Just to build on that, when I spoke about normalizing the use of open data, you're starting off with an assumption that as much as possible it will be public. So you do have to then structure it so that you know, first, what level of data can be shared that will not permit anyone to be personally identified through that data; and secondly, you build in the vehicles and the levers to allow you to share it easily and readily.
I think it's fair to say that up until very recently, most bureaucracies--the ones I've worked in--have never really thought, “Okay, if we're going to build a data set, how are we going to make it public?” That's usually the last thing they actually think about. I think we're now going through a transformation where, quite quickly, that is starting to become the norm. Therefore, as I said earlier, it makes us do better work. You think about it differently, and you set it up differently. Then you don't face the barriers as you try to move to make it public.
The protection of personal information, or the collection of different metrics that could allow you, with some accuracy, to probably predict who a person was, is critically important, but I think with some thought and proper planning at the beginning you can absolutely safeguard.
One of the things that is very important is a good relationship with your local office of your privacy commissioner. Certainly in British Columbia, as a city manager or a deputy minister, having a good relationship with the commissioner of privacy is fundamental to your ability to safeguard the public interest, to do a good job, and to be successful.