Hacker News Reader

Comment by wildrhythms | Hacker News Reader

How the hell did we regress from these immediately recognizable works of art to what we have today: uninspired, barely discernable hieroglyphs? See Material icons, Bootstrap icons, Font Awesome... all boring as hell, and for how many N millions of dollars these corporations spend on artists and graphic designers and UX designers they still can't seem to break out of this tired trend.

"Ah but hieroglyph icons of today scale better!" Well, hieroglyph icons of today scale consistently, but remember the icon designers back in the day would make different sizes of each icon to be clearly recognizable at all scales (the whole point of the .ico format!), and to me that is way better than what we have now.

View on HN

Part of it probably has to do with how sterile modern hieroglyphs are.

Look at those Windows 3.1/95/98/ME/2K/XP/Vista/7 icons: They have character. You try and tell me they didn't have fun drawing the icons for MS Agent and Dr. Watson; the fun and sense of humor the devs and artists had making them just ooze out.

Modern day hieroglyphs? They aim to not confuse or offend anyone, and end up impressing noone. They aim to speak all languages and end up speaking nothing. They aim to inform everyone and end up informing noone.

I have to concur. I find that I am somewhat icon blind these days and can scan over an icon/app that I am looking for multiple times in certain instances before finding it

In my opinion, the most beautiful icon sets are from Windows Aero (Vista + 7) and Aqua (Mac OS X Leopard and Snow Leopard).

Immense character; there were Easter eggs when you zoomed in on any one of those icons. Now everything is a flat dreary mess.

I really like the look and feel of Tiger and Leopard. Also of Haiku and GNOME 2 (around 6.10 Edgy Eft).

They look nice and skeumorphic. Flat design is just lazy design and makes things hard for users.

My parents no longer understand what's a button and what isn't. They find modern desktops really confusing.

I hate modern desktops so much I just pretend they don't exist and run bare X with tiled windows, usually one only and fullscreen.

> Modern day hieroglyphs? They aim to not confuse or offend anyone, and end up impressing noone. They aim to speak all languages and end up speaking nothing. They aim to inform everyone and end up informing noone.

I would disagree as to the supposed 'aims'. Most flat designers don't aim to do anything - they just kept mindlessless repeating the mantra of 'minimalism' to the point where everything looks the same.

Your use of the word 'hieroglyphs' I think is very apt - because that's what they are: hieroglyphs, and NOT icons, as earlier versions of Windows had them.

What is bland is universal, what is universal is bland.

And then there is the Apple Dogcow! ;)

https://en.wikipedia.org/wiki/Dogcow

MOOF!!!

BeOS icons were really fun.

https://iconarchive.com/show/be-box-icons-by-be-os.html

To add to this, I would argue that the Haiku Icons are the proper way to vectorize/modernize icons:

https://github.com/mdomlop/retrosmart-icon-theme/blob/master...

Instead of replacing them with vague minimalist symbols.

But those of Skeumorphic elements, so they don't fit into modern flat design and eliminating any affordances. Our designers are still trying to get the rest of the business to accept hyperlinks that look exactly like the rest of the text, so that the only way you can know what is clickable is hover the mouse on it, or try tapping it. /s

Gradients on buttons to make them appear rounded and clickable are dead these days. Nothing suggests the maximize/minimize/close buttons are buttons anymore other than being abstract looking symbols that are otherwise out of place (at least until you hover over them).

Hiding all scrollbars is totally the trend these days, so the only way to know that something is scrollable is either by trying to scroll it, or noticing cut off content. If the window is sized just right so no content is cut off, it is easy to not realize scrolling is even possible until you eventually do it accidentally.

Those sorts of icons just don't fit into this type of usability harming hyper minimalism that was largely intended to save precious screen space on small mobile devices, but which are simply moronic on full sized computers. sigh

i think increased screen resolution (and smaller sized devices) led to a pixel art identity crisis. as you mentioned, everything today is expected to scale and fit on any sized screen. that wasn't the case when everything was explicitly designed for monitors with a resolution between 800x600 and 1024x768. the nice pixel art of windows 98 wouldn't really translate well to both a phone, a desktop computer, and a big screen TV at the same time.

except i doubt that's the entire story, because even though pixel art doesn't scale well, you could make very nice, detailed, inspired, responsive vector graphics using SVG, which would scale up or down to any size. but that probably would cost way too much money to produce, so we're now in the era where companies have brazenly embraced 'phoning it in' as an aesthetic (e.g. material icons, bootstrap, font awesome, etc.). so far, users have either put up with it, or have been tricked into embracing it too

I'd probably be ok using an upscaler like the ones used for retrogaming. Probably not a very professional image for a mainstream OS though.

I've fed Silk icons into xBRZ and Waifu2x, and the results coming out were not pleasing to the eye (xBRZ can't handle antialiased input well).

I looked up the Silk icons and they appear to be about half the dimensions (1/4 area) and high color. I was thinking that there were upscalers that targeted and worked well with 16-color/more pixelated images.

SVG isn't necessarily the panacea you're making it out to be. If you have a highly detailed SVG that looks great at large sizes, when you rasterize it at low resolutions, it looks like a gray soup.

Haiku's icon format[0] seems to solve this issue: not only it supports per-shape level of detail, allowing elements of an icon to be hidden, appear or replaced based on the icon scale size, it is also much smaller than SVG (supposedly average size for Haiku icons is 500-700 bytes) so it can be used directly.

[0] https://www.haiku-os.org/docs/userguide/en/applications/icon...

You can make SVGs responsive based on their raster size[0] using media queries. So there's no need for a grey soup when the icon is rastered at small dimensions.

[0] http://tympanus.net/codrops/2014/08/19/making-svgs-responsiv...

That's something the browser is doing, not something the SVG renderer is doing. Someone still has to make multiple versions of the same image with different levels of detail, and some software still needs to pick the right one for a given resolution.

The issue of scaling isn't so much about scaling in display dimensions but display pixel density. All the different resolution raster images stored in an ICO file were all displayed at the same ~96ppi of Windows. With high DPI displays and scaling a "pixel" is no longer necessarily a hardware pixel. Raster icons that display without upscaling on high DPI displays balloon in size quickly because they need 2x to 3x the number of pixels as the 1x version. Vector icons can have tiny file sizes but can be rasterized at any needed resolution and density.

The hieroglyphic-style icons remain sharp and distinct at most resolutions and densities. You also need to keep in mind the icon sets you mentioned (Material etc) are meant to be used by third parties. They're sort of indistinct and non-offensive because they're supposed to work with a site's branding. The icons are not supposed to be a site's branding. Windows' icons were part of the Windows product branding.

Designers seem to have no problem ballooning the size of their pages with high-resolution photos and videos through. Perhaps it's because getting high-resolution photos and videos is a lot easier than designing and creating icons in multiple resolutions.

Part of the reason might be caused by the popularity of the consumption web where sites want photo to be king, so they reach for minimalistic icons. Font awesome was invented and the look ended up being fed back into productivity apps.

I fully agree. I’ll throw out a thought: they don’t translate well into the higher resolutions.

Actually I doubt this. But the very simple, dumb icons probably scale more cleanly.

Too many people want "clean" and "modern" designs because they like the way it looks, usability be damned.

A lot of low skilled designers entered the workforce and they really fell in love with "clean ui" because it required so much less skill compared to making icons like these.

Yep. This is the uncomfortable truth: back in the heyday of Aqua, if you wanted a good icon, you had to pay a skilled artist about 40% of the sum that, today, gets you a full app design sketch.

Folks look at the cool designs from that era (and late '90s Windows) and want them back. You can't have that and $0.99 apps. If you try to get that kind of design with today's design budgets, you get exactly the kind of Y2K photo app skeumorphic design that everyone hated.

This may be true wrt. skeuomorphic Aqua-like designs, but pixel art with low native resolution and low color depth is the cheapest style of them all, only literal pencil drawings on the back of a napkin are cheaper. Look at SerenityOS for the kind of polished design that can be achieved with ease even by amateurs.

That is still substantially more expensive than flat/symbolic icons, simply on account of it requiring human effort. A substantial number of modern-style icons and logos that you see in the wild are obtained by (lightly) retouching the best our of a handful of auto-generated candidates. Since most logos are just "stylish" text now, and most icons are just variations on some letters/geometric shapes, there's a pretty lucrative grey market of tools that pump these out programatically.

Haiku had an icon format that is vector and also allows the inclusion or exclusion of detail based on render size. It's a pretty good update on the old way.

Design is countersignalling. Same as how we went from websites that put all the information on the first screen to those that conspicuously waste acres of space. Putting less information on the screen, and making icons harder to distinguish, signals that the site has sophisticated users and is best viewed on an expensive monitor.

I can tell you why I use lazy stuff like FontAwesome and highly reusable assets. The company I work at has a huge focus on small, independent teams (they call it Amazon-esqe). In the long term, I get penalized if I don’t make sure that I close every ticket that was assigned to me for a particular sprint. Pulling in any other teams increases my risk exposure to not closing a ticket for reasons outside of my control, so I just use pre-canned solutions whenever possible in order to hold on to my job.

Reply on news.ycombinator.com