Thursday, March 20, 2014

Is Google 20 percent time same as "Right to experiment"?

A few months back I presented following argument to PGSEM students at IIM Bangalore: Google “20 percent time” practice is same as “Right to experiment (RTE)”. I thought I had made a good case. However, a number of hands were raised and they all argued against the statement. And soon I realized that I was indeed wrong. How exactly is Google 20% time different from RTE? Let’s explore in this article.

What does Google “20 percent time” say? It says that a Google engineer can spend up to 20 percent of his time experimenting with his own ideas. Google wasn’t the first company to implement such a practice. 3M introduced 15 percent rule in 1940s. I asked following 2 questions to the students in the class (all of them had their day-jobs as managers in software companies):
  1. Do you need to take permission from your manager if you have to do an experiment – assuming no impact on your deliveries?
  2. Does your team member need to take your permission for performing an experiment – assuming no impact on deliveries?

Most of the students answered both the questions as “No”. That is, in many of the companies nobody questioned experimentation so long as it did not affect the deliveries. Even for the Google employees, most of them have to work on their deliveries anyway and find additional time for the 20 percent project. This is equivalent to the practice being more of 100-and-20 percent instead of 80-20. If that is indeed the case then how different is it from their situation? Isn’t their situation equivalent to having a 20 percent time? When I raised this question, I expected a mixed response. What I got was a vehement “No”. Here is a gist of what students said:
  • Legitimacy: When an organization formally or informally approves a “20% time”, it makes “right to experiment” legitimate. The legitimacy makes a difference to many employees who don’t have to worry about whether what they are doing is “right” or “wrong”. Even if it is not a formal policy, it still sends a message that management values experimentation.
  • Infrastructure: If you want to run a 20% project, you may need additional help. How do you communicate your intent to other engineers? Google-like organizations create bulletin boards where such projects can be announced where people can enrol etc. It makes it easy to implement such a policy.
  • Senior management attention: Perhaps the most important aspect of 20 percent time is that somebody from senior management is watching the progress of at least the interesting 20 percent projects. As in Paul Buchheit’s story of AdSense, when a project shows significant promise, resources are put behind the project and it is incubated. If such a mechanism is absent, the motivation to work on the project inside the company is low.

Last year we saw a debate as to whether Google has killed 20 percent time. Ryan Tate has argued nicely that Google couldn’t kill 20 percent time even if it wanted to.

1 comment:

  1. This is absolutely brilliant Vinay. Thanks for sharing. I have been pondering about exactly the same question for quite some time in the last few months.
    I think I got the answer - It is not a question of whether I can experiment or not, but rather, whether I am enabled to experiment or not...