settings to get to Programs, when their research indicates that "opening a program" was the primary objective of a user. I know it's an old joke that "you have to click Start to Shut Down," but it does seem far from the way the Apple menu worked at the time for effectively the same thing.
http://www.sigchi.org/chi96/proceedings/desbrief/Sullivan/kd...
"Launching Programs: Start Menu. Although we abandoned the idea of a separate shell for beginners, we salvaged its most useful features: single-click access, high visibility, and menu-based interaction. We mocked up a number of representations in Visual Basic and tested them with users of all experience levels, not just beginners, because we knew that the design solution would need to work well for users of varying experience levels. Figure 5 shows the final Start Menu, with the Programs sub-menu open. The final Start Menu integrated functions other than starting programs, to give users a single-button home base in the UI."
I think it's because people read top-to-bottom. If they had located the task bar on the top of the screen, they could have had programs listed on top and it would work for reading order and mouse movement.
Well by that logic, why's the Start menu at the bottom? (Rhetorical question: I assume it's because then it would have been TOO much like the Apple menu).
> One thing I'd like to see explained is the location of "Programs" at the top of the Start menu, rather than right at the bottom where it would be closer and less mouse movement.
If you move the taskbar to the left or top, this problem is dealt with. (As is the problem that later versions of windows -- before they got rid of it altogether -- moved "Programs" below the dynamically-added links to things that Windows thinks you are more likely to use, which keeps those more-likely things farther from the start button if you keep the taskbar at the bottom.)
I wonder if the taskbar was originally planned for one of those two (top/left) locations, but got moved to the bottom later.
Am I the only one who misses the Program Manager? I've never found a reimplementation of the idea in any OS I've used since 3.1.
Windows 95 actually still included the program manager and you could have used it as your shell, if you really wanted: http://toastytech.com/guis/win95progman.png
1 reply →
Make a shortcut to the start menu folder so that it opens in the file explorer?
Now there's no Programs button at all. Just a list of apps.
This is quite fascinating. All these practices are touted as modern design techniques, however they were being done over 20 years ago.
We're in an industry where absolutely nobody learns anything from the past.
People going to school to learn software development don't spend a week on DOS, a week on Novell 4, then a week on Macintosh 6.0.8, then a week on Windows 98, a week on BeOS, etc. There's no history education at all. No learning from past ideas. Not even any conception of history, really.
Compare it to, for example, a film class. Students learning how to create films watch films. They watch films from the 1910s, the 1930s, the 1950s, etc. They learn how their industry changed and matured over time. They get a solid sense of history, and an appreciation for their forebears. Software developers get none of that.
Anyway, this shouldn't come as a surprise, is what I'm getting at.
> People going to school to learn software development don't
...usually have offered anything that even purports to be a "software development" curriculum.
They usually have a "computer science" curriculum.
2 replies →
It's an unfortunate aspect of the human condition that every generation, in their youthful defiance, ends up discarding some of their forebears' painstakingly accumulated wisdom, only to rediscover it later (of course crediting themselves for the discovery). To paraphrase Santanaya, those who cannot remember the past are condemned to reinvent it.
Go watch (or read the descriptions of) the Mother of All Demos, and weep.
1968.
https://m.youtube.com/watch?v=yJDv-zdhzMY
https://en.m.wikipedia.org/wiki/Mother_of_all_demos
That was really interesting, it reads like a write up of a modern project (other than having to justify going non-Waterfall).
It's interesting to see HTML from that time period too. Did HTML tags in <title> used to be rendered in browsers?
In some, if not all, browsers it did. I remember <marquee> and <blink> used to be pretty popular :-)
It's beautiful, is what it is. It loads instantly, doesn't download megabytes of crap and is very information-dense.
Though I think I really have to write some user CSS or a bookmarklet or something for pages like this, namely:
The Firefox Reader View is supposed to do that but it's way too narrow (30em I think) and way too often messes up some formatting essential to understanding the content.
2 replies →
It probably didn't load instantly using contemporary setups.
<title> tags were rendered in Internet Explorer as a tooltip. Other browsers would only render the alt tag as a tooltip, but there was pressure on them to also render the <title> because IE was the dominant browser at the time.
You're confusing the title element with the title attribute.
Where would a <title> display as a tooltip? And how would that render the <sup> tags like in the <title> on this page?
My main question here is whether this kind of rigorous user lab testing - at a time when things could still be changed - was also applied to Windows 8. That thing seemed such a top down dictated hot mess to me.
IMO the issue of productivity and discoverability has still not been solved on touch based UI, even with our post-iPhone interfaces. Shame that Microsoft didn't go back to these roots and innovated on exactly these issues.
I don't think anybody does that type of user testing anymore. The last project I've seen a good write-up about was Office 2007, and that's a decade old at this point. Which is a shame.
Just using most web-based UIs shows that there's almost zero usability testing going on in that field, and since more and more apps are web-based, well.
You'd think that MS would want to replicate their success with Win95. The task bar and start menu is easily the most iconic UI to ever come out from them and made them jump ahead of MacOS in terms of usability - and apparently it can all be traced back to this lab. The people responsible should all have been promoted to A level where their lab becomes the equivalent of a Steve Jobs product curator.
But it seems like MS at some point got on the high horse and started thinking that they can just shove anything down user's throat anyways. When they lost the entire segment of the fastest growing computing market, a competent CEO would have gone back and study why their 95 UI was so successful that everyone wanted to buy their product instead of being forced to.
It's interesting that they doesn't noticed that beginners often mix elements have to be double-cliked and which single-clicked.