The Decline of Usability: Revisited (2023) | Hacker News Reader

The Decline of Usability: Revisited (2023)

datagubbe.se

97 pointsturrini80 comments

I recently cancelled my Spotify subscription because I just simply can't find my way around the app anymore. This is not an exaggeration—I really don't know anymore how to do basic stuff like getting to an album from a song (without going through a series of obscure right clicks).

I'm 38 and have used computers since I was 5.

Spotify just made the 'brilliant' decision to remove the "Add this song to the playlist" button. Now to add songs that Spotify itself recommends based on the playlist you're listening to you have to, right-click the name, select "Add to playlist", scroll to find the playlist you're currently listening to and select it. Before there was a single button to do this.

But at least I got more whitespace on my screen now...

But why would you want to do that!? All you need is the section with algorithmic suggestions. You're clearly using it wrong.

Spotify’s ultimate goal is to move you completely away from listening to recordings from actual artists, and instead listening to a stream of 100% AI-generated slop that they don’t need to pay any royalties on.

Letting you go from a song you like to its respective album (or really, doing any navigation other than “start/pause this algorithmic playlist”) is counterproductive to that goal and so needs to be disallowed, or at least made as difficult as it can be.

Really? Very hot take. Can you link one AI song that has moderate popularity that they might be shuffling people to?

This is not even software but somehow on/off switches are a challenge. When I was a kid I remember struggling to figure out how to use a power switch on an extension cord. You'd slide it one way and it revealed the word "on", and the other way revealed "off". But did that mean it was on when it said "on" or was "on" telling you to what moving the switch would do - which is how pause/play buttons on video players work today? 30 years later I got stuck with the same problem in my iPhone. Today they have green coloring to help but back then, I think it was more confusing. Computer monitors today are worse - you never know if it's blank because it's turned off (indicated by LED being on (or off) or a certain color) or no signal (indicated by LED being off (or on) or a different color).

Note how this hostile feature tends to appear in places where you turn off ad tracking.

Pause/Play still confuses me sometimes, so I think that not all apps do it the same. Is it showing the current state, or what will happen if I tap it? Just have two separate fucking buttons please.

Same as on/off controls, they should have clear external labels, e.g.:

    Off |X | On

So it's clear what moving the slider will do.

It’s even okay to have a single button, but don’t change its label and instead clearly indicate whether its active, for example by rendering it as depressed, assuming a consistent UI where the user knows how “depressed” looks like.

I’ve dealt with bug reports where users turned out to be completely unaware that they could scroll an area of UI, but yet we continued the war on scroll bars. I thought that users literally not being able to use the product trumped aesthetics but the Jonny Ive inspired minimalism over functionality won out.

My husband was trying to do something important on American Express's website. He was having trouble and called me over. Turns out some asshole designer thought that light grey text looked nice or some shit. Light grey text also makes it look like all the options are disabled and he was trying to figure out what he had to do to enable the form.

Infuriating.

There is a forum that even makes the main text content of a post light grey!

Yup.

"usability forensics"

aka point & click adventure games

> users turned out to be completely unaware that they could scroll an area of UI,

Some _visible_ and _usable_ scrollbars would really help. I really think that people who come with "new ideas" in UI design shall make their name public, so we lnow who to curse.

One other gripe that I have with modern UIs, which I didn't see mentioned (but which I might just have read past) is the scattering of controls in different places. Many are, of course, across the top in either a real menu or (yuck) a "ribbon", but others are down at the bottom, or along one side or the other of an app's window.

My recollection is that MsWord is particularly bad at this, but since I no longer have it installed (one reason is exactly this!), I can't show it.

But I do have Ms's Visual Studio Code on-screen. There is (thankfully) a real menu, with File, Edit, View and Help (the latter no longer exists on most Microsoft products). I happen to have a terminal open; it has six mostly indecipherable icons across the top of its pane. All the panes--the terminal, the file edit panes, and the "bar" at the right-hand side, have a '...', which seems to be the equivalent of a hamburger menu for that pane. Finally, the status bar down at the bottom of the window has still more indecipherable icons near the left end, and a few info things near the right end, some of which are controls ("Select Interpreter", inexplicably highlighted in brown with yet another icon), and some of which appear to be just info (line and column)--except these at the bottom of the window turn out to pull down a special menu item at the top of the window. For example, the control labeled "Ln and Col" (the latter means the character within the line, not the column in a tab-delimited file) pulls down a menu item that allows you to go to a particular line (but not a particular "column").

Firefox is a big offender here. Every single interactable UI element is at the top of the window, until you press Ctrl+F and for whatever reason some asshole at Mozilla decided that the Find on Page UI should go to the bottom of the window. Absolutely cooked.

I don't see a problem with that. If you are using the find in page feature then you are very likely going to type something. Your hands are already on the keyboard. The fact that you said "press Ctrl+F" instead of clicking the menus is evidence for this. All the options are best accessed with keyboard shortcuts, so Fitt's law is not relevant.

However, I just noticed one big UI flaw in this interface. The keyboard shortcuts for finding the next and previous occurrences of the search phrase (enter and shift-enter respectively) are not easily discoverable. They ought to be mentioned in the tooltips for those buttons.

EDIT: And another problem: the next and previous buttons aren't even correctly marked as buttons. It's worse than the "flat" buttons used elsewhere, it's "stealth flat" buttons that only appear when you mouse over them.

they just copy chrome for the past decade tho.

The whole point of the Office ribbon was to consolidate controls, because they were at the point where most of their feature requests were for things that already existed but people were completely unaware of on account of every menu and toolbar having 80 million items.

So instead they moved every feature into 80 million ribbon items, most which are not visible.

At least the ribbon makes the most common features one or two clicks. I have nostalgia for the old menus, and wish they were at least an option. But they often required a lot of dexterity and persistence to discover and use when options expanded beyond a depth of two

it was already hidden.

the ribbon didn't fix it. the ribbon just replaced an invisible right mouse button click on the general toolbar region with an unavoidable huge eyesore.

Part of the problem they had just before the switch to ribbons (i.e. with Office 2003) is how they were semi-hiding not-commonly-used menu items by default, so that you had to click an arrow-button on the bottom of the menu to get the full set if items.

I have to say that from my experience, ribbons make it harder for people to remember where buttons/functionality is located, and harder to remember that it even exists, relative to full menus + toolbars. And yet - newbies used to MS Office keep clamoring for ribbons, ribbons, ribbons (e.g. in LibreOffice).

There is now a toggle to switch LibreOffice to a ribbon UI: https://itsfoss.com/libreoffice-ribbon-interface/

In LibreOffice, on the menus, you can choose View > User Interface... and choose one of several UI modes, including the classic menu+toolbar and a ribbon-like "tabbed" UI. That's been available for a few years already.

For the upcoming 25.8 release of LibreOffice, we have decided to make the UI mode selection part of a welcome dialog on the first run - as a compromise between the majority opinion, which is that the MSO ribbon is undesirable, and the desire to make the transition easier for MSO users who expect a ribbon-like UI since that's what they're used it. The default choice will remain menus+toolbars.

Also, the LO tabbed UI is not "fully-baked", and still has quite a few issues :

https://bugs.documentfoundation.org/show_bug.cgi?id=107237

and more generally:

https://bugs.documentfoundation.org/show_bug.cgi?id=102062

And Microsoft said it would be too hard to maintain both the Ribbon and the traditional menus. Somehow LibreOffice does what Microsoft said was too hard.

"The whole point of the Office ribbon was to consolidate controls..." And now the controls have become un-consolidated, having been scattered to all margins of the Office window.

Thinks you can click on should look like things you can click on. This is why the hamburger menu is bad. Even people that have been in tech for decades often don't realize that's the menu they're supposed to click on. This is why it was bad all those years ago when "UX" suddenly became a thing and designers inserted their craft into everything as if they were in marketing or something.

There is a happy medium between something written by a ham radio operator in 1997 and what counts as "good ui design" today.

Though you probably shouldn't listen to me. I still prefer the look of the Windows 3.1 GUI.

I think it took me a good year or two of seeing hamburger menus to realize that it was a menu icon. Totally shit in terms of intuitiveness. And of course no text to suggest that it's a menu.

At least make it a gear or something to imply "psst you can make changes here".

Three lines? Three lines? That's what we settled on? Okay. Yeah okay.

I kind of assumed that the three lines were to represent a menu with three choices, but it's not at all obvious

> I still prefer the look of the Windows 3.1 GUI.

Although not so good as Win2k UI, it still looks thousand times better than Windows 10 or 11 GUI. I really think that, today's UI developers, leave their brain at the entrance.

I will truly never understand the mind of people who make decisions like "scroll bars should hide". They are like aliens to me, simply impossible to relate to.

I’m the same, except the opposite - I don’t understand people who want it there permanently, instead of gracefully hiding when not needed.

I think the lesson here is people have different desires and priorities, and that’s ok.

I don't have a strong preference either way but the argument is the scrollbar isn't just for moving your position on the page but also for communicating to the user where they are on the page. If you hide it you're removing half it's functionality.

Traditional scrollbars always went away when they weren't needed. The problem now is going away when they are.

I don't think this is correct.

I distinctly remember some scroll-enabled UI elements on Windows displaying a completely greyed out scrollbar when the content wasn't actually scrollable.

Indeed, I can verify this today with PUTTYGEN.EXE even on Windows 11, where if I generate an RSA-2048 key there is an active scrollbar, while if I generate an Ed25519 key which is 1/8 the size, there is a greyed out scrollbar.

That having been said, I think this is the right way to do things. The greyed out scrollbar tells me that control is meant to be scrollable. So if I do something that changes the content, I can expect to be able to scroll it.

I originally had some text in there saying as much. But this always on capability has to be enabled by the author explicitly. It’s rare enough that I felt keeping my answer simple was more important.

“When not needed” is subjective. It’s okay to have an option to hide them when you prefer that, but we are now at a point where websites and apps are adding HTML-based position indicators and “scroll down to see more” labels because scrollbars have lost the ability to serve that function for most users.

I want them always because, if they are hidden, it's difficult to get them back when needed.

You need to get the cursor in the right zone (which is invisible, and quite narrow) and keep it there for long enough (so you have to slow down in order to not overshoot).

This is a pretty annoying thing to have to do many times every day.

Sometimes, for narrower scrollbars, it might take a few attempts until I succeed. It's distracting.

Crazy how often it backfires, too. I wasted 15 minutes a few weeks ago checking a user report that a PDF I hosted was missing exactly half its pages. Eventually, after walking it through my workflow step by step and it not breaking, I realized that it was simply the PDF viewer hiding the horizontal scrollbar which told you that it was a two-page layout. We had both fallen for it. Thanks, GNOME et al! (╯°□°)╯︵ ┻━┻

(This also illustrates the fallacy with the sibling comment blandly asserting that scrollbars should 'simply be hidden when not necessary'.)

But thanks to the scrollbars being hidden, you got about 20 pixels more of your pdf! /s

Invisible scrollbars is probably the number one reason why I scream at the computer when I have to use Chrome instead of Firefox. At some point, Chrome removed the option to make the scrollbars visible. The UX people are completely insane.

Case in point: I tried to install ChromeOS Flex on one of my laptops. After booting from the USB drive, the installer went through a series of screens. On the 3rd or 4th screen, it would hang and make no progress. I rebooted and re-installed. Same thing. Tried a third time. Same thing.

On the 4th try, I accidentally discovered that the dialog box had an invisible scrollbar. WTF. If I two-finger scrolled on the dialog box after moving the mouse pointer into it, it would reveal some additional text on the bottom which indicated that it was not hanging but doing some work.

After I had finished installing ChromeOS, I discovered ChromeOS has a Settings option to "always display scrollbar", but the Chrome browser completely ignores that flag. Awesome. I blew away ChromeOS Flex on my laptop.

I desperately want real, thick scrollbars again. It is becoming close to impossible to use the mouse with browsers and websites.

I do understand why designers don’t wish to clutter their pretty UIs with navigation elements, but surely the need of users to use the site should outweigh the desire of the designers!

Gawd, yes. Modern UIs are far less discoverable, and less usable, than their predecessors.

It's less about the look than the state. The article shows the complexity of Blender menus. But that's not the big problem. For an object to be visible, four things must be turned on. It's hard to figure out which setting is set wrong. Especially when a new release adds something new that has to be turned on.

This is a generic problem in Blender. You can't do something because you're in the wrong state, but nothing tells you that you're in the wrong state.

GNU GIMP has always been much worse than Photoshop in this way. When inserting text, the text is in a default size. If you change the size, and then click on a new insert point, it goes back to the default size. Whether or not the last change has been committed is not obvious, and until it has been committed, many menus and icons do nothing, silently. Selection vs. layers vs. commits are very confusing. Just keeping the dockable menus visible is tough.

IMO, Human Interfaces Guidelines should be primarily thought of as a foundational layers, like a standard library for common patterns. But the true driver of UX should always be the domain, be it word processing, 3d modelling, or graphic editing. Instead, we have company branding and oversimplification.

I've moved to Emacs, TUI, and CLI tools to escape the madness, not because they'r e better, but they let you do stuff and are at least stable so you don't have to alter your workflow every quarter.

You speak of the need for domain-driven UX but you have abandoned domain-specific tools?

Most of my needs are related to computer operations, so I just craft small scripts. But I would happily use a professional tool if needs be. Like Intellij or Affinity Designer. The issue is more about smaller utilities and office tools.

I think they make valid points (they really do describe the problems with many modern UI that I see, and that many older ones are much better), but they did miss a few things, such as:

- Having keyboard commands is helpful.

- Good documentation is very helpful; a program is understandable if it is documented.

- Some of the difficulty seems to be due to the programming environments and libraries that are used for making these programs; due to badly designed UI libraries and programming environments, the result will also be bad. However, this is not the only thing that can cause these problems.

- On a computer it should also be helpful that the operator is able to make other external programs and can interact with them too, with your software. Command-line programs, user configuration settings, API, etc, can also be helpful in doing this.

> Having keyboard commands is helpful.

Not just having keyboard commands, but using standard ways to make them discoverable, such as by tooltips, menu item annotations, and underlined characters in labels.

Another aspect in today’s UIs is that they often introduce latency in operations (due to network communication, among other things) while not buffering keystrokes accordingly, which makes it borderline impossible to press memorized sequences of keyboard shortcuts in quick succession, because you always have to double-check that the application is in the right state to receive the next keyboard shortcut. That goes against developing muscle memory for frequently performed operations, and forces a conscious back and forth and constant ascertaining that the command was correctly received by the application, instead of being able to blindly trust it and thereby reduce cognitive overhead.

> which makes it borderline impossible to press memorized sequences of keyboard shortcuts in quick succession

My favourite is the Windows password dialogue. If you start typing before it appears, your keystrokes are lost.

Even worse, in my opinion, is when you're typing away and a dialog from some other application suddenly decides to pop up and swallow all your key strokes; or, in the worst case, one of your key strokes just happens to coincide with a shortcut key on the intruding dialog so you accidentally accept something without having a clue what the dialog was and what you accidentally accepted.

OS should prevent any dialog from a background application from coming to the foreground of it detected that use is in the process of typing. It's not that difficult of a task after all.

[Edit: auto-incorrected text]

> Having keyboard commands is helpful.

Having some place where they are described is ... priceless. MS keyboard commands: press ALT (bangs head)

FTA: Another widespread source of influence was IBM Common User Access from 1987, which among other things introduced […] the ellipsis ("...") to indicate menu choices that opened a dialog window.

I think that the Mac, possibly even the Lisa had that before the CUA. https://andymatuschak.org/files/papers/Apple%20Human%20Inter..., page 23: “A dialog box appears whenever the user chooses a menu item that is followed, in the menus itself, by an ellipsis (…)”

CUA definitely built on and incorporated Mac innovations, some which were built on work at Xerox. Motif/CDE also converged on the same conventions. And it was glorious for twenty years or so.

So...

* Do you have a link to the IBM CUA document?

* Can you trace a sequence of widely-adopted or highly-regarded documents of this vein, after 1987? And perhaps how they relate?

There used to be a digital document with the CUA from IBM, but it was taken down.

Irreducible complexity is a good term. I think that's the problem with modern IDEs. Writing code is a complex task that can take infinitely many different forms. Sometimes all I need is just notepad or nano. Other days I need to look at four files, debug output, and an SSH session at the same time.

In order to solve this irreducible complexity, all modern IDEs are simply trending toward the simplest, """cleanest""", bare minimum of functionality.

It might look great to your UX design artists that want that promotion you promised, but where the rubber meets the professional with thousands of hours of experience, it falls apart. By removing complexity from your tools, you remove the ability to do complex tasks.

When my work involves processing huge amounts of information as fast as I can and my tools doggedly remove as much information as possible from my screen... Do these 'designers' actually know any programmers? Does anyone involved here actually write complex programs where you actually need the hidden 90% of functionality?

I've given up at this point. I canceled my JetBrains subscription and I'll be using the 2024.1 versions of all tools until someone comes out with a more modern IDE that's designed to be used by professionals and not children or c-suite types.

"Because some VP in a glass office somewhere got the brilliant idea that shopping for a $60,000 car should feel like buying toothpaste online." https://www.linkedin.com/posts/markobijelic_if-clients-thoug...

> What is the "Archive" icon even supposed to depict? The lower part of a printer, with a sheet of paper sticking out?

it's a bankers box (a cardboard box for files) which I guess is not a thing one sees often now? skeuomorphism doesn't work on a digital generation because all the real world touchpoints for information have been replaced by digital

But then they put some lines on the side for no reason and now it looks like a printed sheet.

That's the labeling put on bankers boxes so you know what's in them when they are stacked on shelves.

Those boxes seem to be mostly a US thing, and the author is Swedish (or at least the domain is).

I have hated the removal of text from icons for ages, and it seems like a problem that is going to become more relevant when you realize that zoomers have no idea what the floppy disk icon to save something is.

On a plus not I noticed that the Movistar TV app has a wonderful keyboard (in the context of using with a TV remote). Instead of a querty keyboard, which was designed to make common letter combinations far apart to avoid typewriters jamming, it has a horizontal a-z alphabet. It is so much more efficient to use, just left or right buttons, no farting about trying to locate the key and having to move up down left and right.It's both simpler and more efficient.

Nowadays menus are called "+" or "...".

(Enough said.)

Hamburgers are like a box of chocolates; you never know what you're gonna get. Because finding the features of your product should be an adventure!

First of all the number one thing getting in the way of mainstream usability is growth hacking.

But to the authors point about power tools usability likely the core issue is new customers need an easier entry point into a power tool than existing customers. This results in simplified designs.

If they don’t do that cruft design builds up over time and you end up in a different usability trap like adobe products in the mid 2000s

Aren't Adobe products even worse today? Although I confess I've only ever used Adobe Acrobat, but I recall that the controls became exceedingly hard to figure out, and took up excessive screen space, somewhere around 2010. I now use PDF-XChange Editor, which has a somewhat better UI, to my mind.

> What is the "Archive" icon even supposed to depict?

A labeled archive box. https://kagi.com/images?q=archive+box

This raises the question of the universality of icons. They are contextual to territory and time. Until recently, a diskette was the universal icon for saving. Yet nobody under thirty today has probably ever seen one.

Yeah, I think “what does this icon depict” is a little beside the point, also considering the mIRC screenshot. BUT, what is important is for icons to be easily re-cognizable by shape AND color, and modern low-contrast monochrome icons fail that, being designed to look as uniform as possible. In the Outlook screenshot, the fact that the labels have the same color as the icons also doesn’t help. Furthermore, the label text size there emphasizes the larger-sized icons, although the text is more informative.

I recently wrote about gnome:

https://news.ycombinator.com/item?id=44043157

TL;DR: I use gnome only indirectly. In the past I've cut them slack, because (I thought) they were designing for touchscreens etc. Turns out they indeed destroyed the usability of their desktop but somehow forgot to make it work for touchscreens.

I have a starlite linux tablet where no gnome video player works acceptably, and I've tried around five. More than one looks like it is made for touch, big round corners/buttons, etc, but requires a keyboard for all but trivial functions... like rewind 15 seconds. WTF, can't tap?

Freetube, Netflix, and Kanopy work fine from a browser, so the problem sure as hell ain't the hardware. Defeat snatched from the jaws of victory.

Welcome to the brave new world of “I built the X software/app/website in 5 minutes and 10 tokens using AI!”

Kind of interesting about ranting UI and still getting the margins grouping wrong on his h2 headers.

(They grouped against the first paragraph of his section instead of the fill the section)

The author is talking about usability. Does this impact usability in any meaningful way? I would never have even noticed this and even after you mentioned it I am not completely sure what you're talking about and whether it's a problem.

Elements close to each other are perceived as related. So the h2 should be closer to the content below. Proximity is one of the Gestalt Principles and nothing novel. An article that goes do such lengths to shred current UI (I do find myself nodding along) should get basic typography right.

Now I definitely don't know what you're talking about. There's only one h2 on the entire page and it's above all of the content. I refuse to believe this is causing you a usability problem.

The Slack thing is a feature. Each org can have their own theme. Don't understand why it's desirable that everything has the OS theme. Some people spend 8 hours per day in Slack. Having it look a bit unfamiliar the first time you boot it up is fine, IMO.

Also IRC looks and works like ass. As a teenager I used it without putting in the time to getting used to it, and it was really really impenetrable.

The author isn't talking about IRC the protocol and probably wouldn't defend the more impenetrable parts of IRC interaction. He is talking solely about the mIRC user interface with its familiar File, Edit, View, Help menus, colourful icons, beveled buttons, visible scrollbars, draggable windows, high-contrast focus indicators, etc.

The theming is okay as long as it’s user-controllable, so that every user can select their preferred contrast and color scheme. Personally I would prefer all “orgs” to use the same OS-native theme, and at most a different logo somewhere.

The problem is when we are getting inconsistent themes with wildly varying contrast levels shoved down our throats, in what is supposed to be productivity applications.

Reply on news.ycombinator.com