But lately, over the past 5 or 10 years, it seems to me that perfection in UI is just as arbitrary and mutable as people's tastes and preferences.
It's hard to admit it to myself, but I think my love for the early Mac OS and Windows 9x UIs was mere puppy love at first sight, and now is simply nostalgia.
To me, it seems very related to the idea of how to fall in love with a person. There seems to be nothing you can measure it against. You simply either do or do not feel a connection with the person, an inexplicable infatuation. And if you do, then that love cools and settles into something more subtle but just as real over decades, until you're holding hands on your deathbeds. Yet I can't for the life of me figure out how it begins, or what its metrics are, or where its catylists come from. I suppose this is what Randall Monroe wondered all those years ago when he came up with his blog's subtitle. If only I could ask him, perhaps I would have the answers to everything.
The Macintosh Human Interface Guidelines, especially the old ones, were seriously well thought out. Apple just doesn't follow them anymore, and that leads to demonstrably poor design choices. "Consistency" is still one of the three banner concepts (see https://developer.apple.com/design/human-interface-guideline...), and Apple just fails wildly at paying attention to itself.
Windows 2000 and Mac OS 9.x.x were contemporary with each other, are both considered to be strong examples of principled interface execution, and they were very different from each other in both behavior and appearance. They were also notably both the last generations of iterative executions of principled designs before both companies started veering off into the wilderness.
By that logic, you shouldn't have any particular preference for newer UIs (apart from how similar they are to your "primed" UIs) and you shouldn't ever able to discover new UI patterns you particularly like. You should also be unable to articulate why certain UIs are good or bad, or your reasons should be wildly inconsistent.
I don't think this is generally the case. There lots of articles like this one, and usually the takeaways are similar: UIs should be predictable and consistent and allow the user to reliably find actions and elements. Ideally they should also have markers and "fast paths" to allow more experienced users to find and do an action quickly. They should not overwhelm the user with too much irrelevant things.
That's the very high-level gist of it. There is actually lots of research in the details of it, which you can get a glimpse of in books like "The Design of Everyday Things".
They just don't seem to have a very high priority in current tech development for some reason.
That a good part of the industry seems to have essentially given up on GUIs, left them to the fashionistas and engagement maximizers and retreated to the command line and TUIs instead also doesn't bode well.
I do agree that Steve was somewhat revolutionary in this, bringing his personal quest for beauty into early GUIs through his (overly) perfectionistic command. I remember a decade or more ago being struck by just how genius it was for him to bring professional typography into personal computers, complete with kerning and everything.
GUIs are tools first and art second.
Any GUI that looks good, but gets in the way, or fails to make the task at hand easy is not a good GUI.
The books and research are about making GUIs functional in an effective way. No one is claiming that a GUI that follows all the principles will look good. However it will make the task at hand easier to perform than it otherwise would be.
The problem is when people in charge of GUIs focus exclusively on Graphic Design instead of UI design. They are different fields, with different goals that only sometimes overlap.
Also there are certainly better UIs than others, otherwise people wouldn't complain about software having bad UIs (see GIMP, older versions of Blender, etc).
For example, I also am old enough to have used DOS before Windows 3.1 came out, which was my first GUI. When Windows 95 came out, it was a clear improvement, but retained the same principles as Windows 3.1 began. Windows 98 iterated on it, and Windows 2k perfected it in my eyes.
So that when Windows XP came out, it abandoned the principles, going for a look that felt cartoonish and childish to me, but for others was perhaps casual and inviting. It was during this time that I discovered Compiz and other Linux eye candy, and although it abandoned the fundamental principles that Win 3.1 planted, it almost admitted this with pride, submitting a new set of principles altogether.
So when Windows Vista came out, clearly trying to compete in the arena of that new set of principles of beauty, I was ambivalent but mostly impressed. That's when I found out about Mac OS X, which Compiz et al. were clumsily imitating, and I fell in love with them, which I realized had perfected those principles before Microsoft and Linux even began to imitate them.
It almost seems like the same concept as the original purpose of MMA (mixed martial arts). There is a perfection particularly to a set of principles. You can be the best at boxing, or the best at Brazilian Jui-Jitsu, and it's almost comparing apples to oranges because they're so fundamentally different that they don't actually mix well (the current UFC being proof that the experiment has failed and created a monster).
It's the same reason movies exist like Home Front: it's that age old question, "who would win if ...", in this case London gangsters vs Southern American gangsters. Or Freddy vs Jason, or Alien vs Predator. I wish I could remember more, because those are some of the most interesting types of movies, with different real human cultures being pit against each other. Like David and Goliath, the top champions of two cultures facing off for the world to see.
I think MacPaint is one of the most beautifully designed GUIs ever created.
Consistency is always better. App A and App B should use the same save icon, without any caveats, regardless of what it actually is. If you offer me a consistent experience, I'll take it whether it's Windows 95 or Tahoe.
Based on the measurement of these outcomes you can absolutely say whether an interface is better or worse than another.
Anyone in software design in the 90s/early 00s was quite familiar with this subfield of human factors. Very few people making software today are. As vibe coded UIs become more and more widespread, it's only going to get worse.
No. Perfection exists, but it's not how something works or looks. Perfection is stability. A bad solution that you know inside and out is preferable to a "better" solution that you have to spend time to think about every time you use it.
One issue of the recent redesigns is that they degrade rather than improve “how it works”.
Yes, it is. UI is not not about stable form, alone.
> Otherwise any old solution that does not change would be good.
UI design is nuanced, regardless of the large number of people who dismiss it as borderline irrelevant. eg The most common used elements should be the easiest to recognize and access. Culture driving change is the most useful, while random association is less useful and statistically worse for user experience.
Apple has taken a great leap in misinterpreting that if the most common functions benefit from iconography, there must be a benefit the icons convey, as opposed to the other way around. It's disturbing.
