Tumblr Likes

Look at just the first digit and the number of digits.

• science: 32914, 11566, 4989, 3743, 968, 814, 673, 482, 286, 2811
• black and white: 1694, 1167, 1108, 988, 919, 639, 596, 591, 580, 544
• lol: 22627, 18100, 17688, 14374, 13459, 12045, 4711, 3779, 3670, 3393
• fashion: 955, 581, 486, 435, 402, 303, 279, 279, 278, 275
• architecture: 1426, 461, 433, 251, 230, 219, 194, 194, 175, 167
• art: 7492, 2965, 2761, 1316, 544, 435, 413, 331, 307, 296

Snapshots taken between 9:30-10:30 on 5 April 2011.

To the victors, go the spoils. (popularity?)

R code:

> science <- c( 32914, 11566, 4989, 3743, 968, 814, 673, 482, 286, 281 )
> bw <- c( 1694, 1167, 1108, 988, 919, 639, 596, 591, 580, 544 )
> lol <- c( 22627, 18100, 17688, 14374, 13459, 12045, 4711, 3779, 3670, 3393 )
> fashion <- c( 955, 581, 486, 435, 402, 303, 279, 279, 278, 275 )
> architecture <- c( 1426, 461, 433, 251, 230, 219, 194, 194, 175, 167 )
> art <- c( 7492, 2965, 2761, 1316, 544, 435, 413, 331, 307, 296 )
> require(RColorBrewer) > accent = brewer.pal(8, "Accent")
> leg.txt <- c("science", "black & white", "lol", "fashion", "architecture", "art") > leg.col <- c(accent[1], accent[2], accent[3], accent[4], accent[5], accent[6])
> par(bg="#fafaff")
> plot(science, type="s", log="y", lwd=2, col=accent[1], xlab="x-th most popular blog post", ylab="# likes", main="distribution of LIKES on tumblr", cex.axis=.8, col.main="#444444", col.axis="#333333", fg="#332211")
> points(bw, type="s", lwd=2, col=accent[2])
> points(lol, type="s", lwd=2, col=accent[3])
> points(fashion, type="s", lwd=3, col=accent[4])
> points(architecture, type="s", lwd=2, col=accent[5])
> points(art, type="s", lwd=2, col=accent[6])> legend("topright", leg.txt, fill=leg.col, title="TAG", text.col="#393939", title.col="#222222", border="#f0ffff", box.col="#666666")

Question: is there lag-1 autocorrelation in the likes per tag over time? If I were scrolling down from the top, I’d be more likely to quit (or skip to the bottom, not AR[1] or AR[3]) if the last few pictures were boring.

I wonder how this distribution of Likes compares under the Explore vs Directory régimes. I would guess Likes are more focused under the new régime.

Tags: statistics, R, Data, Blogs, Data Visualization, distributions, Internet, distribution, web, Tech, charts, Data Viz, power law, tumblr, facts