Submitted by “A statistical analysis shows that icons with less detail score better in terms of usability. It seems to be an easy truth: too much detail in icons confuses the users. So we wondered whether we could find any evidence for this truth in the data of our large scale test of the LibreOffice Icons.”
I would say that, on the other hand, bland icons with too little detail are just as bad.
I don’t know, I’ve seen some minimalistic line-art icons that popped and were crystal clear in meaning, yet were nothing more than monochrome lines on a consistent background, elegant in their simplicity.
Like the meaning of the word itself, an icon must accurately represent its subject. There is simply no excuse for a visually appealing yet confusing icon.
My biggest problem, I think, is confusingly-similar icons (along with blandness).
For example–on my phone, the cell signal and “Wi-Fi On” indicators are just bars, increasing with size. Well, the Wi-Fi icon is a dumb indicator, it doesn’t actually change; it’s either off or on, but it looks similar. On a PC, the wireless network/Wi-Fi indicator looks pretty much the same; increasing numbers and sizes of bars for better signals.
Oh, and the traditional volume icon? Well… it used to look distinctly like a speaker with sound waves of increasing sizes coming out, but now that modern operating systems and desktop environments are “simplifying” that icon, it looks pretty indistinguishable to the others. And battery power? Heh, even that is often indicated by the same style of increasing-size-bars icon, as it traditionally was before all the wireless stuff came out.
I’m running KDE right now and its “Notifications” and “Desktop Search File Indexing” files are incredibly bad. The notifications icon is just a lowercase “i” and I don’t even know what the other is supposed to be. And KDE’s two-dimensional monochrome volume icon is so bland it barely even looks like a speaker.
Edited 2013-01-17 02:21 UTC
Being bland kind of already says it ain’t got enough flair. I’ve preferred simple, clean icons and user-interfaces for a long, long time, and some of my favorite icon sets of the moment are the Sticker – ones by David Lanham — see e.g. http://iconfactory.com/freeware/preview/stkr2 — and Stylistica – icons at http://dryicons.com/free-icons/preview/stylistica-icons-set/ .
The sticker-like icons are colourful, yet they don’t actually sport all that many details, and they scale well both for high and low resolutions. Similarly, the Stylistica – icons scale well both for high and low resolutions, lack any unnecessary details, but unlike the sticker-like icons they don’t use colours.
I think no matter how much or little detail in an icon, what matters is if it scales well. If you scale a very detailed icon that is beautiful at 256px down to 32px and the result looks nothing like the original, you’ve failed.
Yes, that’s what I think of this “study”..
The “study” is, as the comments on the blog page say, flawed and useless. The low-resolution icons are for common functions. The high-resolutions icons are for not so common functions.
Also, there’s a low resolution for save, but no high resolution icon for the same function.
You can’t make a serious conclusion from this “study”.
Yeah, as much as I want to believe in the conclusion, there isn’t any credible evidence here to support it.
But I love their advice for icons:
At first it seems like a great statement showing how the highly detailed icons have violated the most fundamental rule, but it really depends on what you mean by “iconic”. Simple things can be iconic, but so can highly detailed things. Like images of people. Mohammad Ali ( the boxer), has several iconic poses and pictures, but I wouldn’t want any of them as icons in a software setting.
This study is an example of sloppy thinking. My guess is that author tried to prove his opinion which was stated in title. Biggest sin though is that this pseudo scientific mumble can intimidate some people and make them think that author proved something.
*What are the criteria for icon classification ? (“amount of detail” is very vague)
*Where is differentiation between representations of simple and complex actions that icons are linked to.
*Where is differentiation between well known icons/actions (like “save – floppy disc”) and uncommon (like “quick print – printer with lightning” and “print preview – printer with magnifying glass” – this may not be best example though
)
Stay skeptical!
This is retarded. I don’t understand why it got posted here. The “study” seems conducted by a moron.
Maybe the point is: LibreOffice, and similar projects, will have utter shit design as long as they have wanks wasting their time with pseudo-scientific studies that don’t even measure what they think they are measuring but belying their complete lack of design skills and usability testing skills? Maybe, I don’t know.
All this modern trend of not using icons anymore in applications. While it is true that some apps over-abused them to the point of annoyance, i think in menus and other areas they are easy to remember than having to read text. Nowadyas I find apps more confusing due to the lack of them.
Fedora 18’s new installer had me looking all over the place and not finding stuff. This craze of low-contrast themes, minimal icons, and frosted glass looks needs to die.
Previous Anaconda was: to that, now do that and now do that. New installer reminds me a Slackware, where you decide what to do in what order you want. And if you don’t want to customize software you can skip this part entirely. As well as setting root password.
Concept is nice, but installer home screen is a little confusing.
Edited 2013-01-17 09:20 UTC
Slackware LOL! I don’t think I could imagine anything less Slackware-like if I tried. The Slackware installer is simple and intuitive; the new Anaconda is a glitzy, confusing, low-contrast mess.
If they’re complex, they’re harder to understand in my opinion.
I think simple and bland are not mutually exclusive things. You can be simple, but use color to focus attention within the icon or add flare.
If you look here:
http://blogs.msdn.com/cfs-filesystemfile.ashx/__key/communityserver…
I think it shows a good example of color being used effectively within icons to make them simple, but not so bland that they’re hard to distinguish.
The comparison tries to split the icon set, this is mistake no 1. The investigator should compare one icon set against another, like crystal against galaxy or human against oxygen.
More is worse is very good title but the main problem is elsewhere. There are roughly 4000 icons in LibreOffice, obsecure icon names (Example in folder res/ lx03123.png, lx03125.png, lx03126.png, lx03127.png, lx03129.png, lx03130.png, lx03135.png … Go Figure out!) and I have found the meaning of many only by trial and error. Many similar icons exist with no documentation to guide which icon go into which dialog / toolbar.
Making an icon theme is horrible for LO for the same reason. People are trying, I have contributed a few to crystal and now working on an other. Hopefully I can decipher icons and their use before the interest is lost.
Cheers.
For me an icon must have a distinct silhouette and colour to be identifiable. Anything that disrupts this, like too much detail (essentially camouflaging the icon), uniform shape (which one of the many identical squares in my peripheral vision is the one I want) or mono-tonal (tuned out as screen furniture) negatively impact this.
For me the Adobe application icons are a disaster of icon design. Without knowing the applications can you even tell what they might do?
You’re right.
The human eye easily identifies general shapes and colors much faster than it does the interpretation of the shape.
This is similar to how a child or someone new to a language will identify the letters in a word and examine their context carefully to derive the word’s pronunciation and meaning (like putting together a puzzle), while an experienced reader will immediately recognize the “shape” of the word and leap directly to the meaning without requiring further examination.
In similar fashion, icon design should make enough sense so that it’s possible to interpret its meaning upon further examination, while also having an easily recognizable shape (i.e. silhouette) and color for the more experienced users to recognize and traverse quickly. IMHO, as long as the icons are designed with these principles in mind, whether they have less detail or more detail is irrelevant.
Edited 2013-01-17 16:38 UTC
Actually, I prefer well-chiseled, Michelangelo-style icons that looks like heavenly angels and fart fairy dust when I mouse over them. Only them do I recognize the iTunes icon.