Hacker News Reader

Comment by graypegg | Hacker News Reader

A designer I worked with years ago had a great explanation for why consistency is important.

It’s not about a limited colour palette or a careful selection of fonts no one will ever notice. Chasing the specifics makes horrible software. Some people equate less diversity in their UI to more consistency.

It’s about letting someone become an expert in your software.

Microsoft office was always his example. People pride themselves on “knowing” Microsoft office. They jump between all of the apps because they have a feel for the inertia of “office”. I have the exact same feeling in Vim. I just know how things are expected to be, and new (well made) plugins tend to respect those patterns.

Office and (Neo)Vim aren’t exceptional examples of UI, but they are uniquely stable.

View on HN

As a non-"Designer" something I never shut up about is how powerful it is to bootstrap that expertness by leveraging the UI elements and concepts your users already know from elsewhere, for instance, literal native UI widgets, and more broadly, highly-recognizable simple fundamental widget types. When you have the opportunity to present 4 options, you could use your brand's themed version of [Insert UI Library]'s searchable combobox with checkable items, or you could show 4 normal labeled radio buttons or checkboxes. The latter beats the former in every way:

1. User knows whether they can pick one vs many based on the circles vs squares.

2. User can see all four options, even if the widget ends up unfortunately placed in the viewport (I've had to scroll inside these and only be able to see like 2 options at once, so many times).

3. No issues with mobile keyboard wanting to open to let you "search", further obstructing the tiny viewport.

4. Accessibility will always be 100%.

Yet this option "fails" in the category of "looks cute with our Design System" so usually designers choose the first choice, and worse, "standardize" on doing that, so that using normal widgets is a "bug."

Ultimately too many "UX designers" are hacks, who worship and pursue aesthetics and branding at the expense of everything else. A UI widget isn't the time or place to assert your brand or creativity, any more than you should design your own font that forces all letters to be shapes from your logo. Your customers don't care about your brand more than being able to complete their tasks.

I had an interesting discussion on Twitter with Michael Darius about this very topic recently. He is an "Apple pioneer, skeuomorph & protégé of Steve Jobs" who epitomizes the excesses of aesthetics and branding at the expense of everything else, so I had to call him on his brash claim and ask him some pointed questions, and brought up the User Interface Hall of Shame's review of the QuickTime 4.0 player, as an example of the excesses of skeuomorphism.

https://twitter.com/darius/status/1741188955985604867

Michael Darius @darius 9:06 PM · Dec 30, 2023 22.1K Views:

There is just no reason why an interface should be anything but beautiful.

Don Hopkins @xardox:

Tell that to an airplane pilot who takes the lives of their passengers in their hands every day. Are you really saying that airplane cockpits should be beautiful, at the expense of safe and usable and accessible?

Michael Darius:

Beauty for the sake of beauty is a mistake but beauty for the sake of user enjoyment reduces cognitive dissonance. Don't forget that pilots are people too with emotional needs.

Don Hopkins:

Beauty should take a back seat when it comes to safety, usability, and accessibility, not only in airline cockpits, but in most other interfaces, too. Your statement that "there is just no reason an interface should be anything but beautiful" is extreme and dangerous.

Michael Darius:

An interface is a habitat for a persons soul to dwell. Some habitats are better than others and there some awfully designed habitats out there.

Don Hopkins:

There are many reasons why, and many things more important than beauty. A spreadsheet is more beautiful if it hides the ugly truth about the formulas, relationships, numbers, dependencies, and complexities. Its purpose isn't to be easy on the eyes, it's to be correct.

Are your earpods supposed to be beautiful, at the expense of being comfortable and actually fitting in your ears and not itching and poking and flaking paint and decorations into the holes on your head? You don't need to see them, just hear them.

The shape of your ear canals isn't beautiful, but your earpods have to conform to it nonetheless.

Michael Darius:

I'd sooner fly a luxury airliner with dashboard complexity that has been thought through for me than assume that safety was the tradeoff for getting to fly with simpler controls.

[photos of extremely complex airliner cockpits -- not sure which he's claiming are beautiful or ugly though]

https://twitter.com/darius/status/1741193514032267758

Don Hopkins:

Beauty is in the eye of the beholder, but safety and usability and complexity is scientifically measurable, you can't claim that thinking through safety and usability and complexity always results in beauty. Most people would consider those examples cluttered, complex, and ugly.

ESPECIALLY anyone who subscribes to the school of minimalistic design like iPads exhibit, like hiding scrollbars because they are ugly, even though they tell you how much more there is to scroll, and what to do to scroll to it. Airplanes and spreadsheets need ugly affordances.

Michael Darius:

You can tell when something has been thought through on the inside, just as much as it has on the outside and most of the people who disagree with me in practice have simply never had the opportunity to work with engineers who care more about what is under the hood than what can be seen by the naked eye.

I see no difference between beauty and the reduction in the margin for human error.

[Photo of two children piloting an airplane.]

https://twitter.com/darius/status/1741195625767772529

Don Hopkins:

You just have an extreme personal opinion about beauty if you think removing scrollbars and buttons and even cluttered features from iPad interfaces, and making airplane cockpits look like playschool toys, always equals beauty. Your original statement is extreme therefore wrong.

It's wrong to assume that "There is just no reason why an interface should be anything but beautiful." I'd rather be lazy and safe than wrong and dead because some interface designer though he was smarter than the FAA.

And you're also wrong because you're presuming there is only one definition of beauty: yours. And that it can be measured. It can't. Address that, please.

Michael Darius:

When effort has been put into an interface to simplify a function, the function itself becomes more beautiful. We see this when writing code. If I can perform an operation by writing 3 lines of code instead of 10, the margin for error decreases.

[An image with an example of "Ugly vs Beautiful" code claiming that a one-liner is more beautiful than 6 lines of equivalent code.]

Don Hopkins:

Again, you're pretending that beauty is not subjective, and that you know the one universal definition of beauty, and I would have expected better from you, and even more from a first year art student. Would the Mona Lisa be more beautiful with less paint?

Would the Sistine Chapel be more beautiful with stick figures like an XKCD comic? Get off your high horse of thinking you can define beauty. Especially after championing the overly minimalistic iPad user interface bereft of scrollbars, buttons with text labels, and even features.

Michael Darius:

Here's a quote that sums up Shaker philosophy:

“Don’t make something unless it is both necessary and useful; but if it is both necessary and useful, don’t hesitate to make it beautiful.”

[A photo of a "beautiful" chair.]

Don Hopkins:

But for the n'th time, you've again ignored me when I pointed out you're moving the goalposts from your original outrageously extreme and incorrect statement that "There is just NO REASON why an interface should be ANYTHING BUT beautiful."

Michael Darius:

I'm not ignoring you, I can't keep up with how much you are writing.

Don Hopkins:

Well at least go back and read the words you wrote in your initial tweet and try defending THOSE, instead of trying to move the goalposts and express the one true definition of beauty. I ask you again: define and defend the words you first chose and wrote yourself.

Your original statement is indefensible, so go back and try defending it instead of trying to move the goalposts and pretend you have a monopoly on the definition of beauty.

Your original statement didn't say anything about necessity or usefulness, as I said before. In fact it PRECLUDED them by claiming interfaces should ONLY be beautiful: "There is just NO REASON why an interface should be ANYTHING BUT beautiful." Stop deflecting. Defend THAT.

What exactly do you mean by "NO REASON" and "ANYTHING BUT". Usability and safety are reasons, and something but.

Some random guy:

Sorry to interrupt you guys. How about you two actually go to FAA or some shit and consult with them if it's possible to make a cockpit more beautiful but retain its practicality. I feel like the argument wouldn't go anywhere without trying to do something for real.

Don Hopkins:

You suggest us just walking in the front door of FAA headquarters, waltzing through security, and insisting on a meeting with the heads of the FAA to redesign their interfaces? Should Darius and I drop all we're doing and volunteer our time to do that for free?

Michael Darius:

The images themself do a better job of communicating what I mean by "beauty standards for excellence".

I've never met a designer who didn't feel that their responsibility was to defend beauty standards for excellence and I'm not different.

The mechanics of a jail door might be functional and usable by a door guard but there is nothing beautiful about building more prisons.

Habitats for humanity cares more deeply about the ACTUAL lived experiences people are ACTUALLY having.

[... A side discussion about the definition of beauty, in which Darius channels Deepak Chopra, and I try to get him back on track to defend his own original words, and he stands by his statement but doesn't justify it: https://twitter.com/xardox/status/1741212753056944187 ]

Michael Darius:

If beauty were so subjective then walking through a meadow would feel no different than doing jail time.

Don Hopkins:

Now you're sounding like Deepak Chopra and his quantum faith healing mumbo jumbo hand waving woo woo. I would have expected much more from you. You made a clear unambiguous brash statement using the words NO REASON and ANYTHING BUT. Defend those words.

Michael Darius:

We can agree to disagree.

[...]

The subjectiveness here may not be around the definition of beauty and more around the ‘definition of an interface’. An interface can be defined as ‘the surface between Air and water’.

Don Hopkins:

More woo-woo, Deepak. Address points 1 and 2 which are directly about YOUR WORDS: "There is just NO REASON why an interface should be ANYTHING BUT beautiful." (How many times do I have to quote them back to you before you address them? I must be on #10 by now.)

[...]

Don Hopkins:

Of all those cockpits, which fit your definition of beauty and which don't, and aren't you aware that there are many hard won FAA regulations learned from the study of accidents, mass death, destruction, and human error, that vastly trump your personal opinion of beauty?

Or are you arguing that all iPad interfaces should look like 747 cockpits?

Or do you believe that beauty automatically implies safety, ergonomics, usability, visual unambiguity and perceptibility? Then how do you measure and prove that beauty is not subjective and not in the eye of the beholder and not just your personal opinion?

And what if your personal opinions about beauty happen to require more weight, more expensive, less durable, more bespoke and less modular materials? Do economics not come into the picture, or do you only design interfaces for extremely rich people?

Of course Donald Trump thinks that solid gold toilets in his luxury airliner are beautiful. Do you agree, because you think beauty isn't in the eye of the beholder and he's right, or because you want to make the customer happy no matter how bad their taste and idea of beauty is?

Michael Darius:

In a field where cognitive safety needs to be the priority and a world where user enjoyment reduces cognitive load, it is intellectually lazy to assume that beauty is the tradeoff for ergonomic safety.

Don Hopkins:

The UI Hall of Shame's review of the notorious QuickTime 4.0 player proves the perilous consequences of a UI designer's personal opinions about skeuomorphic beauty trumping usability and ease of use. Were you involved with that, or was it before your time?

http://hallofshame.gp.co.at/qtime.htm

That article should be required reading for aspiring UI designers. Hacker News discussion:

https://news.ycombinator.com/item?id=18212478

My comment in that discussion:

"Apple's long romance with skeuomorphism peaked with Bob Bishop's APPLE-VISION, and went downhill from there." I posted several other comments about VLC's terrible UI, too.

https://news.ycombinator.com/item?id=18219757

Bob Bishop's Apple Vision:

https://www.youtube.com/watch?v=CU_qKQL5PVk

[At this point I took it to private DMs.]

Don Hopkins:

Are you familiar with that review and the problems that caused the widespread negative response to the QuickTime 4 Player user interface? Have you taken the time to read the whole article? Were you involved with that product, or was it before your time?

http://hallofshame.gp.co.at/qtime.htm

I have written about it on Hacker News, and compared it to the even worse WinAmp user interface. Other commenters compared them to WinAmp, but I don't believe WinAmp's infinite and easy skinnability is nearly as maliciously terrible or professionally irresponsible as the QuickTime 4 player (I expected SO MUCH MORE from a company like Apple that published Tog's original UI guidelines), or VLC (which is an open source project without any money or professional UI designers).

HN discussion:

https://news.ycombinator.com/item?id=18212478

One of my comments (among several), which includes a thread about VLC:

https://news.ycombinator.com/item?id=18219757

You can see that I actually give a shit enough about VLC to have spend quite a bit of time analyzing the problem and writing up bug reports, but the culture of the project is so insular and "NIH" that they don't give a shit about user interface design. So I have little hope that VLC will ever improve. But at least Apple finally improved their QuickTime 4 Player in response to acceoss-the-board criticism.

A counter example to the rule of thumb that open source user interfaces are terrible is Blender. Especially when you compare it to GIMP (which is like shooting fish in a barrel, but still is an instructive comparison of cultures).

Blender was infamous for its complex non-standard confusing user interface design, but they LISTENED TO THEIR USERS and vastly improved it!

But of course there is no way of getting around the fact that it's an extremely complex tool that's used differently by a wide range of people, so a lot of that complexity is necessary, and you can't just simplify it away and dumb it down without destroying its usefulness.

One thing Blender does have excellent support for is pie menus! And there are some great pie menu editors, which are important because every different Blender user has their own workflow and set of common commands. So Blender users need to customize and define their own pie menus. Just like HyperCard enabled normal users to construct their own task oriented user interfaces.

(Of course Steve Jobs hated pie menus, which is a story I can tell you about later, or maybe you can tell me the Apple side of that story: Why didn't Apple ever adopt pie menus for MacOS or iOS?)

Pie menus have the potential of being both beautiful, efficient, reliable (low error rates), and easy to use, but it takes a lot more design and programming to pull off than a traditional linear menu.

Simon Schneegans is a brilliant user interface designer as well as an accomplished programmer, and he has developed not only beautiful pie menus (for example the Coral and Trace menu he did for his Master's thesis years ago, and he more recent work on Gnome-Pie, Fly-Pie, and the cross platform Kandu pie menus for desktop app launching, window management, text selection, and many other tasks).

The Trace-Menu A short demonstration of a Pie-Menu I developed for my Bachelor thesis.

https://vimeo.com/51073078

The Coral-Menu A short demonstration of a Pie-Menu I developed for my Bachelor thesis.

https://vimeo.com/51072812

The thing about those examples (and his later work) is that the cool graphics don't spoil the underlying good Fitts' Law friendly design -- they actually increase usability and reinforce the gestural navigation user interface metaphor.

His most amazing accomplishment is the WYSIWYG drag-and-drop interactive pie menu editor he designed and implemented for Gnome-Pie, that he's reimplementing across platforms (Linux X11 Gnome, KDE, Wayland Gnome, etc, Windows, and Mac).

Fly-Pie 7: GNOME Shell 40+ and a new WYSIWYG Menu Editor!

https://www.youtube.com/watch?v=sRT3O9-H5Xs

>Fly-Pie is an attractive marking menu for GNOME Shell. Fly-Pie 7 supports GNOME Shell 40+ and brings a completely rewritten menu editor.

I hope this proves to you how important I believe beauty is. but how I also believe it doesn't need to sacrifice usability.

Here's the latest demo of his current project, Kandu: cross platform super customizable pie menus for Linux, Windows, and Mac. I just sent him my old Mac laptop so he can support the Mac desktop well. The way it works is by using Electron with a full screen transparent overlay window, so it can draw the pie menus with html/css/svg/canvas or any other standards based web technologies.

Kando becomes useful for the first time!

https://www.youtube.com/watch?v=7vVdJ9LORAM

>With Kando, I am creating an open-source pie menu which runs on Windows, Linux and (eventually) macOS. In the future, you will be able to create your own men...

Michael Darius (excerpts):

[...] Tim Wasko was the designer for the initial QT player and still remains one of the best interaction designers I know." [...]

I don’t consider something unusable or barely usable to be beautiful so if the work hasn’t been put in on the usability side or if something isn’t working it doesn’t matter how beautiful the interface elements themselves are. But in that case what makes it not beautiful is not that the interface elements aren’t beautiful. What makes it not beautiful is that what is under the hood isn’t beautiful and what is under the hood is more often than not not thought through the way it should be.

My definition of ‘beauty’ is inseparable from my definition of functional design:

An interface should go beyond usable, it should delight, create enjoyment, provide a sense of safety and control over one’s environment, it should be beautifully engineered inside and out and when any one of those pieces go missing beauty itself goes missing.

"you could use your brand's themed version of [Insert UI Library]'s searchable combobox with checkable items, or you could show 4 normal labeled radio buttons or checkboxes. "

Ahhh ... that's likely a Stilbruch (break of style) my design professor would have exclaimed. Designers hate those and I came to understand why and try to avoid it wherever possible. It is something that can be irritating and bringing in confusion, if suddenly there is a element out of place.

But you can have "highly-recognizable simple fundamental widget types" without breaking the style. It is just harder and of course, a functional ugly design is in my opinion still way better than a good looking broken design. But the goal should be something consistent - in terms of functionality and style.

It's possibly one of the reasons why so many people were bent out of shape when MS introduced the adaptive ribbon shenanigans that tried to 'help' by only showing the most likely options. Really knocked my ability to find stuff

Expectation should precede consistency in order of importance.

Meaning, if a new user is learning your software, how would they expect the next [flow] to go?

Consistency often shapes expectations but when things go how you expect them to, you don't need to learn new mental models.

UX designer here. I've learned that these changes are a delicate balance between allowing existing user to remain experts while improving the retention of new users.

This can be accomplished by making small changes over time that break up all the new info a user has to learn and avoiding the big UI reveal which people universally hate because all that new learning is required at once and they need to get stuff done.

Often times its required because as a product moves into mass market phase of its life cycle, it needs to be simple to use for lots of people, which mean it doenst work great for any specific goup.

Which is why I'm really in favor of allowing customizations to the user where appropriate. It allows experts to have control over their flow and new users can enjoy the UX optimized for them.

As a fellow UX designer, i've learned the same.

I still hate ribbons.

MS Office UI peeked around versions 97-2003 or so. Everything in a menu, actions grouped / categorized so they were easier to discover. QUICKLY accessible by underlined keyboard combinations (alt+menu letter THEN item in menu letter) and with any actions that had a direct keyboard shortcut annotated. Everything easily discoverable.

Ribbons, I've no idea how the categorize what's popular or not, but 'related' things seem weaker to me, and the context switch price is much higher. Plus there's a need to hunt and kill with a mouse instead of direct with the keyboard.

My first job was on a particularly verbose set if software that's hard to learn. But, everyone that uses it does so daily and quickly become experts in their workflows.

Number of clicks was the #1 metric because a "pretty" UI would generally add steps and slow everyone down.

Verbose UIs are hard to learn but once learned have much better end-to-end workflows with fewer steps. Simple UIs hide information behind interactive (i.e. slow) workflows and universally hated by customers.

You can still access the items in the ribbons with Alt key sequences (although I agree that they've become far less discoverable), but the full-screen(!) "file menu" abomination that they added in later versions is even more hostile and offensive.

Especially after they made everything cloud-first, making saving my damn documents to my local drive a pain in the ass.

F12 brings up the old "Save As" filebrowser.

Ribbons for Word used to have actual research behind them. Since then it's just cargo culting.

Sadly, many images are broken, but the blog series "Why The New UI" survives on web archive: https://web.archive.org/web/20080316101025/http://blogs.msdn...

There's an old MIX08 video where some guys from MS actually talk through the design process and everything that went into it with a lot of depth as well as looking at alternative ideas and why they didn't work.

https://www.youtube.com/watch?v=AHiNeUTgGkk

I think I've seen a video where they take the metrics they ask to collect (and everyone I've seen opts out) where the more buttons were clicked in office the bigger and further to the left they moved.

I find it weird based on my comment that windows and office is built around people who choose not to opt out either by ignoring the dialogs or by knowing what it's used for and hoping their metrics make some difference. Unfortunately I think it's mostly just people wanting to be a part of the product who really participate in user voice a lot of the time so we get weird design choices sticking like the mystery meat cut copy paste in windows 11

> It’s about letting someone become an expert in your software.

This is brilliantly said. Watching someone, who does not think that they are technical, zoom around Excel is always special to watch.

> Microsoft office was always his example.

The Microsoft UI design rules are pretty amazing. The consistency over the years allows people to upgrade every 1-2 years and continue to use their software. (I do not write this a Microsoft fanboi.) The key for each upgrade: Incremental changes to improve the UX. One thing I never understood: When Win 95 introduced the concept of "right click everywhere for properties (and deeper settings/details)", why didn't the design team reject it? After all, it is invisible to user (to indicate right click is possible). It seemed like no one understood right click the first few times they used Win 95.

I'm a programming language designer, so I think about this stuff all day every day.

I believe you're right that expertise has something to do with it, but I don't think your comment quite connects why consistency leads to expertise.

Consistency is about taking what users already know about a system and compounding that expertise across other parts of the system. When a system is consistent, once I spend the effort to learn X and Y, then I also know Z and W because they'll behave the same way. In an inconsistent system, I can't extract as much value from my existing learning investment.

I fully agree. it's also how the user takes what they know and making it work for them.

Windows 95 had years of user testing behind it. They'd sit people down who were used to DOS or Windows 3.11 and give them tasks and see where they struggled or where it worked and iterated on it and the talks by the ux teams really stuck with me. every color and design choice was important even the awful but good real background wallpaper colors

I agree with the importance of consistency, but disagree with "Office and (Neo)Vim aren’t exceptional examples of UI, but they are uniquely stable."

Office is not stable, at least over the long term. The hated "ribbon" marked a sad departure from the "stable" and efficient Word UI, and Microsoft's clueless regressions across the Windows platform have compounded the problem. For example, the deletion of the menu bar from applications. WTF.

We are about a year or 2 from the Ribbon interface existing in Office longer than the pre-Ribbon interface.

I will agree that the Ribbon and what you can customize or use in it/looks has changed drastically since 2007.

But the Ribbon “paradigm” will soon have existed longer than the non-Ribbon “stable and efficient” interface so it is more “stable” in one way.

Stabilizing on shit is a sad thing to double down on. But Microsoft is doing that at every opportunity now. Witness their offensive, relentless hounding to log in, log in, LOG IN WITH YOUR "MICROSOFT ACCOUNT!!!!!!"... or you can't do anything, including install Windows.

https://news.ycombinator.com/item?id=38925695

Thanks. I'm going to watch that MS video and see if I'm still as pissy about the ribbon afterward.

Ribbon aside, there are still lots of regressions in Word. The handling of styles, for example, which used to be excellent. Now Microsoft has stuffed the style list with dozens of asinine canned styles (most of which consist of different-colored underlines) that you can't get rid of. So instead of being able to format your document efficiently by selecting styles, you must now wade through a giant list of shit, looking for your own styles every goddamned time you want to apply one.

Just wait until the default template gets hacked up by some fluke in word and it's permanently stuck to weird formatting. you can revert it back but it's weirdly hidden and took me a Google to do it instead of just finding and clicking a "default word style format" theme thing.

This crap really drives me to use markdown and pandoc to crap out word files when I'm done. I also really hate how word deals with document formatting and images.

I have been screwing around with Microsoft C 3.0 in DOS. What a breath of fresh air it is to install Vim for editing and still have nearly all of its modern functionality (macros, syntax highlighting, folding, splits, regex, visual mode) with absolutely NO keystroke changes.

How can MS Office be this example when shortcut modification is awful there, as well as Ribbon customization??? People pride themselves on all the nonsense they've invested a huge deal of time into, that's not an indication of any quality or UI excellence Vim is another prime example of extremely user unfriendly UI design, breaking most of the principles from the linked article (but similarly, with folks who've invested a few metric tons of effort into fixing it will pride themselves)

> Vim is another prime example of extremely user unfriendly UI design

I know it's going to come off the wrong way but as an editor, I would take plain vim over VSCode/VS/IntelliJ etc any day. Give me neovim as an option and I would be even happier.

I don't spend "a few metric tons of effort into fixing it", there's nothing wrong with a modal editor with key chords as the primary interface. It's not something you can "open and just use with 0 experience" but neither is your car or any other useful complex tool.

The fact that Vim & Emacs design hasn't really changed in decades yet they still have a very vast userbase should indicate something.

> Give me neovim as an option and I would be even happier.

That's fine, the issue here isn't someone's personal happiness, but good design (my 2nd sentence already covers this apparent contradiction)

> there's nothing wrong with a modal editor with key chords as the primary interface.

There is this whole list of UI design principles that tells you what's wrong

> It's not something you can "open and just use with 0 experience" but neither is your car or any other useful complex tool.

Yeah, that's not the issue, >0 experience doesn't require bad interface. Like, a car doesn't require pressing 6 buttons with no feedback to turn left, instead you have a wheel with immediate feedback (still requires experience to get the feel for the connect at different speeds)

> The fact that Vim & Emacs design hasn't really changed in decades yet they still have a very vast userbase should indicate something.

It doesn't indicate what you imply it does. But to answer in kind: the fact that both have multiple (in total) close to complete rewrites of the whole interface (full set of key bindings, commands, tabs, plugin languages, etc.) should indicate something, and not something positive about those interfaces

The conversation we’re having is about consistency only. That’s why I said “they aren’t examples of exceptional UI”.

They just respect the fact that people have become experts in using their tools.

Sorry about broadening the scope, but even in the narrow one Ribbon broke stability in a big way consistency when it was introduced, so that's not uniquely stable

Vim is stable in a close-to-tautological "if plugins respect stability, they're stable", otherwise vim ecosystem has a lot of packages wildly departing from the default (it's definitely uniquely customizable, though)

Not to discredit, really I am actually interested and may even agree with you.

Give me an example of a text editor that you concider to have good UI design.

That's not a very useful approach, there are big issues with all of the stock configs, though with enough customization you could pick various design patterns to make it good (but still you can't collect all of the best ones in one app unfortunately, and doing that is a lot of effort). Thinking from "first principles" like this list is more appropriate

But a few examples:

The "visual-first" approach of kakoune/helix is good UI, better than the "keep the whole sequence of ops in your head". Then emacs' which-key displaying a helpful menu of key chord continuation, but only if you're stuck (after a delay) is good UI design, better than Helix's constant popups.

That's "informative feedback" right there

Then any GUI's tab management approach where you can simply drag&drop tabs around is good. Though if you could "cut/copy&paste" that would be better as that would be consistent with how you manage files in a file manager

Command palette like with fuzzy search like Sublime Text's is good UI design, that helps in reducing your memory load as you can find an action without remembering a shortcut (but then if you invoke it frequently enough you could remember the shortcut from the tip and use that instead)

Then keymaps that prioritize convenience of key locations for most commonly used keys is good UI design (don't think any text editor has it by default, most prioritize English-only semi-abiguous keycap abbreviations)

Etc. etc.

See I can agree with all of this and ironically the exact changes I make to my neovim setup address quite a few of these.

I'm going to look into helix again, I never gave it a good go since I didn't really see any benefit ovwe what I already have but maybe I was premature on it.

I just wanted to make sure we aren't on two completely different trains of thought here where potentially your answer would have been "Visual Studio is the pinnacle"...

Everything is a trade-off and for me personally I really don't want to have to touch my mouse to do common things, I'll take a slightly more intuitive UI where I need to memorize things to make doing the things I do the most often easy.

Neovim (only slightly modified, telescope + lsp and completions effectively ) + tmux or a tab based terminal is where I am happy.

I happened to spend 3-4 months in Visual Studio and fleet when I switched back to Windows (C++/golang ane some python/js) and when I eventually could dedicate the time to setup neovim correctly on Windows it felt like a breath of fresh air compared to that.

I worked with Ben Shneiderman at the UMD Human Computer Interaction Lab developing pie menus, and one of the important principles of pie menus, especially in comparison to both traditional linear menus, and invisible gestures as used by the iPad and mobile apps, is that they smoothly TRAIN novice users to become experts by using "rehearsal".

Pie menus can lead, follow, or get out of the way. The way a novice uses them is actually rehearsal for how a more experienced and experts use them.

Unlike invisible gestures, they can pop up and show users the available items. They also support reselection and browsing, which gestures don't. They also utilize 100% of possible "gesture space" as meaningful predictable actions, as opposed to gesture recognition which squanders most possible gestures as syntax errors.

Gesture Space:

https://donhopkins.medium.com/gesture-space-842e3cdc7102

Unlike pull-down menus that have keyboard shortcuts, pie menu "shortcuts" are exactly the same action a novice takes to use them in the first place, only quicker, so using them in the slow way trains you to use them in the fast way. While selecting from a linear menu with the mouse is a totally different action than selecting a menu shortcut with the keyboard.

Ben Shneiderman introduces Don Hopkins' work on pie menus in Spring 1989 on a Sun Workstation, running the NeWS window system:

https://www.youtube.com/watch?v=8Fne3j7cWzg

After an 1991 intro by Ben Shneiderman we see the older 1989 demo by Don Hopkins showing many examples of pie menus on a Sun Workstation, running the NEWS operating system.

This is work done at the Human-Computer Interaction Lab at the University of Maryland.

A pie menu is a menu technique where the items are placed along the circumference of a circle at equal radial distance from the center. Several examples are demonstrated on a Sun running NeWS window system, including the use of pie menus and gestures for window management, the simultaneous entry of 2 arguments (by using angle and distance from the center), scrollable pie menus, precision pie menus, etc. We can see that gestures were possible (with what Don call "mouse ahead" ) so you could make menu selections without even displaying the menu. Don uses an artifact he calls "mousee" so we can see what he is doing but that extra display was only used for the video, i.e. as a user you could make selections with gestures without the menu ever appearing, but the description of those more advanced features was never published.

Pretty advance for 1989... i.e. life before the Web, when mice were just starting to spread, and you could graduate from the CS department without ever even using one.

This video was published in the 1991 HCIL video but the demo itself - and recording of the video - dates back to 1989 at least, as pictures appear in the handout of the May 1989 HCIL annual Open House.

The original Pie Menu paper is Callahan, J., Hopkins, D., Weiser, M., Shneiderman, B., An empirical comparison of pie vs. linear menus; Proc. ACM CHI '88 (Washington, DC) 95-100.

Also Sparks of Innovation in Human-Computer Interaction, Shneiderman, B., Ed., Ablex (June 1993) 79-88. A later paper mentions some of the more advanced features in an history of the HyperTies system: Shneiderman, B., Plaisant, C., Botafogo, R., Hopkins, D., Weiland, W., Designing to facilitate browsing: a look back at the Hyperties work station browser Hypermedia, vol. 3, 2 (1991)101-117.

PS: For another fun historic video showing very early embedded graphical links (may be the 1st such link) + revealing all the links/menu items + gestures for page navigation

• HCIL Demo - HyperTIES Browsing

https://www.youtube.com/watch?v=fZi4gUjaGAM

https://news.ycombinator.com/item?id=11319498

DonHopkins on March 19, 2016 | parent | context | favorite | on: Motion Design Is the Future of UI

User interfaces should always be able to lead, follow, or get out of the way. Animation should never delay interaction, and it should never interfere with gestures and mouse-ahead (or whatever the input device is).

The user should never have to wait for animation to finish before they're able to do something, and the interface should never be disabled during animation, or ever ignore the user's input under any circumstances.

User input should always pre-empt and interrupt feedback and animation.

The interface should always support quick gestures (mousing ahead, touching ahead, or whatever), without ever requiring the user to pause and wait, or focus their attention on the screen to watch the animation play out before they know it's safe to make the next move.

I developed a gestural pie menu tabbed window manager for the NeWS window system in 1990, which supported mousing ahead, suppressing the pie menu display and pop-up animation until you stopped moving, showing light weight feedback on the overlay plane, and executing commands instantly without any animation or even popping up the menu, when you make a smooth quick gesture without hesitating.

NeWS Tab Window Demo:

https://www.youtube.com/watch?v=tMcmQk-q0k4

https://donhopkins.com/home/movies/TabWindowDemo.mov

Transcript of the relevant part of the demo:

Now you can press the right button to pop up a pie menu on the tab or on the frame itself. And that has commonly used commands like front and back in mnemonic directions. Back is down, and front is up.

When you make a menu selection by mousing ahead, it doesn't display the menu.

As long as you're moving, it suppresses the menu display.

And it gives you feedback on the overlay plane of the slice that you're in, and the label of that slice, so you can actually see what you're going to get before you choose it without even seeing the menu itself.

And when you wait, it pops up the menu once you stop moving.

So if you waste some time by just waiting around, it will waste a bit more time by giving you some stupid animation.

And this is meant to be negative reinforcement, to encourage you to mouse ahead.

The sub-menu pops up. This is "move to" which is unconstrained move.

You can always get that from the tab by mousing left and right.

That's an easy gesture. Just quickly...

Or mouse there and wait. There it is. It pops up the one you're at first.

This is constrained horizontal move.

And this is constrained vertical move.

So constrained horizontal... We'll wait.

Constrained vertical...

So, I mean, once you're there, and you know what you want, why wait?

This is "beam me up": put it in the next layout position. To tidy the windows.

So, if you've clicked the menu up and haven't moved, it will just spin it, because it's confused, and doesn't know what you're going to do.

----

In other words: As it pops up and scales up the round menu, it also tilts it along the axis perpendicular to the direction of movement to reinforce the selected direction, or spins around the center if you haven't moved to show no direction is selected.

And you only ever see any animation if you actually stop moving -- once you make a selection, the command always executes or the submenu always activates immediately.

You can mouse ahead smoothly through multiple levels of sub-menus, without popping any of them up or seeing any animation, as long as you never hesitate.

By "lead, follow, or get out of the way", I mean that pie menus can lead novice users by giving them feedback and animation when they pause, follow intermediate users who move in the right direction then pause for the feedback to make sure they got it right, and get out of the way of expert users who know the right direction and can quickly articulate gestures without pausing or waiting for feedback.

----

Here's another demo showing pie menus, mouse ahead gestures, and display pre-emption in SimCity:

X11 SimCity Demo:

https://www.youtube.com/watch?v=Jvi98wVUmQA

----

And here's a really old demo from June 1986 of the "uwm" window manager for the X10 window system, that I hacked to support pie menus with mouse-ahead and display pre-emption.

X10 Pie Menu Window Manager:

https://www.youtube.com/watch?v=IJhvB6kwmog

---

More info here:

The Design and Implementation of Pie Menus -- Dr. Dobb's Journal, Dec. 1991:

https://donhopkins.medium.com/the-design-and-implementation-...

Do you know of any work about how to combine pie menus and keyboard usage? Mousing is very uncomfortable for me, but if a pie menu could be used with a keyboard-attached joystick, it might be a really quick way to work.

Pie menus work well with trackpads and trackpoints (keyboard clitoris), as well as analog and digital 4- or 8-directional joysticks, and even numeric keypads and arrow keys.

If you arrange your menus into 4- and 8-item pie menus, they are uniformly navigable and memorable for many types of input devices including keyboards. Four and eight items are ideal for muscle memory, and also cognitive memory too. So using pie menu layouts that map directly to keyboards and digital joysticks works really well.

The ActiveX pie menus I implemented for Internet Explorer a couple decades ago supported full keyboard navigation:

https://www.youtube.com/watch?v=nnC8x9x3Xag

Basically, you use cursor keys to select the direction. So you can just memorize the paths to the correct items.

Modern console games use the same idea for communication wheels, that are usable from game controllers.

Great for game controllers and keyboard use but I find them horrible for mouse use, personally. moving a bound mouse cursor around a donut is really awkward when you're using to x or y movement

Thanks. A lot of good stuff here that may go overlooked.

both vim and neovim are examples of UI, just not (G)UI :)

Reply on news.ycombinator.com