FluffyFreak wrote:@lwho, first of all, let me apologise [...]
Apology accepted. In some of my posts I was sailing quite at the edge of personal attacks, and I apologize if something I wrote was understood as such. I want to stress that I did not assume that anyone involved here has malicious intents with the data collection. But some people (especially here in Germany is my impression) are very sensitive to data protection issues.
FluffyFreak wrote:Tell you what we'll just put it on Desura/Steam instead then we get all of it for free without having to do anything and no-one complains.
Actually, that's even worse. Now it's not us who control the data, but commercial companies. They probably collect more data than they make available and probably also correlate data from other games. IMHO FOSS and Steam (not sure about Desura) are "spiritually" incompatible if you get what I mean.
On the data items to be collected: There was one very problematic field: user_id. I know, you are very keen on knowing how many users there are and this field is what enables it. Unfortunately, this allows to tie all data of one person to one profile,which might break the anonymity when correlating data. As an example, if we are collecting data about abnormal program terminations (which technically a sane thing to do with such a feature) and a user is affected by a seldom bug and reports it on the forums / github, you immediately have connected the user_id to the forum/github account.
One way around it: Let's say we have a config option collect_usage_data, which is set to -1 initially. In that case, the user is asked if he wants to allow the data collection with three possible answers:
- "Maybe later", which lets the setting at -1, so he will be asked again.
- "No", which sets it to 0, disabling the collection and not asking again.
- "Yes", which sets it to 1, enabling the collection and once sends a "new_user" field in the collected data.
This does not give an exact user count, but a good estimate.
For the record, I still think such a "phone-back" feature will be more harmful for the reputation than technically helpful. I'm also not sure, how applicable data protection laws of our countries (and EU etc.) are to what you are planning.