Counting Things, and RPEs

On an unrelated email thread this morning I got to thinking about how I quantify the Flickr engineering team, and counting things in general.

Depending on how I’m counting I tend to place the Flickr engineering team at ~20 people. In that group I include everyone on our team who writes code (including HTML, CSS, Javascript, PHP, Java, Perl, Python, C, C++, XUL, or Objective-C). Additionally I include our operations team (aka sysadmins aka “service engineering”), our “tech support” team (technical customer care/qa/researchers), and various folks with “manager” in their title.

(a more traditional count would probably put the Flickr engineering team at 5 application/backend engineers, 4 front-end engineers, and 4 technical manager types.)

Which got me thinking about a new metric, the RPE or “roughly per engineer”. Mostly it’s a useful thought tool (for me) to think about what sorts of things scale up with economies of scale, and what doesn’t. Here are a couple of quick RPE metrics I pulled tonight.

• We’ve got roughly 2.5 million Flickr members per engineer.
• Roughly 200 million photos per engineer.
• 28 user facing pages.
• 23 administrative pages.
• 20 API methods, though only 7.5 public API methods.
• 80 API calls per second.
• 250 CPUs.
• 850 annual deploys.
• 16 feature flags.

Photo from siliconmonkey