One more anecdote. I found this screen for the first time when I got my laptop to demonstrate something to a student. I didn't know what to do, started to do random things, trying to figure out what happened, and the student interrupted my attempts, got mouse from my hand and swiped with it. I felt myself old and stupid. It is the thing like 20 years ago when I taught my parents to use standard UI. Only new it is me who needs help.

I asked, she didn't see Ubuntu before, and nevertheless she managed it better than me. I think I'm growing old, and just couldn't keep up with a pace of changing.

Because any hope of guessing it comes from knowing that phones do that, so the less like a phone you know it to be, the less likely you are to try that. Notice that even on phones it isn't intuitive if you've never done it before. If you tap the lock screen on Android it shows a message that says "Swipe up to unlock" -- because how else would you know that for the first time? But at least there it serves a purpose.

On the other hand, when my iPhone suddenly would connect with a caller, but neither party could hear the other, redialing didn't help, turning it off/on didn't work, I remembered the ancient trick of "cold boot". Which resolved the problem.

After some digging, I found a new trick that I guess is implemented at a lower level: press and release volume up, then volume volume down, then press and hold the main button until it powers off.

It is pretty confusing the first time, and annoying every time after that, didn't conciously know about the "Enter" trick before now.

1. What do you need to do to invoke a wakeup? Press a key? Are there any keys that don't wake the machine? Move the mouse? Click a mouse button?

2. Multiple monitors: During the wakeup sequence, you first have one display turn on, then you think you can log in but surprise! Another display turns on, and the main display briefly flickers off. For some reason, after 25+ years, display driver writers still can't figure out how to turn a second display on without blanking and horsing around with the first display.

3. Once the displays are on, some systems require some kind of extra input to actually get a login prompt to display. Is it a mouse click? A drag? Keyboard action? Who knows?

4. Some systems allow you to type your password as soon as the computer wakes. But there is some random delay between when you invoke wakeup and when it can accept input. What this usually means is I start typing my password, the computer only recognizes the last N characters and rejects it, I wait, then type it again.

These are some irritating bugs that affect everyone who uses a PC every time they log in. Yet OS vendors choose to spend their development time making more hamburger menus and adding Angry Birds to the Start menu.

So MS has managed to make an interface that works just the same on desktops, laptops and touch-enabled devices, and the UX isn't bad on either.

It's the shitty ergonomics that have been pervading software UI design for several decades now.

From the huge number of responses to this article, it's clear the software industry has a very major problem. The question I ask is why haven't users/user complaints been successful in stopping these irresponsible cowboys.

Seems nothing can stop them.

Most users don't complain, because technology is magic to them and they have no point of reference; they assume things must be the way they are for a reason. From the remaining group that does complain, many do it ineffectively (e.g. complaining to friends or in discussion boards that aren't frequented by relevant devs). And the rest just isn't listened to. It's easy to dismiss few voices as "not experts", especially today, when everyone puts telemetry in their software (see below), and doubly so when user's opinions are in opposition to the business model.

Finally, software market isn't all that competitive (especially on SaaS) end, so users are most often put in a "take it or leave it" situation, where there's no way to "vote with your wallet" because there's no option on the market that you could vote for.

The problem with telemetry is that measuring for correct things and interpreting the results is hard, and it's way too easy to use the data to justify whatever the vendor (in particular, their UX or marketing team) already thinks. A common example is a feature that's been hidden increasingly deeply in the app on each UI update, and finally removed on the grounds that "telemetry says people aren't using it". Another common example is "low use" of features that are critical for end-users, but used only every few sessions.

I also don't like things measuring "dwell time" when scrolling, as it encourages attention-grabbing gimmicks and rewards things that are confusing as well as things that are useful.

As in, if you are not a UX professional, your opinion is inconsequential.

See: replies on most Chrome UX feature requests over the last decade

Of course. The people complain. They say fork it and change it yourself or use what we give you.

The people that can't just suffer through it. The ones that know enough use something else.

I login in linux tty. And startx starts dwm. No fancy login screen for me.

This is why I don't touch GUIs from the major binary distros or gnome3 with a 10 foot pole. If I can avoid it I don't ever install anything from those projects.

[0] is the example that always comes to mind. I guess this made sense to somebody at the time, but it adds overhead to a process that was simple before, and isn't enabled just for "Enterprise" deployments, it's just dumped on the user to figure out how to configure screensaver hack settings by creating / modifying a theme.

[0] https://wiki.gnome.org/Attic/GnomeScreensaver/FrequentlyAske...

Instead of spending the substantial donations they received[1] on who knows what, the GNOME foundation should have spent some of it conducting proper focus groups.

1: https://www.gnome.org/news/2018/05/anonymous-donor-pledges-1...

I think it's an interesting and worthwhile experimental path; I just wish it wasn't the "default" as much as it is. But I also feel that way about Ubuntu. And Windows. xD

One of my least favourite was when it was not possible to configure the screensaver timeout to never turn off the display. IIRC you had a choice of several fixed times, from 5 minutes to 4 hours, but no "Never" option.

Not useful for systems which display information and are infrequently interacted with. That use case was completely ignored, and for no good reason.

oh no doubt, I have another comment from 4+ years ago about the same topic https://news.ycombinator.com/item?id=10883631 and even then it was ancient history IIRC

man, just looking at that page again reminded me that Windows Registry for Linux^W^W^W^W gconf exists.

>Icon Themes can change icon metaphors, leading to interfaces with icons that don’t express what the developer intended.

Icons were never sufficient metaphors to start with which is why we have text labels.

>Changing an app’s icon denies the developer the possibility to control their brand.

What does this even actually mean.

>User Help and Documentation are similarly useless if UI elements on your system are different from the ones described in the documentation.

This is only true if the user is completing an action that is solely based on clicking an icon with no text which we have already established is bad.

>The problem we’re facing is the expectation that apps can be arbitrarily restyled without manual work, which is and has always been an illusion.

Why has this worked generally fine in lots of ecosystems including gnome?

>If you like to tinker with your own system, that’s fine with us.

Earlier discussion seemed to suggest that lots of gnome developers were in fact not fine with this because it hurt gnomes "brand identity"

>Changing third-party apps without any QA is reckless, and would be unacceptable on any other platform.

Reckless?

> we urge you to find ways to do this without taking away our agency

Your agency?

> Just because our apps use GTK that does not mean we’re ok with them being changed from under us.

Nobody cares if you are OK with it.

Because it’s now possible to run multiple VMs at once (containers, etc) perhaps it’s time to run a simple, minimal, admin friendly hacker vm inside Ubuntu desktop?

Let Ubuntu configure all that it needs to get a good functional machine out of the box (working sleep mode for laptops, WiFi management, GPU support, systemd if that’s what it wants.) I then deploy the minimal VM I actually want to poke around with inside that installation.

This is pretty much what many people do in macOS. Apple’s OS supports the bare metal, vagrant / VirtualBox give me my tractably scrutable dev environment.

It’s not a particularly ground breaking concept but it might cheer me up a bit when battling with the volatility of user facing Linux distributions.

> Let Ubuntu configure all that it needs to get a good functional machine out of the box (working sleep mode for laptops, WiFi management, GPU support, systemd if that’s what it wants.) I then deploy the minimal VM I actually want to poke around with inside that installation.

If there's anyone like me here they might be happy to know that KDE Neon exists and is something like:

- Stable Ubuntu base.

- Infinitely (almost) customizable KDE on top.

- And (IMO unlike Kubuntu) sane defaults.

There is never going to be a unified GUI for Linux; that requires a dictator. KDE tried to provide the carrot of development-ease, Gnome tried to generate some reality distortion, but nobody cared. Carrots don't work. As far as I'm concerned, the experiment is over and it is time to embrace the chaos.

Now, this is easy for me to say, I'm mostly a command-line person anyway, and have spent most of my working life dealing with horrible UI. But it does have a lot of implications for Linux that I think a lot of people are not ready to accept.

You are spot on, and your 'book analogy' is perfect. If it works perfectly don't change it — that is unless an innovation arrives that offers a significant improvement and that's just as easy to use.

Unfortunately, most so-called UI improvements over the last 20 or so years are not improvements at all, in fact many have been quite regressive. They've annoyed millions of users who've collectively wasted millions of hours relearning what they already knew (and in the end nothing was added by way of new productivity)—and that doesn't include the developer's 'lost' time developing these so-called improvements. It's time that would otherwise have been much better spent fixing bugs, providing security improvements and or developing software for altogether new applications that we've not seen before.

The question I keep asking over and over again is what exactly are the causes behind all this useless 'over innovation'. Why is it done on such a huge scale and with such utter predictability?

Is it marketing's wish for something new? Are developers deliberately trying to find work for themselves or to keep their jobs or what?

It seems to me that many a PhD could be earned researching the psychological underpinnings of why so many are prepared to waste so much money and human effort continuing to develop software that basically adds nothing to improve or advance the human condition.

In fact, it's such an enormous problem that it should be at the core of Computer Science research.

Promotion & NIH management sydrome.

New shiny gets a promotion. Fixing a niche bug in a decades-old stable system does not.

And by the time all the new bugs you've introduced are found, you'll have a new job somewhere else.

So essentially, project managers' bosses not pushing back with a hard "Why should we change this?"

I’m probably missing some stuff, but I think people out to at least be able to “feel” their way around a UI. Lately there’s been so much push for minimalism like omitting scroll bars and such that make it confusing.

But, again that experimentation will root out what works and doesn’t. And new devices like VR of course have yet to be discovered paradigms.

> the stacking window managers work well from Windows 95 and XP why change it?

To get something that works better.

Despite all evidence to the contrary.

Well said. Is your machine shop stocked by a single brand of tools all in the same color, or is it a mix of bits and pieces accumulated, rebuilt, repainted, hacked, begged-borrowed-and-stolen over the course of your development as an engineer?

A free software Unix workstation is exactly the same. It’s supposed to look untidy. It’s a tool shed.

Apologies if I’ve touched a nerve with the Festool crowd with my analogy.

Agreed, but I can never get to the bottom of or reason why developers do not provide alternative UI interfaces (shells) so that the user can select what he/she wants. This would save the user much time relearning the new UI (not to mention a lot of unnecessary cursing and swearing).

For example, Microsoft substantially changes the UI with every new version of Windows—often seemingly without good reason or user wishes. This has been so annoying that in recent times we've seen Ivo Beltchev's remarkable program Classic Shell used by millions to overcome the problem of MS's novel UIs.

Classic Shell demonstrates that it's not that difficult to have multiple UIs which can be selected at the user's will or desire (in fact, given what it is, it has turned out to be one of the most reliable programs I've ever come across—I've never had it fault).

It seems to me that if developers feel that they have an absolute need to tinker or stuff around with the UI then they should also have at least one fallback position which ought to be the basic IBM CUA (Common User Access) standard as everyone already knows how to use it. If you can't remember what the CUA looks like then just think Windows 2000 (it's pretty close).

It's because everybody wants you to use their thing and not some other thing. If people have a choice then some people will choose something else.

This is especially true when the choice is to continue using the traditional interface everybody is already familiar with, because that's what most everybody wants in any case where the traditional interface is not literally on fire. Even in that case, what people generally want is for you to take the traditional interface, address the "on fire" parts and leave everything else the way it is.

Good change is good, but good change is hard, especially in stable systems that have already been optimized for years. Change for the sake of change is much more common, but then you have to force feed it to people to get anyone to use it because they rightfully don't want it.

Linux was all about chaos and herding cats until a short number of years ago.

It's the "standardisation at all costs" brigade who have killed the goose that laid the golden eggs. It's now far worse than Windows in many aspects. Freedesktop and GNOME deserve the lion's share of the blame, but RedHat, Debian and many others enabled them to achieve this.

Over the last decade, we have experienced a sharp loss of control and had certain entities become almost absolute dictators over how Linux systems are permitted to be run and used.

Linux started out quite clunky and unpolished. It could be made polished if you wanted that. But nothing was mandatory. Now that's changed. A modern mainstream Linux distribution gives you about the same control over your system that Windows provides. In some cases, even less. Given its roots in personal freedom, ultimate flexibility, and use as glue that could slot into all sorts of diverse uses, I find the current state of Linux to be an nauseating turn off.

And I say that as someone who has used Linux for 24 years, and used to be an absolute Linux fanatic.

Exactly, but what perplexes me is why aren't these issues that are so obvious to us not obvious to them. Why do they think so differently to normal users?

Animations that block use input are the sort of stupidity that becomes evil in its own regard. There was even a calculator but with this. This is a massive failure at the management level; somebody actually codes these things, but that it's not caught anywhere before shipping shows that the wrong people are in charge.

Don't make your user repeat something twice :)

I do not log-in frequently enough to remember how it behaves on log-in (even my laptop has ~90 days of uptime).

FWIW, moving away from GNOME 2.x to either Unity or GNOME 3 was a hard move to swallow, though in all honesty, Unity was better (though pretty buggy and laggy until the end)!

Now when it is gone, I'm using xfce, which seems be the last decent desktop environment.

Besides waiting for animation, in Windows 10, if you type the password fast enough the first character gets selected and the second character you type will replace it.

This happens frequently.

These things used to work reliably. I think most of the problems are caused by introducing asynchronicity into apps without thinking about how it affects keyboard input. Keyboard input should always behave like the app is single-threaded.

Here's an example I encounter whenever I use Microsoft Teams at work. I go to "Add Contact", and the entire screen becomes a modal entry box into which I have to enter a name. There's a single entry field on the screen. It's not in focus, even though that's the sole action that I can perform at this time. I have to explicitly select the entry field with the mouse and then type. It's such a basic usability failure, I really do wonder what both the application developers and the testers are actually doing here. This used to be a staple of good interface design for efficient and intuitive use.

Presumably Gnome copypasted it from Windows, because otherwise where did that idea come from into multiple distinct projects simultaneously? Windows has always had ctrl+alt+del to logon, Ubuntu hasn't had a precedent of having to do something to get to the logon prompt, IIRC.

Of course not.

This is the same this Win10. I was super annoyed they add one more step to a very frequent action, for no benefits on a PC computer. Hopefully I don’t have to use Win10 too much, but this is symptomatic of the mobilification of computers.