Pretty Visualizations From R, Explained

If you’re interested in learning what people go through to create beautiful graphics from basic statistical output, then spatial.ly is a site worth bookmarking.

This post, Improving R Data Visualisations Through Design, is especially worth reading. It has a handful of well-explained examples from geographer Dr. James Cheshire’s new book, London: The Information Capital. I haven’t worked with R, but data visualisation is dear to my heart. In a former life, I created animations in Flash to demonstrate things like data flows for a security network. If I ever delve back into that sort of work again, I will definitely count on Dr. Cheshire as a resource.

commute_flows_before_after

Curation, Algorithmic & Human

Here’s a great quote from Mike McCue of Flipboard in an interview by Robert Scoble at the launch of their version 3:

algorithmic & human curation woven together is extremely powerful

Marcy and Mike McCue, Flipboard’s co-founders, set out to create a personal magazine. The Zite acquisition earlier this year brought them closer to that vision. Now they have a better topic-based model for surfacing the content that people indicate they want to read. The cool thing here is the emphasis on making human-generated data an important part of the mix.

There are lots of flavors of weaving together algorithmic and human curation.

Compare Flipboard to Mashable, which orders itself completely by an algorithm. Well, it kind of does. Its “Velocity graphs” show which content is trending in number of shares. That data is pulled back into Mashable, merged with their predictive engine, and the articles that are likely to go viral are fed to the top of the stream. So you could say that there’s a human metric built into the algorithm: people naturally share what they like, and everyone gets to see more of that sort of thing. Flipboard has a similar feedback process, but they designed it so the pivot of control is closer to the individual readers.

With Twitter, you can already curate your own content, just by following interesting people. It takes more effort, but you can use lists and search to surface the content you want. Twitter is cautiously moving in a more algorithmic direction with their main feed, and they got some flack about it, too. Twitter should start thinking more like Flipboard and give people better tools, to let them curate their own main feeds.

Playlists are a more sophisticated kind of human-curated content, since content is in a specific order, in a context. Humans love playlist features like the ones from Songza (acquired by Google) and Beats Music (acquired by Apple) because we like to tell stories. We also do it better than machines.