Design evolution of the Windows control panel
versionmuseum.com
versionmuseum.com
It's form over function so hard it hurts, in a place where function is all I want. I'm not on a "journey" or having an "experience" or whatever they're aiming for. I just want to change a setting, so show me ALL the settings programs, RIGHT NOW. It's laughable how much you need to click on random stuff to see where stuff may be hiding.. Search is great, if you can remember what it's called, but often, I can't and I need to look through the options to be reminded.
The font, buttons and how everything is spaced is just perfect. I wish we could have a minimalistic Windows, that have this UI.
Just x3, x4 integer scaling so it can display on modern screens.
No crapware included by default. And then you download software you need form the net, like browsers, mailclient, torrent client, games, or what ever you need.
Does ReactOS tick the requirements? It is FOSS.
I think the 98 side panel added to 95 would be the best usability. Not every user knows it is safe to open the programs in the control panel (maybe it’ll change something if I run it!) so the select icon to view description was a really good idea imho.
That's why old Windows versions had Cancel buttons in every configuration window.
- opens a folder
- opens a document
- executes a program
I think it is reasonable for a user to assume those icons in the control panel are programs and worry whether they’ll open a dialogue with an “okay” and “cancel” option or just do something.
We have the same problem on the command line. We can reasonably assume that `config-program -—help` won’t execute anything, but we don’t know until we try. However, we are assured that `man config-program` won’t execute anything.
The sidebar brings intermediate users closer to assurance and helps novice users understand where they are.
At least that’s how I see it as a good UX addition to windows. I wouldn’t be surprised if most scared users wouldn’t even look at the sidebar.
If you design an app which writes anything into a document file as soon as you open the document in them you are doing it really wrong.
This given, the confusion you describe comes straight from lack of awareness of the UI metaphor logic in the software engineers. A well-designed GUI app (a non-GUI app should not be executed by double-clicking at all) would do virtually nothing to leave any traces (let alone break anything) right after you run them, by just launching them or "launching" any specific document to be opened in them.
But I get your point. From the practical point of view - given the reality, what we actually have, you are perfectly right.
The proper solution would have been to refresh the old one. Make a full-featured control panel that uses just slightly more modern UI paradigms and technologies. And with zero weird extension points for third parties.
But no, instead we get half the controlpanel made into a usless minimalist web design look, and the other half is still the original.
I don't know what parts of the control panel must remain for backwards compatibility (with what!?) or why someone even thought it was a good idea to have extensibility points that become a nightmare of compatibility of maintenance. Or why it's apparently more work to rewrite a single control panel page than it is to ship a whole major release of Office. But more importantly, I don't care.
It is good to look back at what they did wrong, but also what they did right. For me the most notable things:
BAD: too many windows and dialogs. The 90s GUI paradigm was all about pumping out a new window or dialog for every interaction or feature. In this regard Microsoft Windows was actually the least offender, most Unixes and Macs went even further and had each toolbar on its own window, it was madness. In this regard modern single-window apps are much better, the modern tabbed web browsers being the best example (I love that even the config dialog was turned into a tab).
GOOD: the clear visual separation of "layers". Modern GUIs are too flat for their own good, there is little visual cues separating the app window vs app content, everything has almost the same off-white background so the app toolbars and the app content sit on the same visual domain. The late 90s Windows had a very clear distinction: content have white background, app (window,menus,toolbar,widgets,...) had grey background, so the eye was automatically drawn to the one you wanted. Looking at the screenshots in the article the difference is jarring, the Windows XP/8/10 panels are mostly visually flat, there is nothing guiding your eyes to what you want (content vs toolbars), while the 95/98/2000/ME and Vista had clear visual layers.
Because programs and drivers can and did ship their own control panels, it's fairly easy (https://www.codeproject.com/Articles/29264/Control-Panel-App...). Java comes to my mind, as did some virus scanners, and drivers for all kinds of hardware - and the latter is the key why MS keeps the architecture in place, they take backwards compatibility extremely serious, to the extent you can (generally) use Windows 95-era games and other software on a Windows 10 machine!
The other part is that there is an utterly absurd amount of things that can be configured in Windows, with a lot of ways to do so: text-based configuration files on the disk (e.g. /etc/hosts), registry settings, group policies that are mapped to registry settings, classic control panel applets, new control panel applets, the whole mess that is the bootloader config... all of this needs serious effort to re-organize, for something that 99% of end-users won't ever touch.
And that explains why there is such a split: the stuff that end users regularly need (display resolution, wallpapers, audio, ...) is in the new crap look, and the stuff that even nerds rarely touch (user-wide environment variables) is in the old school CPL applets.
There was just no care for actually improving user experience, everything was shoehorned into the modern ui mainly as a push for ARM windows devices which were so slow they just never took off. MS wanted to force an enclosed ecosystem like apple iOS, but just didn’t have the hardware to make it happen.
I remember one story asking why it was so difficult to manage switching to Bluetooth speakers for the surface which was still somewhat secret at MS at the time only to be told that they did follow up with my question and only 0.005% of users ever used Bluetooth speakers with windows and it wasn’t a priority. (Obviously the number is t right, but it was something utterly tiny)
successfully? How many tried? This can btw. be an issue with telemetry: if people don't manage to do something, maybe you will never know.
On the other hand it is seemingly impossible to use the microphone of these Airpods and listen to music at the same time. The quality is atrocious. Something that I assume is possible on a Mac. (Anybody knows of a solution? I read it's because of bluetooth profiles or whatever...)
When you start to see people going in loops or doing the same things over and over but not quite reaching the destination, then you know you have UX issues.
Of course this is good only for "near misses" - you may have users who don't even know where to start - that is harder to identify.
Here's an idea: take that wireless controller thingy, the logitech thingy that talks to the keyboard and mouse, and the jabra one that moves audio between a headset and base, put them all inside the computer instead of in dongles, call it "greentooth" and charge people whatever you want for it.
If you want to connect to a pair of previously paired Bluetooth headphones in Windows 10 you can press Win+K to open the connect quick action.
But I don’t know. 7 year old me could figure out the classic control panel. I don’t think it was that hard.
Imho, it's better to observe the generalizations we each fall under, and use it to learn more about ourselves.
Great website ! Thank you for the memories.
May I suggest you to add links to archive.org downloads where the software you show is available ?
As a side note somehow (and as I see it) "evolution" has a positive connotation, i.e. an evolution step brings something that is seen as "better", and unfortunately, in the case of Windows Control Panel, this is very hard to state.
By reading the title only I had the at first sight impression of the article being some form of MS fanboysm.
Something like "How the Windows Control Panel design changed over 36 years" would IMHO be more neutral.
So: in Windows 7's control panel I've counted ten different kinds of windows—with ten somewhat different approaches to their design. Some of the widgets in there aren't used anywhere else in the OS. This only includes MS' own built-in settings windows—no third-party stuff, which frequently also has custom one-off looks and widgets.
And apparently, things became even worse in Win 8 or 10, with half-assed attempts at a redesign.
The fact that Updates are under Security, and not under Apps does say something about Windows, though...
It was never what to do there.
Still better than the stupid dog when trying to search files.
What frustrates me about Windows 10 is that each one has settings which are exclusive to it, so you always need both, and never know where to look.
More importantly, they've gotten pretty good about linking into legacy Control Panel applets from the modern settings app-- so, even if a setting hasn't been migrated yet, you should be able to find a link to it in a logical place from the new Settings app. That's not new in Windows 11 (Windows 10 is pretty good about it too, for the most part), but the Settings reorganization in Windows 11 helped this along a bit.
Especially all the network configuration variants went from bad to worse over the time.
One more issue is that they kept "evolving" the design (classic, settings view, etc) without retiring previous interfaces. Now you have access to everything and you keep wondering how the controls and components all interact. Almost as catastrophic as explorer.exe and "shell" integrations, resulting in the most laggy explorer.exe yet.
They just borked the whole execution and burned the bridges that many of us had, with the multiple rewrites since Windows 8 came to be, dropping support for devices and finally dropping even .NET Native and C++/CX.
The irony is that Windows 11 is so rushed out, that they needed to still make use of UWP for the new UI regardless of WinUI 3.0 marketing.
Not only was the new name less descriptive and harder to search, but it also made that setting to no longer appear as one of the first icons (they are sorted alphabetically).
I'm still mad about that change, and to this day I still take a couple failed attempts to find it.
Thanks Microsoft
Only Microsoft could unsolve a solved problem by segregating the settings into two paradigms, neither of which is a superset of the other. Even the duplicated functionalities are not equivalent—the list of programs you can uninstall with the new UI is not the same as old UI.
No sane person would ever intend to engineer this outcome, but it’s what we get. Very strange how that can happen.
Microsoft really could use some product/UX zealot in the driver seat for a while.
They should also work on reducing the latency of their products in general... Everything feels like laggy trash now. Someone should do a study on stress levels between windows XP and windows 11 user groups for basic daily office tasks and then throw it into the boardroom.
Anyways, the ironic thing today is that that whole UI mess Windows is in was caused by something that doesn't even exist anymore. These two disjoint worlds that lead to these jarring moments where you're thrown from that modern flat-clearly-coming-from-a-mobile-design-paradigm settings window into another dialog coming straight from the 90s. The new one wouldn't exist without Microsoft's botched attempt of re-entering the phone market! Without WP, we could have simply gotten more thought-out iterations and improvements to the plain old Windows GUI, with a desktop-centric approach, but more modern design philosophies. But now we have a mobile-first UI that is being spent lots of effort on to make it usable on a desktop OS.
And sure, I'm over-dramatizing a bit here. We got more touchscreens, convertibles, yada-yada that still benefit from these changes. But it was approached in a really twisted, stupid way, and I still don't know anyone who prefers UWP apps. Especially my old folks are rather confused by this and would prefer Windows still looked like XP or 7. Younger people might not care so much but spend all of their time in a browser anyways if they don't use their phones. Everyone else is a professional, and professional software doesn't exist as UWP apps.
I hate to say this, but I think it would just result yet another UX camp in the system. The weight of legacy is heavy.
To clean up the UI mess, Microsoft would need to heavily deprecate supported APIs and conventions, and I don't think that's going to happen.
But that's certainly the approach they always try, yet it never works. Because that declared winner always turns out to be yet another superficial subset full of broken promises. And this declared but failing winner is always the latest incarnation.
What they have never tried: going back to the oldest and extending that with new additions. There are always far fewer new additions than the backlog of existing settings that would have to be ported to the latest variation, doing the reverse, properly, wouldn't be that hard.
Uh-huh... unless you don't know what the option is called. Ever tried using translated Windows? I spent hours upon hours wading through those dialogs gawking at the oddities that were supposed to be computer-themed terms in my native language. If you tell me I need to formulate those freaky things in my head, you're gonna get a laugh in your face.