[neepi]

This reminds me of something that happened to us recently. So one of the internal things we have is a giant turd that sits on top of some Oracle/SAP mess. The user interface is circa 2001 and hasn’t changed really. Looks like a cross between Motif and GTK1 it’s that bad. The icons are off the shelf ones from some clip art collection.

So they need to rewrite it and got a consultancy to do it. The whole UI got rewritten using react and used an off the shelf flat design. A week after launch they sent out a user survey and nearly every respondent complained about the flat UI. So they made it look like the old one again very quickly.

So now we have a shiny new React app that looks like something from 2001.

I have to wonder if anyone likes flat user interfaces or just user studies are broken.

[cardanome]

I think modern UI is mostly optimized for onboarding new users. For that it makes sense to have a more minimalist UI that doesn't make the user think too hard and does not overwhelm them.

But if you need to do actual work you want maximum information density. You want icons that are easy to tell apart by color, not some sleek minimalist grey in grey.

If you use a tool every day for multiple hours your UI needs will be vastly different. We have forgotten how to build tools for power users.

[kunzhi]

> But if you need to do actual work you want maximum information density. You want icons that are easy to tell apart by color, not some sleek minimalist grey in grey.

IMHO Photoshop is still the classic example of this. The UI can feel overwhelming at first, like dropping into a helicopter cockpit. But once you start getting a hang of what you're doing, anything more "minimal" just feels like dumbing things down for the lowest common denominator.

From what I've seen in large enterprises, it's also why OG users are so attached to their mainframe terminal UIs. Yes, it's very hard to learn, but once you've developed some facility, everything else feels unusably slow.

I've never had a bad experience designing like I respect the users intelligence. Humans are insanely smart and capable, treat them that way and good results occur.

[jimmaswell]

GIMP's icon redeisign and new tool layout were a massive mistake IMO, first thing I do on a new install is disable tool groups and change the color scheme to "legacy"

[GuinansEyebrows]

> IMHO Photoshop is still the classic example of this. The UI can feel overwhelming at first, like dropping into a helicopter cockpit. But once you start getting a hang of what you're doing, anything more "minimal" just feels like dumbing things down for the lowest common denominator.

while i agree, i wish more dense applications like Photoshop took the Rhino3D approach of integrating a CLI directly into the interface. yes, you can click the icons or select tools from the menu, but being able to just type a command and arguments (or have it prompt for the arguments ex-post-facto) feels just incredible in an otherwise-GUI application, in a way that memorizing keyboard shortcuts just doesn't compare.

[falcor84]

> Humans are insanely smart and capable, treat them that way and good results occur.

My experience isn't quite that. While most humans can be capable when they want to, in typical situations they often don't and aren't. People who have put in years to become proficient in mainframe terminals aren't representative people in a typical situation; most people (myself definitely included) perform most daily actions on autopilot.

[skydhash]

And autopilot rely on consistent environment. Full focus isn't sustainable. Training and practice rely on this capability, ingraining something in muscle memory so you don't have to pay that much attention. When someone wants to fill a form or process some data, the less he/she pays attention to each action, the better it is.

EDIT

And that's why I like Vim and most TUI that much. I don't need to follow the cursor or wait an abitrary amount of time because "reasons". It's all muscle memory, and my attention is more on what I'm trying to do than how I'm doing it.

[neepi]

The mainframe terminal user interface was shown to me in all its glory by Italian gate staff at the airport in Rome. Literally hammered through a security check after I was flagged as having two return flights. It was no less than a form of kata.

I can imagine how shit it would have been if you have to log into windows and open a web app and use the mouse and stuff to click through a web form hacked up to do the job.

[rondini]

How is photoshop a classic example when all of the icons and controls are quite literally grey in grey like the person you quoted was denouncing?

[skydhash]

It's not grey on grey. It has cleanly delineated section, and most of them has been in the same place for ages. And there's thing like tooltips that help.

[DiggyJohnson]

because of the information / functionality density and the fact that it's optimized for power users

[starkparker]

> Humans are insanely smart and capable, treat them that way and good results occur

Obviously, maybe more obviouisly now than ever in recorded history, not all humans are smart or capable.

Regardless of capability, however, many humans excel at memorizing complex routes across obscure paths that they experience through spaced repetition, which research suggests can alter memory pathways in the brain to facilitate easier recall[1] and also engages memory functions in our nervous systems beyond our brain.[2]

Any UI, including bad ones, can foster efficient workflows in any user _if_ it accomplishes things compatible with repetitive use:

- the UI's behaviors and interactions are minimally interally consistent

- the UI has pathways from a starting point to a result that are discoverable through those behaviors and interactions

- the UI's reactions to input are sufficiently efficient to avoid arbitrary or dynamic pauses, which can disrupt effective repetition

- the UI's interactions are minimally accessible to people; if they use buttons, shapes, colors, sounds, controls, etc., a person can consistently distinguish between and physically access them when necessary

- a person interacts with the UI long enough to find those pathways from starting points to results, and does so repetitively over long time spans

Modern UI design often attempts to reduce the time to value for users at arbitrary experience levels, at the expense of maintaining the consistency of pathways that reward longtime users who have accumulated training.

The only people using the UI when the change happens are people with a non-zero amount of accumulated training. Any change disrupts consistency. It's a net negative to the people who are around to complain about it, and also resets the often competitive field of users; not only do experienced users have to relearn their workflows to avoid committing errors or wasting time, they also have to compete with new users who have easier access to results that previously required experience through repetition to efficiently reach.

For example, a UI designer might change the UI to surface a feature that they want users to access more easily by making it require 1 or 2 interactions to reach, but a veteran user already has "easy" access to that feature even if it takes 6 or 7 interactions to reach it, some of them obscure. If the change removes the result from the end of the old pathway and moves it to a new one that experienced users don't know, the new UI becomes less efficient for them no matter how smart or capable they are (or aren't). Both the new user and experienced user might be smart and capable or stupid and incompetent; the differentiating factor is experience.

Arguably, the "smart and capable humans" who use complex UIs are either the ones who achieve a level of power to prevent UI changes that degrade consistency of existing pathways to preserve their productivity at the expense of less-experienced users needing more time and training (at which point they probably don't need to use that UI anymore anyway, and the act mostly rewards other experienced users), or the ones who divert time that might be spent complaining about UI changes toward adapting to the new UI's pathways.

The truly disruptive UI/UX changes for repetitively used workflows are the ones that introduce unpredictable delays between interactions. Repetition rewards rhythm and consistent feedback, and unpredictable interaction delays destroy both.

1: https://www.nature.com/articles/s41586-024-07425-w

2: https://www.nature.com/articles/s41467-024-53922-x

[robertlagrant]

It's worse - modern UI is for onboarding new customers. The calculus of what to prioritise should change when your customers aren't your users.

[KronisLV]

> A week after launch they sent out a user survey and nearly every respondent complained about the flat UI.

When have most users ever enjoyed a new UI in a system that they're used to? Genuinely asking, because while I enjoy things like new icon themes and even the UI of Windows 11, most of the time I've seen people complain about any new UI that displaces something that they're familiar with.

If I'm wrong and it's just the flat design that is the real issue (which might also be true), then wouldn't the solution be to pick any other modern look and feel, instead of necessarily reverting to the very old one? Not that there's necessarily anything wrong with the more old timey UI design, I think that Windows 9X versions had really good design, perhaps despite some usability issues like no proper fuzzy search in the actual OS etc.

I quite like how themable PrimeVue/PrimeReact/PrimeNG is and swapping themes shouldn't be something impossible, though I don't doubt that with many of the libraries out there that ends up being the case: https://primereact.org/inputgroup/ (click the little palette in the corner to switch themes)

[cardanome]

> When have most users ever enjoyed a new UI in a system that they're used to?

Blender.

Is has seen some drastic changes in UI but barely any backlash. Even holy-cows like right-click select got mercilessly slaughtered and I am not even mad about it, in fact I love the changes.

The main thing is that they are focusing on providing value to users and are dog footing their own software to create movies.

But yeah, generally people hate change and you should avoid changing things as much as possible. Sadly that doesn't work with the way incentives in most companies work.

[b3ing]

Yeah, you have to change things and make things look good and on trend to get hired or promoted. If you work on something that looks a few years old in your portfolio, you will be on the job market longer

[KronisLV]

Honestly, Blender is probably the best example anyone could bring up - thanks for that, they really did wonders with the UI/UX!

[neepi]

Well the problem was that they got a new UI anyway when they didn’t need or want one.

I think that was the problem.

[skydhash]

Flat design doesn't help with discoverability, because you're never sure what's an icon and a button, and if it's one, what it is for. But familiarity is another constraint, especially spatial relations and action flows.

[mrweasel]

> I have to wonder if anyone likes flat user interfaces or just user studies are broken.

My impression was that it was an attempt to get engineers to be able to do the design, rather than involving graphic artists.

I also think that we often conflate pretty with usable. There's nothing more interesting than these user interfaces that has grown organically for 20+ years. They look "bad" or at least old, but that doesn't mean that they necessarily have poor ergonomics. Some people, myself included, have tried to force that old-school, hodgepodge look, but you can't really do that either, it doesn't work. You just end up with ugly and confusing. Those interfaces has to evolve organically.

[williamdclt]

They actually polled for user feedback, listened to it and acted on it quickly? That sounds like a great consultancy!

TBH I’d guess that people would have complained whatever the new design was, people hate change. But yes, flat designs are often on the worse end of the spectrum

[neepi]

No we dug out a faculty statistician and did the analysis and paid them even more money to fuck it off.

The consultancy resisted this horribly because the tech lead is a performative bullshitter and it doesn’t look good on his portfolio page that it looks like something from 2001.

[michaelmrose]

He should show both the prettier and the ugly version side by side and give it as an example of him being adaptable and willing to please his client instead of himself.

[dahjelle]

> I have to wonder if anyone likes flat user interfaces or just user studies are broken.

I don’t have any experience in running user studies, but it sounds difficult separate the momentary frustration and drop in efficiency that a change in _familiarity_ brings from an actual difference in long-term _usability_. Do you know if the user survey the consultancy did tried to account for this?

[giva]

Sane people don't want to invest time learning a new UI just because it's prettier.

[esafak]

But it might improve the UX for others. You can't please everyone.

[nkrisc]

Doing a proper, unbiased usability study is difficult. I only worked at once company that had a dedicated user research team. The rest the time we designers had to run remote usability tests ourselves, for our own work, which is of course a pretty poor way to get any kind of objective result. No one wants to say, "hey, I tested my design and it sucked hard. Can I have another four weeks to try again?" Furthermore, we often didn't really have much of a say in the fundamentals of the visual appearance, we had to follow the brand standards and GUI pattern library that was chosen by other people. The brand guidelines were often delivered from on high from an outside consultancy, while the GUI pattern library was often produced in-house and robustly tested, but never perfect.

[xandrius]

Sounds like a win for the tech team at least.

[hulitu]

> Looks like a cross between Motif and GTK1 it’s that bad

You should see Material design or Windows 11.

[b3ing]

Nobody wants to relearn an application, that’s most likely what it was

[RajT88]

> So they need to rewrite it and got a consultancy to do it.

Why? Was it broken?

[bazoom42]

The best UI is the UI the users are already familiar with.

[gherkinnn]

Familiarity bias and coping mechanisms.

https://xkcd.com/1172/