It seems from the finer details, a lot might be AI generated, which at least to me would defeat the message about how our interfaces are less human. I'd be very glad to be proven wrong, and it's just the style.

Aren't these visuals just AI generated fillers?

I agree with other commenters that it seems AI generated (patterns/noise in fine detail, nonsense text, and inconsistency in style, particularly in the faces of the "dolls").

The first few images are impressively well curated and consistent though. I think the "rotoscoping"/"compositing" helps quite a bit to make it feel more cohesive and intentional - certainly it's a step above the usual "blog post header image" slop.

Maybe if I can make a counter-point: a lot of these patterns are common place right now! And much more so than whatever golden era we want to imagine existed long ago.

- Gestures in a lot of applications have made things more confusing by hiding functionality that you now need to stumble into to discover.

- Sound cues are used all over the place. Anyone who's ever worked in a kitchen hears the godforsaken ubereats alert sound in their nightmares.

- About ten minutes ago, I got startled by my phone deciding that the "you should stand up" vibration pattern should be three long BZZZZ-es... amplified by it sitting on my hollow-sounding printer.

- If another fucking god damn website asks me to chat with an AI agent in it's stupid little floating chat bubble, only appearing AFTER I interact with the page so it's allowed to also make an annoying "chirp!" sound, I WILL become a chicken farmer in some remote forest eating only twigs, berries, and improperly-raised chicken eggs.

All of these things annoy me, and actively make me hate computers. A silent glass brick can go in my pocket because I know it's not going to bother me or beg me to talk outloud to it. If it was some sensory-overload distraction machine (which, by default, it is) it would find itself over the side of a bridge rather quickly. It's getting in the way of my human experience! The one where I'm the human, not the computer!!

This is a beautifully designed and illustrated page.

But I couldn't disagree more with the premise. It complains that computers have been reduced from physical, tactile, hulking mainframes to neutered generic text interfaces, but I've watched the opposite happen over the past two decades.

My phone is physical -- I swipe, pinch, and tap. It buzzes and dings and flashes. I squeeze my AirPods, I pay by holding my wrist up to a sensor, I tilt my iPad to play video games and draw on it with a pencil.

Everything the article complains about, we've already solved. All of its suggestions, we already have. It wants "multi-modality" but we already have that too -- I can change the volume on my iPhone with physical buttons while I dictate. I can listen to music while I scroll.

Our interfaces haven't lost their senses. Our interfaces have more senses than they've ever had before.

Disagree: Our malaise is not boredom from simplicity, but fatigue from inconsistency.

"Flat" interfaces aren't bad because they lack an ineffable whimsy of embodied human experience, they're bad because they threw out the baby the bathwater, tossing decades of conventions and hard-learned accessibility lessons in the name of supporting a touchscreen.

Compared to 20 years ago, everyone is shipping half-website-half-desktop abominations (e.g. with Electron[0]) and reinventing UX wheels. Too many apps/sites impose "their own look" instead of following what the user has already learned. [1] Often users must guess whether certain things are even clickable, how a certain toggle looks when enabled, whether certain settings are a single-select option or a multi-select tickbox... And memorize those rules with per-app or per-website granularity.

> You can talk while clicking, listen while reading, look at an image while spinning a knob, gesture while talking.

Those are all things people do after "make computer do what I want" has become automatic.

Now when--for example--trying to find the 21st item they just added inside a list that is vertically limited to 20 and the custom grey-on-grey scrollbar is always hidden unless you've currently hovering a mouse exactly in the right 5-pixel-wide strip between two columns of the interface.

[0] A sample listing of software readers may be familiar with: https://www.electronjs.org/apps

[1] That may be due to deliberate "remember us" branding, whatever was fastest-to-ship, because things to look new to get somebody a promotion, because they want to create a switching-cost so current users feel bad trying to use a competitor's product... Or because someone like the blog-poster has misguidedly tried to make a "richer experience."

These beautiful images (AI generated, perhaps?) make for a great showcase, but I find myself disagreeing with almost everything here - except for the core desire to make interfaces more engaging.

The real challenge is that UI designs are ultimately constrained by their hardware. This means major interface innovations often limit where the software can actually be used.

Take tablet-optimized apps, for instance. They can fully embrace touch interaction, but this leaves desktop-only users completely out of the loop.

So unfortunately, truly revolutionary interfaces tend to require equally revolutionary hardware to match .

Fantastic design. Normally pages with funky scrolling behavior and boxes whizzing all over the place and all that are annoying but it really works here. Not to mention the adorable visuals.

That being said I think it misses what made the old physical interfaces so appealing and useful. It's not that there's something inherently superior about multimodality; it's that physical interfaces are permanent, with defined edges and definite shape. Unlike screens you know exactly what's where, building muscle memory every time you use it. There are no hidden menus or moving parts.

Multimodality - such as being able to see the position of a slider at a glance, or feel its position by touch - is useful because it reinforces the absolute existence of a control and its state across multiple senses. Interfaces using voice and gestures like suggested are the exact opposite of that, because each point of interaction becomes even more disconnected and vague.

This kinda reminds me of how, in the wake of the smartphone, for a few years every company thought they needed to boost engagement with their product. Even if their product was something in the background that people are happiest not thinking about. Do we need to engage with our oil filters? With our clothes washers? With our insurance policies?

Some things are best if they stay simple, efficient, reliable stable, and quiet. Not needy, demanding, high-maintenance, attempting to ensnare us through as many of our senses as they can get their claws on.

Some things are an experience, other things should just be quietly useful. Do we ask ourselves which we should be, before adding another colorful icon, with a red dot in the corner, with a number inside the red dot, to the poor user's screen?

And I hate haptic feedback. I keep my phone on silent 24/7 just to not feel my phone creepily zapping my fingers, and for some reason silent mode is the only way I can accomplish that.

Hm, no reference to Bret Victor?

https://worrydream.com/ABriefRantOnTheFutureOfInteractionDes...

It's a lovely set of sentiments. I think another aspect of UI that has been lost is discoverability - finding out how to do things in a new interface seems harder than it used to be when there was one app-level menu bar. Too many things are hidden in context menus, found only by right-clicking or long pressing on just the right spot. A set of multi-modal interfaces might just make discoverability even worse.

I was reflecting on something similar to this this while photographing the recent lunar eclipse with a Fujifilm X-T5, a highly tactile camera that is just an absolute joy to operate.

I was on my roof in the dark at 1:30 in the morning in the cold and wind. I'm tired, can't really see much, but still need to actively work with the camera's controls. Thankfully, the X-T5 is covered in physical dials, switches and buttons. Without looking at the camera's screen, I can quickly change shooting modes and the majority of the settings I care about and be confident that I changed the right things.

The same cannot be said about a large number of modern cameras, which opt instead for a more digital approach.

In terms of modern "computing" devices, my cameras are an absolute joy to use compared to most of my other hardware.

So much so that I've recently been finding myself looking to recreate this tactile experience on my general purpose computers. I've been looking at weird bespoke dials, switches and various input hardware to make processing the photos (among other tasks) feel more tactile.

I got agitated looking through that due to the excess of flourishes. Fancy elements should punctuate focal points. If there's too many, the focus is lost.

I think the article laments over a lack of something that interfaces have legitimately embraced for some time. Gestures, audio control, interactivity, visualizations, and so on are all things we’ve seen an increase in over the decades, not vise versa. Whether it’s done to a degree and in a manner that suits the author is another matter. That in itself leads to another rebuttal: As someone who is easily overwhelmed by their senses, simplicity and accessibility should be the priority. Surely, there are times when a rich interaction can be extremely useful (why just talk about physics when you can also let the reader interact with the concepts[0]). On the other hand, it’s easy to become flustered when someone imposes their artistic flair, conceptual model, or worse (when businesses weaponize interfaces against the user). I look at the author’s note organization mock up and I feel legitimate anxiety as it looks like little more than chaos on the screen.

0 - https://ciechanow.ski/airfoil/

There is a certain beauty of a webpage about user interfaces failing to load under strains from traffic volume. I couldn’t read much, but it would appear the best interfaces are the ones that work!

Yes and no.

Yes, flat design is too flat, and AI chat is too devoid of friction.

But mobile and tablets are better at certain things [1], and we shouldn't get rid of that either.

I saw somewhere (Bret Victor?) that tools have two parts: the part that fits the problem, and the part that fits the human. The example was a hammer; the head fit the problem (the nail), and the handle fit the human (the hand).

Notably, the two parts must fit their respective things, but they also have to work together.

That is what we should be doing: creating harmonious tools that fit the problem and the human. What that looks like will be different for every tool.

Our interfaces currently have two problems:

* Because they can have any appearance, appearance gets more attention than being a good tool. Example: flat design (good appearance) overriding skeuomorphic design (human fit).

* No one wants to redesign everything, so we all reuse the same base stuff (Electron, Qt, etc.) even if the result won't fit (one or both ends) or harmonize.

I would love to fix both of those problems, but because people are lazy, it essentially means creating a GUI framework that is flexible enough to fit almost any problem and any human (accessibility included) while making sure that flexibility does not destroy harmony.

While I am working on that, it is a tall order, and I am almost certain I will not succeed.

[1]: https://news.ycombinator.com/item?id=43350339

This feels like a article against the fur trade that was written on a rare animal skin.

Reminds me of Bret Victor's article,"A brief rant on the future of Interaction design".

https://worrydream.com/ABriefRantOnTheFutureOfInteractionDes...

I guess one on the reasons why he's building Dynamic Land.

This is a nice and visually pleasing manifesto.

It is also hard to read.

An article on user interfaces that is barely usable on a mobile phone due to scroll hijacking is hardly making a convincing point.

I think of this trend every time I try to connect my bluetooth headphones to a third device. They'll tolerate two just fine but if you want a third you have to puzzle out which other two they're connected to, go find one of them and disable bluetooth on it. Then you can power cycle the headphones and your third device will now be your second.

I want some kind of magical piece of string which I can touch to both devices as a way of saying:

"you two, communicate now"

And then later, to break the spell, I'll just touch the ends of that string together.

I don't want to have to dig through settings, I want to manipulate physical objects around me.

It might just be me but I find the thesis of the article to be very confusing.

> but we should have made typing feel like painting.

Maybe painting should should feel like painting and typing should feel like typing? I don't know about others but when I type, I just want to type, as efficiently and quickly as possible. I definitely don't want typing to feel like painting.

By the way, loading 92 MB of images to make me read 6 KB of text is brutal!

There is a reason we are using a keyboard to interface with this stuff. We’ve been writing to think for millennia. Using a keyboard to do it is just marginally more efficient and less of a strain on your wrist.

In The Great Flattening section of the post the author literally argues that the way we interacted with computers back in the 50s-70s was better because it was more of a full-body experience. That's a silly argument to make. As far as the status quo HCI paradigm goes, we've obviously made a lot of progress over the last 50 years.

However, I think the post is striking a chord because it's pointing to a deeper truth: after 70 years, we are still only scratching the surface of all the ways that humans and computers can potentially interact with each other.

Seems to be a call for the return of skeuomorphic UI, and combining it with things like haptics (actually, fairly classic).

TBH, I'm not especially against the idea, except that, if you make something look like a real-world object, it's important to make it behave like one.

There's a hell of a lot of digital interfaces (not just touchscreen stuff -digital dials and switches can also have the issue), that look like they should behave a certain way, but don't actually do it.

I understand the sentiment, but I believe Brad Woods explained it better in https://garden.bradwoods.io/notes/design/juice .

Flat design is much more of a blank canvas than old desktop UI toolkits were, but companies are loathe to invest in "Juice" because it requires a level of attention to detail that works against fast iteration.

There is a niceness to more kinesthetic input devices, to dials, knobs, and pens. I'm always experimenting with trying more. The unfortunate thing is they tend to be niche and unsupported. Try finding a nice dial to control things like zoom or volume, it's harder than it should be and costs over $100 or is not a great experience.

First of all: this is utterly beautiful, thank you!

There is an important thing about computer interfaces which does not hold for playing an instrument, or woodworking, or dancing, etc: it very often needs to be done privately and discreetly enough, both to avoid publicizing your private affairs, and to allow several people work near each other, or work near people who are not working, without bothering them.

This significantly limits the use of the body language (finger gestures are OK; gestures, so-so; poses, no), sound (listening is OK provided headphones, speaking or singing, much less so), haptics (some force feedback is OK, noisy vibration and clicking, less so).

This does not mean that all these things are useless! Not at all; audio feedback may be welcome, haptic feedback may be welcome, but both should be optional (e.g. in a game), unless sound is what you work with anyway (e.g. in a DAW).

What I miss is the richness of erstwhile interfaces, with their colors, shapes, textures, pseudo-3D effects, the occasional skeumorphism. They gave so many clues, helpful hints that encouraged exploration, and provided large amounts of subtle feedback. Sadly, they were considered "noisy" and removed by æsthetics purists, likely the same people who designed the sharp, wrist-biting "clean" edges of certain Macbook models. This regress needs to be revested.

Something has to be said about non-keyboard controls. They are few, and non-standard, with the exception of touchpad / touchscreen and mice. They allowed direct manipulation and gestures to flourish! But, unlike knobs, these cannot give reasonable haptic feedback. Musicians have all the best devices like that, and those who need such controls repurpose MIDI devices with their knobs and faders; GIMP, a raster graphics editor, even has a special MIDI device configuration section in settings. Another good source of haptic controls are mice with their wheels, and, more rare, game pads with joysticks. Neither is assumed to be connected to an average laptop though, which limits their use in interfaces, however optional.

> This is a beautifully designed and illustrated page.

Hard disagree. It's incredibly distracting and the constant movement of text, the introduction and disappearance of images within the medium makes it incredibly difficult to concentrate on the message.

It screams 'look at me, I'm really smart with all these neat effects'. But you know what interface for articles like this has served us pretty well for > 1000 years? Just the words. Please, just display the words rather than this conceit.

Our interfaces have more modalities than before but they are disconnected from both physical and emotional reality. Buzzes and dings and flaps are nothing like hearing a happy shout from a friend or feeling the 'clunk' of an actual motor starter engaging.

I totally agree with OP that the 'flat' visual style is appalling. (And gray-on-gray text is an obscenity.)

The article reads like a description of personal computing in the late 90s to early 2000s. It also reads very similarly to Apple’s early marketing around multitouch displays.

> This is a beautifully designed and illustrated page.

The artwork on those is stunning. It's hard to imagine someone would spend so much time and effort illustrating an article with so little content. (the other articles on that page are similarly well illustrated)

If this article were created 5 years ago, I would be downright impressed. Sadly, I've defaulted to assume the artwork was generated by AI those days. (even though I have no evidence of that)

It doesn't help that the author claims they "empower devs with AI" in their home page, and their older webpage from 2019 [0], while still very beautiful, isn't illustrated to the same standard.

I just wish authors were upfront that they generated the artwork with AI (with a little caption, footnote, something; the same kind of thing newspapers use to credit their photojournalists). I really have nothing against using AI for this kind of thing and regardless of whether AI was used, the author of the article for sure has a lot of artistic merit for the composition as a whole.

[0] https://2019.wattenberger.com/

There has been a number of attempts at making screens create tactile bumps and provide direct feedback which haven't yet worked which might improve physical interaction somewhat so we can get buttons and switches and knobs in a programmable way that isn't hardware specific to the task but we aren't there yet.

Beautiful? It looks like utter garbage to me. I really can't abide that twee visual style. The designer is trying way too hard and completely lost the plot.

Honestly using GenAI slop pictures to illustrate the article about the soullessness of modern computing clashes with the message in a way I don't think the author intended.

  > Our interfaces haven't lost their senses. Our interfaces have more senses than they've ever had before.

Hard disagree. Let's take a very simple example, Wikipedia[0]. Took way too long to build in a dark mode and when they do they have the options "light", "dark", and "automatic". YET the default value is "light". WHY THE FUCK IS THERE AN AUTOMATIC IF THIS ISN'T THE DEFAULT!? Obvious stuff like this is everywhere.

I find a lot of interfaces INFURIATING. My car wants to do things with touch screens while I want to feel because I want to keep my eyes on the road. My iPhone won't capitalize the letter I and will change not just the word I'm typing but the word previous to it making swipe style texting painful to use. Speaking of the iPhone, it's 2025 and there's no universal back. I still don't know how to exit the YouTube popup asking me to activate my free trial of premium other than swiping close the whole app and reopening[1]. Or I scroll through an app with threads (e.g. Twitter) and I move slightly left or right and bam I'm on a different tab and when I move back I'm not where I left off but somewhere completely new.

You may say "well that's a 'you' problem, I'm happy with the way things are" and my point is that humans are all different. There's no one size fits all. Maybe that swiping thing happens because our thumbs are different sizes or our phones are different sizes. Maybe you like light mode and don't open any websites with the lights off. But that difference is what makes us human. The problem is that things are converging to things that are bad for everyone. Design matters a lot and getting used to a design is very different than designing things around people. A well designed product needs no instructions (obviously not absolute), just see the "Norman Door."[2] We shit on backend developers for making shitty UIs (as a 'backend' person, I agree, this deserves criticism) but I don't think the front end people are at all concerned with design now a days either. There's a special irony with Apple, considering the magic was the interaction between Jobs and Woz. The magic is when the good backend meets good frontend. Yet now we're just doing both like it is a competition of who can create the worst thing the fastest.

[0] https://en.wikipedia.org/wiki/Hacker_News

[1] I now use Orion browser. The video quality is lower but it is better than dealing with this bullshit.

[2] https://99percentinvisible.org/article/norman-doors-dont-kno...

To add insult to injury, not only is everything inconsistent thanks to incessant wheel reinvention, nearly all of the reinvented wheels are halfassed at best and missing functionality compared to what they’re replacing. When a company writes a new widget to match their theme they only build the bare minimum necessary to visually match mockups. UI controls have become vapid and devoid of function.

Agree. The constant UI reinvention of lately is super strange to me. Companies want to save money but most developer time in simple apps is spent on it.

I remember 20-25 years ago mostly using Windows widgets to make Enterprise apps. It was fast to make, fast to run, and users back then knew how to use. Didn't look the best, but at least was consistent.

The next 5 years we sort of tried to do our best, but most things were still sort of standard-ish.

Then for about 5 years or things like Bootstrap, Material, etc, dominated. It was nothing special, but at least consistent between apps.

But in the last 10 years pretty much every company I worked on had a custom UI built ENTIRELY from scratch by a designer and a small army of developers to implement it. It looks "the same but different" in an uncanny valley way.

I honestly feel like this is the worst possible use of frontend developers, period. Not only from a financial perspective but also from an end result.

But hey jobs are generated so what do I know...

The "great flattening" fashion has made computers harder to use for many people.

Older UIs separated computer output from where user input is expected. Compare Windows 7 or 2000 to Windows 10 or 11 and the older Windows versions win. It's the same with Android.

UX designers follow fashion as much as the garment manufacturers on the catwalks of Milan, even if the fashion is uncomfortable.

Just because something is new, that doesn't mean it is better.

I'm currently using Gnome and their UI may not be the most beautiful or complete, but they've gone all in on consistency. I don't mind software like Blender, Audacity, and others having their own design systems as their domain is much more complicated. But a lot of software only needs a few controls and the native ones should suffice.

There is also a big elephant in the room that we are sort of ignoring with the whole AI stuff, which is when flat design came about, a lot of the designers who weren't really good now suddenly had jobs because everybody could put a flat thing on the page and call it a "button".

Good designers still exist but they are simply crowded out.

The same is happening today with the AI generated apps. Most front ends now in another 10 years will be filled with AI generated apps. But good design and applications will be around but they will be crowded out.

And you see this in almost other industries as well. For example, architecture has simply gotten worse. A building from today looks much, much worse than let's say a building from even 300 years back.

So we will simply have worse software and worse performing software which breaks down all the time in the near future and we will all suffer but there is no solution out of this.

Things don't always get better.