Thursday, September 26, 2013

The Web's Most Viral Colors

For those unfamiliar, Virality Score measures the virality of web content through an algorithm that measures the changes in sharing over time and compares that viral content against an index. Think of it like the viral web content stock market or a Klout for content. However, within this large research tool and ranking algorithm also exists a lot of rich meta and semantic data -- including color palette information.
I ran a simple aggregate query in the database over all viral content for Sept. 19th through the 24th, I came up with some interesting results. The dominant colors in the photographs for the most viral content for this period of time are as follows:
They seem to get less viral after that (and these are in order of virality, left to right, right being less viral). I'm also not counting white or black in this case. Though while that tan you see is a bit bright, I left it in because it's actually a very light shade of yellow/orange. My query didn't remove those colors, I had to manually skip over things like #fefefe (off-white). I need a better query (map/reduce) in the future if I wanted to automatically remove those values. In fact, I have luminance details so I could do it that way.
You'll notice that the colors aren't that different...That's because I'm aggregating by hex value. I should probably aggregate by color range. For example, all "blues" and "reds" etc. Then you would see color categories or ranges instead of various shades of the same color. This is possible, but again it's going to take some time. I figured I'd save that for another research project. Though I expect the end result, in this case, to be the same.
Already though, just at a quick glance, we can see that blues and oranges are the most viral. This is actually pretty interesting because there's already been quite a lot of research and theories on these colors in particular.
In this case, we aren't dealing with movie posters here. These colors come from photographs associated with viral stories. However, it would appear that people select photographs for editorial in much the same manner. It would also appear that people may respond to these types of colors as well.

No comments: