This is not something within GSOC scope in my opinion, I expect you will need 6 weeks alone to get up to speed with the concepts of msi and patching.
In general I think it's better to identify a small set of APIs to implement or improve instead of aiming for a broad goal like "improving .net 3.5", which is hard to quantify and more likely to fail.
-Hans
Thanks for the input, Hans. Looking back at previous GSOC proposals, many of them were "Improve x" (see http://wiki.winehq.org/SummerOfCode/PreviousProjects).
As far as the issue of quantification, I think it's pretty easy to say "if .NET 3.5 installs, that's a success. If applications that require .NET 3.5 run (or aren't blocked by anything .NET), that's a success." Hell, even if I don't get .NET to install -- so long as I've made progress on it and it makes it that much easier for another developer, I'd call that a success.
The issue of difficulty is something I've thought about. As I've mentioned, I don't have any Wine-specific experience so I am not well able to gauge how reasonable a given project is.
I'm hearing very mixed opinions on some of them so I think at this point my best bet is to submit several proposals for projects I'm interested in and then do my best to see whichever is accepted through (if it is even accepted). I remember reading somewhere that as long as I work hard, not completely meeting my goals would not be an issue. Given that, I'd rather set the bar high as I think that would be better motivation.
Again, thanks to everyone for their input. I will submit the proposals I have written and will probably write/submit a few more.