More on 20%

Fri, Apr 15, 2005

The 20% BoF session at CFP last night was fun.

The most interesting thing for me was the questions about how 20% time works.  Hopefully I can repeat some of them and the answers here. 

How will this scale as Google gets bigger? I don't know.  What I've seen of Google management is that they are committed to keeping the same atmosphere as the company grows.  As problems crop up (and I'm sure they will) I'm sure that Google will look for "non-traditional" solutions.

How do you track this?  What if someone is spending 30% of their time on a side project? There isn't really official tracking of this sort of thing.  It really comes down to trust of the employee.  At the end of the day (or year) you have to list what you've accomplished for your review.  If you've wasted your time then you have nothing to talk about on your review.  Keep in mind that failures aren't the same as doing nothing -- valuable knowledge is gained.

How do you make sure that people are using the time wisely?  How does this relate to the goals of the division/group/etc? Well -- that is just it -- you don't know if the time is being used wisely.  You want the engineer to do what he/she thinks is the next big thing, not what management thinks.  Most people want to do the next cool think that will fire people up both inside and outside the company and may end up making Google money.  If you think that the best way for you to add value with your 20% time is to slack off and do nothing than that really speaks for itself.

What is the goal for this?  Isn't this just a morale booster? For some people I'm sure that this is just a morale booster.  My impression, from the short time I've been at Google, is that 20% time is more than that.  Most people want to add value by doing stuff that is far easier to do inside someplace like Google than anywhere else.  My feeling is that I want to use the time wisely to add value in new and unexpected ways for the company.  There is a pretty fluid path for these projects, if they show promise, to get funded with full time engineers.  That path is critical if a 20% project ends up being super successful.

Where do you find 20% to spare? I can't cut 20% of my features or slip my project by 20%.  I think that there is more time then you think.  If you do "research" in  a lightweight way like this perhaps you can take the time elsewhere.  Also keep in mind that happy engineers are more productive engineers.  And stop going to so many meetings -- and if you do have meetings, don't allow people to sit in the back and read their email.

What about using 20% time as a reward for exceptional performers?  What about taking away 20% time for those who don't use it wisely? I'm not a huge fan of these ideas for a couple of reasons.  First, some of the best ideas come from people who are looking at something with fresh eyes.  If we make it a reward it may be too late to capture that.  As for punishing people, I would think that this would have a chilling affect.  We want people to try wacky things that no one else has thought of.  Having some missteps is bound to happen and we don't want people to be afraid to experiment.

Why 20%?  Why not 50%? or 10%? I'm not sure.  My guess is that either someone took it out of thin air or some math PhD at Google has a proof somewhere.  If anyone reads this and knows I'd love to here the answer.

[Added in response to comment from 

One thing that I want to emphasize (and it came up multiple times last night) is that a lot of this comes down to trust.  One of the big parts of the secret sauce here is that Google really trusts its engineers and, from what I've seen, the engineers live up to those expectations.