> On 11/23/06, Mike McCarty <Mike.McCarty@xxxxxxxxxxxxx> wrote: > > Mike McGrath wrote: > > > So we're looking for better ways to find out exactly who is out > > > there and what Fedora is being used for. I've been compiling > > > thoughts and ideas on a metrics site. Please take a look and > > > comment. What are you, the community, most comfortable with? > > > Where's the line where you'd check the check box to not > participate? > > > > > > http://fedoraproject.org/wiki/Infrastructure/Metrics > > > > I wouldn't give out any of it. > > > > You're call, thats whats great about this open source stuff. > As far as support goes though, you're only hurting yourself. > Especially if you have odd hardware. I mean its not like > we're gathuring this information to kill your kittens. We're > trying to figure out who uses our stuff and for what purpose > so we can focus better on that. > > -Mike > I've read the wiki page and I come away thinking that nothing the project does will be satisfactory in the long run. As far as metrics go, I'd offer information that would be asked for, given some provisos: -- The information request has to be precise. Tell me exactly what information you are going to collect. Don't hide it under phrases such as "non-personally identifiable". If I am able to give that information to you, I will consent. If I'm not able to, perhaps because my employer or someone else forbids it, then I won't. But I can't make a decision if I don't know, because it's safer to decide any ambiguity in favor of privacy. -- The rules of disclosure must be complete. I often see a request that reads, "We won't disclose your information...except as allowed by applicable laws." This is an automatic 'no'. I don't know what the applicable laws are. I don't want to have any hassle if the laws that you think apply aren't the same ones that I think apply. -- Tell me precisely what you want to do with the information. Offering me a "better user experience" or "focusing on users" doesn't cut it. Why does the project care about what hardware I have? Why do you care which packages I have installed? Before I get flamed, I _know_ why this information is important. My point is the need to explain it if you want the data from users. As an example, we don't have to ask users at install time to know that Intel and Broadcom wireless cards are popular. Red Hat could obtain such data directly from manufacturers or computer vendors who sell Red Hat software on their systems. -- Tell me when the data gets destroyed, if and how I can check the data I give you, and how the information will be used in the future. What happens when Red Hat enters into its next technology partnership? What happens when Red Hat experiences a change-in-control? Who controls and uses the data then? The real risk is not in disclosing personally identifiable information. It is in disclosing non-personally identifiable information, such as location or ZIP code, that can be combined with other sources of data. Those other sources -- not collected from the user -- can together reveal or estimate personal information. For example, a geographic location such as a postal code can be combined with public records (such as employment data and home sales) to estimate one's income and net worth. The location is not personally-identifiable or "private", but that doesn't matter. When combined with other commercial or public data sources, the combination is personal enough. It's almost never the original use of user data that is the problem; rather it's the second and third use that causes trouble. So I don't think it's possible to be too specific and too up-front about collecting data. Erik