On 5 June 2015 at 10:38, Dan Andreescu dandreescu@wikimedia.org wrote:
Gotcha. Reading that proposal it appears to be a proposal for a methodology that will enable future proposals; where are the future proposals?
Well, so the geo cube has to guess a bit at who would find it useful in the future.
It also says "in many countries, disease monitoring must be carried out at the state or metro-area level" - which countries have to be metro-level? Who are we risking the entire reader population for, here? Is it one country, or ten, or?
For what it's worth I love the idea of this kind of live stream. But I want to make sure that how the various chunks are being prioritised, and how critical they are to the outside world, is correlated - and is correlated with the underlying data's sensitivity, at that. If we're introducing risks by going down to city level and the actual use cases for city level data are limited, let's not do that - but this proposal doesn't provide thoughts on how limited those use cases are. It just says that it's required in some countries.
I agree with you, but I'm not sure the data is risky if it's k-anonymous. Most likely, just doing that will limit the countries for which metro level data is available.
I don't think it is if it is! As you said, though, we need to hammer on it for a while to make absolutely sure it's okay, and using lower-resolution data would not only make this easier but also reduce the cost of getting people wrong (geolocating people to MA is less dangerous than geolocating them to Arlington)
Analytics mailing list Analytics@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/analytics