hhba2013

View Github Repository Open presentation in a new window

abelsonlive

See all presentation from abelsonlive

hhba2013

0 0

hhba2013

Slides for my talk # Hacks/Hackers Buenos Aires Media Party

On Github abelsonlive / hhba2013

Creating Metrics For News

Brian Abelson - @brianabelson Mozilla-Knight OpenNews FellowNew York Times

James Watt / Steam Engine

Horsepower

Horsepower now

What metrics are for and the effects they have.

  • Metrics are for communicating complex concepts in interpretable, actionable terms
  • As a given metric is codified, it comes to actively shape and define its context, often being manipulated in ways that lead to unforseen outcomes

This guy

"What's an externality?"

What does this mean for News?

  • Pageviews
    • Closely related to circulation size
    • Widely implemented and easy to measure
    • Of course, there have been many consequences:

So how do we construct an alternative?

  • A good metric should:
    • Reflect an organization's well-defined goal(s).
    • Be widely applicable.
    • Provide actionable intellgence while remaining interpretable.
    • Minimize externalities.
    • Also, read: stdout.be/68

Pageviews Above Replacement - PAR

Editors and Journalists constantly ask "How did it do"? Raw numbers don't account for the promotion a story recieves. By placing pageviews alongside promotional data (Twitter, Facebook, Homepage, Emails, etc), we can put these numbers in context, and make more meaningful comparisons
Promotion vs. Performance

Predicting Pageviews

Potential externalities

  • In my initial tests, "viral" articles rose to the top
  • Optimizing for this metric may not necessarily reflect newsroom priorities

In Conclusion:

  • Metrics both reflect and shape the realities they attempt to capture.
  • Building metrics for news requires that we think hard about what the potential side effects of their implementation.

Thanks!

  • @brianabelson
  • brianabelson.com