Over the last decade I’ve had the privilege to work on two topics for which I am absolutely passionate; the democratization of analytics and the ethics of data science. To many people these two topics may seem to be at odds with each other. On one hand I’m trying to make more analytics available to more people, and on the other I’m trying to control, and in some cases curb, access to analytics. In fact, they are opposite sides of the same coin. Analytics benefits from strong, clearly communicated, and well implemented ethical principles. In this post I want to talk about the Personal Social Dashboard as a good example of the delicate balance between empowering employees through access to analytics while respecting privacy.
The Personal Social Dashboard provides every IBMer access to analysis designed to help them be more effective and get more value from their use of the enterprise social network. The analysis is varied because different IBMers want access to different types of analysis; some might want to build their network (people with common or complementary interests) or identify interesting content (based on their own information needs) or grow their reputation (more effectively share or grow their knowledge). Now on the surface this might seem like any other social network analysis project, however here is where things get interesting…
Unlike most analytics projects, this one was driven by the users themselves. Over two years ago, employees began asking how they could better understand – and benefit from – their use of IBM’s enterprise social network. It has taken shape over the past two+ years in response to employee wants rather than management direction. It’s truly a system built by IBMers for IBMers. IBMers join the Personal Social Dashboard community to share with the development team the type of analysis they want, their votes decide what features get prioritized, and all aspects of the project (technical, policy, privacy, algorithms, …) are openly and transparently discussed within the community.
The community from the outset made privacy a priority. Accordingly, the entire solution makes all individual analysis private and confidential. I am the only person who ever gets to see my analysis and I don’t get to see anyone else’s unless they choose to share it. Management only ever sees aggregated and de-identified group-level reports. The restrictive nature of this privacy model initially raised some eyebrows, however it soon became clear that the benefits would far outweigh the restrictions. Here I have to thank my colleagues in IBM Legal and the folks on the various European country work councils who really challenged us to look at the problem differently and not to consider privacy as just another obstacle to overcome.
If you are interested in this topic, you might want to check-out my earlier blog post “Why privacy shouldn’t be considered an impediment to innovation but an opportunity to innovate” or my “Privacy by Design” TED Talk.
So has the community-based and privacy-centric design of this project really driven value? Our initial data tells us it does.
- Today the project is continuing to evolve and the community has now grown to tens of thousands of IBMers globally; IBMers who are now really engaging on the subject of analytics and are no longer passive consumers of analysis. They are asking for new and more innovative types of analysis, challenging us where what we have isn’t good enough, and driving us to create better and more relevant capabilities for them.
- We’ve seen a steady growth in engagement across Users of the dashboard so we know that it is increasing engagement within the enterprise social network. This is impressive for two reasons: 1) Due to our focus on passive indicators (what other people do around you and your content) as opposed to active indicators (what you do), this is a system that is hard to game, and 2) This solution is totally private and confidential so there are no external incentives for gaming it.
- We’ve also found, through de-identified analysis, that for certain activities, like innovation (measured through patents and publications), there is a high correlation between engagement and business outcome. So by helping an innovator in IBM to be more engaged, to improve their connectivity with other innovators across the business, we are increasing their chances of generating innovation; a result which is good for them personally and good for the business.
I consider our inclusive employee-centric approach to building analytics just the first step in a journey that (I hope) will lead us to place where we, as employees, are empowered by analytics, but in a way that is sensitive and respectful. The Personal Social Dashboard shows the value to both the user and the organization of an inclusive and privacy-aware approach to design.
FYI: My personal commitment to ethics in data science: The Hippocratic Oath for the Data Scientist