Data Visualization Shows Patterns IRL

This post from O’Reilly Radar fascinates me.

It’s about network operations data pr0n. If you know someone who works in network operations, you’re probably familiar with the usage and traffic graphs, the alerting and monitoring and the abrupt nature of the job. One minute you’re having a conversation, then the phone goes off, the laptop comes out, and you can kiss that chat goodbye.

Network operations is a very tough job. The people who do net ops are like the ER doctors of the ‘tubes, always on call, even when they’re not.

Anyway, the graphs in the O’Reilly post are very interesting to me because they show show a behavioral pattern manifested in site usage, not because I’m heavy into capacity planning or net ops.

Image from jspaw on Flickr used under Creative Commons

Image from jspaw on Flickr used under Creative Commons

I would reproduce them all here, but it looks like a few were used with permission. So, you’ll have to check them out and come back or just take my word for it.

The post includes traffic graphs for Flickr, Last.FM, Google and Twitter during time periods that include the inauguration of President Obama. Flickr, Last.FM and Google all experienced drops in traffic during the inaugural oath, whereas Twitter shows a serious spike at that same time.

What these traffic visualizations show is a move from standard web activity (searching, listening to music, looking at photos) to watch the inauguration. It also really underlines the social nature of Twitter; people flocked to Twitter (ha, unintentional pun) to express their feelings, mostly joy if I remember correctly, and tweet with friends.

It was a very social moment for the U.S. and maybe the rest of the World, if they were awake and online.

Further, my guess is there’s a pretty high overlap of usage between these services, i.e. if you use Twitter, there’s a good chance you use the other three. So, you can hypothesize from these data that people stopped using Flickr, Last.FM and Google *to* move over to Twitter exclusively.

There’s intent baked in there. This is why I’m fascinated by these data.

People intentionally stopped using certain sites and intentionally started using another during the inauguration. There are loads of implications here.

Because these are businesses on the ‘tubes we’re talking about, this could lead to ads and targeted marketing. I’m much more interested in the effect that an IRL event had on online behavior.

For example, I’d love to see traffic of various sites mapped against events like the crash of US Airways 1549. Twitter search (formerly Summize), does a decent job with its trending topics, but that only encompasses Twitter traffic.

Someone needs to start a data-behavioral mapping analysis project using the traffic from major sites so I can get my fill of data pr0n.

Another interesting sidebar is that the O’Reilly post seems to have taken shape around a Flickr group created by John Allspaw, Operations Engineering Manager at Flickr, specifically around the image shown in this post. You can trace a pattern to the Last.FM image, where the O’Reilly post begins to take shape. Another IRL effect and its virtual causes. Good stuff.

Any takers? Don’t pretend like you don’t find this interesting.

INSERT INTO “comments” SELECT thoughts WHERE user = “you”;



One comment

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.