On Fri, 2009-11-20 at 15:50 -0500, Paul W. Frields wrote: > On Thu, Nov 19, 2009 at 05:14:50PM -0500, Máirín Duffy wrote: > > On Thu, 2009-11-19 at 21:16 +0000, Colin Walters wrote: > > > The client parts of this are not too hard; the Fedora Infrastructure > > > part is harder, last I talked with them about something like this they > > > had a lot of concerns about storage space, etc. In the big picture > > > all of infrastructure at the moment is contributor-scale and not > > > user-scale, and that's a big leap even if we're saying the user is > > > likely to be a contributor. > > > > If scale is a problem, could we, rather than tying this into first boot, > > have some kind of email-based invitation system where only maybe 250 > > users get emails and are allowed to send data into the program? > > Am I right in thinking we'd want to randomize these in some way to > ensure that we're not getting a selection of people that's too > insular, beyond the obvious selector of saying "sure, I'll > participate"? Absolutely. The other thing is, though, if we're studying people using Fedora, we're going to be insular. We're not capturing the side of people in our target audience who aren't using Fedora. That's okay though, as long as we contextualize the data collected via this mechanism appropriately. In a perfect world, we'd be able to write clients that Windows and OS X users could install that would capture data too. But I don't know if we have the development (and legal?) resources to do that. ~m _______________________________________________ fedora-advisory-board mailing list fedora-advisory-board@xxxxxxxxxx http://www.redhat.com/mailman/listinfo/fedora-advisory-board
