1 March 2008
Last week the Writely team went hiking on Angel Island. It's a beautiful island located in San Francisco Bay, north of Alcatraz. First let's see the pictures, then let's play with some math.
Now, as promised, some math. One question which gets asked frequently is how many people are on the Writely team. This is relevant because it might be used to gauge the importance of Writely within Google, or to make comparisons against certain other word processor firms whose staffing levels claim to be known. Google is well-known for not divulging this type of internal information. Thus when taking pictures for this website I had to be careful to avoid any shots of the entire group.
However, a picture of part of the group is mostly harmless. For instance the only thing one can deduce from the photo of us looking at the Golden Gate Bridge is that we have at least eight people (plus me). The second partial group photo also has eight people. Extremely close analysis will show that four people are common between the two photos (noting that some people have shed their morning jackets). Thus one can raise the known count to at least twelve people (plus me).
Using basic math one can use this data to determine much more information. One way of counting the fish in a lake is to catch 100 fish, tag them, release them, then catch 100 more. The percentage of fish from the second catch which are tagged can be used to calculate the total fish population. This is called mark/recapture. The formula is:
Unfortunately Google employees are not fish. There are certain differences which make the above calculation impossible to rely on. One is that many people tend to cluster around the people with whom they work most closely. Thus we might simply be seeing members from the same sub-group of (for example) Opera compatibility engineers, and that the total number of people on the hike was in the hundreds. Another critical issue is that no information is known about what constitutes membership of the group. Given my presence, it presumably includes engineers. But what about graphics, infrastructure, user support, testing, marketing, etc? That could have up to an order of magnitude impact on the total figure. Finally there's also uncertainty about the completeness of the attendance. Now is cold and flu season, so some people might have missed the hike. Certainly some people must have stayed behind to deal with any production emergencies.
Thus based on those two photos one cannot come to any conclusions about the size of the Writely team. However it is an interesting technique, one which probably has applications in other areas of data mining.
Update: It has been five years since this was posted. I can now reveal that despite us not being fish, the above math was shockingly accurate. It predicted a team size of 17, which was only off by one.