Friday, November 28, 2014

What you think about desktop colors

Ten days ago, I asked you to tell me about your impressions of different colors. By responding to a quick survey, you described colors using words other than colors. For example: descriptions such as sad, happy, sea, forest, or sky. 58 of you responded to that survey. I wasn't looking for an academic exercise, so I didn't need a large N. I'm very happy to have almost 60 people respond to the survey. Thanks!

The survey included fifteen colors. I chose these colors to approximate the colors used from MacOS X Yosemite, Windows 8, and GNOME 3.14, although I intentionally added some randomness and similar colors to provide a blind.

My goal was to see what emotions or thoughts came to mind based on the colors used in each operating system. What did these colors mean to you?

First, let me briefly explain the analytical process. My survey asked you to "Please use a single word to describe each color." Not everyone followed that instruction; some of you entered phrases or lists of individual words for each color. Thanks, but that does make the analysis more difficult. Fortunately, in these cases, the first word in the phrase or list was representative of the response. So in analyzing the data, I only used the first word provided.

Then, I only had to group colors together to each operating system—including desktop, icon, and window colors. Based on the colors in the survey, I only needed to match five colors to each operating system. Color names (in quotes) are borrowed from the X Consortium color names, most of which are also W3C web color names:

MacOS X Yosemite (from Wikipedia)
MacOS X Yosemite
color:"Dodger Blue"
color:"Orange"
color:"Light Pink #1"
color:"Silver"
color:"White"
Windows 8 (from Wikipedia)
Windows 8
color:"Dodger Blue"
color:"Royal Blue #4"
color:"Dark Green"
color:"Dark Red"
color:"Sky Blue #1"
GNOME 3.14 (from gnome.org)
GNOME 3.14
color:"Black"
color:"Dodger Blue"
color:"Sky Blue #1"
color:"Silver"
color:"White"

From these colors, what associations do you make with each operating system. I find it helpful to demonstrate through word clouds. I used the word cloud generator from ABCya.com to create these images:

MacOS X Yosemite
MacOS X Yosemite: top 5 associations:
8 blank
7 cool
6 snow
6 sky
6 girl
Windows 8
Windows 8: top 5 associations:
22 sky
13 forest
8 cool
7 brick
6 calm
GNOME 3.14
GNOME 3.14: top 5 associations:
22 sky
16 dark
9 night
9 blank
8 cool

This provides an interesting comparison:

  • Mac OS X colors provide affiliations with open, soft, and calm adjectives.
  • Windows 8 colors compare to dark but calm descriptors.
  • GNOME 3.14 colors correlate to open, calm, and moody attributes.

As I mentioned in my previous post, colors have an important role in good design. Designers must remain sensitive to how different people perceive colors, while honoring typical color conventions. In user interface design, colors also affect the mood of an application. As seen in this comparison, users perceived the darker colors used in Windows and GNOME as moody, while the lighter colors used in MacOS X suggest an airy, friendly interface.

This is not to say that colors dictate or predict the usability or user friendliness of a user interface. However, we do know that colors do play a role in how users perceive the usability of a system. As uncovered in a 2003 study, users perceive aesthetically pleasing interfaces to be easier to use. In the study, the same website rendered with an atypical contrasting color scheme fared poorer usability than the same website with a pleasing, harmonizing color scheme.

This may be why users at large perceive the GNOME desktop to have poor usability, despite usability testing that demonstrates GNOME actually has pretty good usability. The dark, moody colors used in GNOME provoke feelings of tension and insecurity. Bang(1991) included color in her analysis of how people perceive images, mentioning that white or light colors feel "safer" than darker colors, because we can see well in the day but not at night. This is especially true of background colors. Of all the "dark" colors, the top associations were forest, brick, deep, and night. Also, the top associations with "black" were dark, night, darkness, and deep.

Open source software projects can learn from these color associations. GNOME currently uses a dark color scheme. These colors correlate to open, calm, and moody attributes. Meanwhile, the lighter colors used throughout MacOS X provide affiliations with open, soft, and calm adjectives. It is not surprising that users typically view MacOS X as having stronger usability. Certainly, Apple has put a lot of effort into usability testing in MacOS X, but I believe their choice of light colors was a conscious one. Even people who do not use MacOS X as their primary desktop describe MacOS X as easy to use.

I recommend that GNOME adopt lighter colors in future releases. Avoid dark colors, especially in backgrounds or the desktop wallpaper. Embrace light, airy colors in the interface instead of somber, melancholy color tones.

Examine the GNOME 3.14 screenshots at gnome.org. I am using the new GNOME 3.14 release, from Fedora 21 beta. I like the light background image; this is a blue that suggests sky, summer, and clear (from the color survey results). The light grey application backgrounds are also good, but the dark grey application backgrounds are not. GNOME should eliminate the black background in the top bar, in favor of a black-on-white theme. Similarly, change the black background in the Favorites launcher (at left) to a lighter grey or white background.

By adjusting the color theme, GNOME can influence how users perceive the usability of GNOME. Without major changes to the interface or design, GNOME would sharply improve its reception, and attract new users while encompassing the needs of current users.

2 comments:

  1. I like this. It would be interesting to see this color comparison done with more than 5 colors too.

    ReplyDelete
  2. In a few days, I'll post a note about how the survey tool could be improved, to be more robust for research. But those methods would require a larger N to be useful, and I wasn't interested in that for this blog post.

    ReplyDelete

Note: Only a member of this blog may post a comment.