Hacker News Reader

Comment by systemvoltage | Hacker News Reader

The amount of scrutiny UI changes would need to go into production should exponentially increase with how long the existing UI has been in production.

We really need to stop changing UI for the sake of changing it. Nothing in the laws of nature, nothing of any fundamental significance, nothing in any rule book or of any engineering consideration says that UI belongs to a particular era or it’s “outdated”. More often than not, it’s purely some designer’s personal taste.

Even if the current UI sucks, changing it should require serious justification and weighed against retraining millions of people.

View on HN

We need to treat GUI as a specification and a contract to the users. With APIs, we version them, we're incredible careful to not break the client's code. If there is a new change, we'd create `v2/foo` to access resource foo while still maintaining v1.

I think graphical interfaces are exactly the same. Instead of `GET` request, it's our eye balls consuming data from a 2D space of a computer monitor. Whenever GUIs break, we suffer from the disparity between expectation and what we're seeing. Breaking GUIs should be treated with the same level of rigor and respect as REST APIs (and other specifications/contracts).

My team does this. We don't really track the layout of a GUI (read: web, desktop, or CLI) but we do track inputs, output, and behavior. That said, it is very expensive to do this in terms of time and the amount of overhead it will add to each iteration of the SDLC so I can see why teams would rather not.

Oh my god, thanks for articulating my feelings and frustration in this way. I'm gonna steal the "GUI is an API for the user" line.

I think an interface that could be defined and iterated on is what Mozilla used to have with their chrome UI specification layer, and I think they mothballed that a few years ago? At least I think it was called chrome, it's not exactly easily searchable anymore.

Edit: Actually, I think I'm thinking of XUL? I'm not sure how much it's still the standard for Mozilla UI specification or not.

Scratch all that, let's just change it anyway and optimise for the following metric: looking good in screenshots (notice how trends like ungodly amounts of whitespace, flat widgets, flashy animations, are all explainable by optimising for subjective style over usability). For fuck's sake.

Maybe the design team should have to personally retrain 100 computer illiterate users before the redesign goes live?

History is full of failures where they did almost exactly that, just without the "re" in front of "training". The creation of that weird MS Office toolbar thing for example must have consumed a large fraction of the last remaining blank slate non-users in the experiments that led to it's design.

The MS Office toolbar can hardly be considered a failure in retrospect. It's like Python 3, a very painful transition with long-term positives.

Funnily enough, I just installed windows 10 on a old computer for somebody, and the new office online that's installed defaults to "ribbon simplified" view which looks like the old office before ribbon

This is your point of view.

For me the ribbon is completely stupid as you have to constantly click 2 times for the features you want to go from one toolbar to another.

And worse, you often have to click through multiple of them just to find the function that you are looking for. It is even more difficult as each one will have its buttons of random size, aspect and randomly placed.

On the other hand, Windows 8’s full screen start menu didn’t quite pan out and Windows 10 reverted back to the classic start button.

If every "modern UI" designer was required to go through a single hour of senior citizen IT support, most would realize they need to confess their sins and beg for forgiveness.

It's incredible how much modern UI design principles just... blatantly conflict with reality.

Heck, I'd be happy with them using their app on five year old hardware.

Similar, nothing in the laws of nature, nothing of any fundamental significance, nothing in any rule book or of any engineering consideration says that UI should never ever change.

It doesn't seem like this new UI does much in terms of changing UX though? Besides compact mode being hidden this looks like more of a new "skin", so there's nothing new to be learned to use it.

What will all the designers and frontend folks do though if we don't keep tweaking, tweaking, and tweaking...?

Come up with new design paradigms. You can't tweak yourself out of local maxima.

Design and implement the UX and UI for actual distinguishing features. Or improving the settings menus, the "customize" view, adding more tab management features (adjustable minimum tab width?), et alia.

Designers in charge of UX is how we got into this mess. We need more HCI researchers and fewer designers.

Work on fixing compatibility for safari on windows

People use Safari on anything other than Mac?

It used to work on Windows. They dropped support years ago.

I think it's more about, what will they be able to claim as an achievement on their resume. None of them want to say "kept the site majestically usable in the same way so that ever larger ecosystems started depending on it", even though that's what we want.

I can clearly feel a mild dopamine rush every time the UI of an app/service changes - independent of whether that being an actual improvement. I suspect this effect to be even stronger for a majority of users. From that pov changing the UI is a marketing instrument.

Like a UI Lindy effect[1]?

[1]: https://en.wikipedia.org/wiki/Lindy_effect

Reply on news.ycombinator.com