NPR logo Happy Accidents: The Joy Of Serendipity Days


Happy Accidents: The Joy Of Serendipity Days

Inspired by old-school skunkworks, Google's 20-percent-time policy and, most directly, RSA's animation of a talk by Dan Pink about employee motivation, NPR Digital Media staff members recently jumped into Serendipity Days for the second time.


The goal of the Serendipity exercise is to "tap the creative ideas of the overall team and create a vehicle for getting small, cool projects/research explored." Put another way, we sought the quality of serendipity: "the faculty of making happy and unexpected discoveries by accident."

Our first sessions took place in May; with a few logistical issues smoothed out and even broader participation, we met again in the last days of September.

The rules are Outback Steakhouse simple: take one-and-a-half-working days to investigate whatever you feel like investigating. You may work alone, or in a small team. Give a three-minute presentation on what you found, and leave a link in the wiki to any artifacts: design comps, wireframes, functioning software prototypes. Can't get something to work? That's good, too: fail fast.

Were we motivated? Youbetcha! At least one colleague took his project home with him and had his prototype running by the next morning.

Come presentation time, many of us had difficulty condensing findings into a three-minute talk. A handy countdown timer kept us on schedule. Presentations themselves included working code, videos, flow charts, HTML with CSS, tables of data and photos of whiteboards(!).

In the demo/presentation sessions from May and September, several themes emerged. These ranged from professional development to interest in new technologies, like the MongoDB database, a popular entry in the NoSQL sweepstakes.

Another theme that emerged was exploring new ways to respond more nimbly to fast-moving news events with liveblogging tools and new approaches to presenting breaking news. Several people presented ideas to improve the quality and reliability of the tools that NPR uses to produce digital content: turbocharged automated tests, a revision control system for stories and a better way to manage the multitude of pieces that appear on the home page.

A sample of other ideas explored include:

  • Proposed enhancements to music events like our Tiny Desk Concerts
  • Techniques to make Web pages load in the browser faster
  • Ways to deepen NPR's relationship with the audience and community, enabling listeners to tell their own stories
  • New training material for our own staff to use
  • Prototyping a reduced-clutter experience for financial supporters of member stations

Again, on the technical and software development side, there was a lot of interest in growing the linkages between the information embedded in NPR's digital content (and social media presence) and the larger digital world.

We pushed toward both ends of the structured/unstructured data axis. On the structured end, team members researched Linked Data and the Resource Description Framework. On the other end, our colleagues mined data from Twitter feeds, for instance, plotting NPR mentions across the planet on a world map. As software developers, this is really energizing — this is right in our team's wheelhouse.

Even more exciting (and a little bit mystifying, since not all of us are designers) were two presentations that involved responsive web design. Mobile and other portable devices continue to explode the received wisdom about page dimensions, and it's critical for us to keep up.

And, yes, there were some failures. More than once we heard, "this tool really doesn't work," or, "I was too ambitious and didn't accomplish what I had hoped." September's session introduced a special recognition for these pioneers, the Penguin award, (a concept from Randy Pausch's Last Lecture) named to honor the bold first bird in the flock who jumps off the ice floe (knowing that he risks being eaten by a leopard seal).

The last event in our 48-hour festival of creative mayhem is a brief retrospective: What worked well? What worked not so well? Retrospectives are a technique we depend on to keep our development process agile, to make sure that all the instruments in our toolbox remain sharp until we use them the next time.

Some of May's innovations are already being incorporated into active development projects. Late Friday afternoon, yet another Serendipity Days project was being demonstrated for an editor. It's likely to appear on in the near future.