|No such thing as a small change|
My own experience
I have the luxury (ahum) that I need and write my own projects for my research. So whenever I need something, I just sit down and start to write in perl. I need these projects right away, so the satisfaction comes immediately after the first succeful test. The downside is that there is little use for these projects outside my own research.
Nevertheless, I start to get a selection of modules that may be interesting for more people. I eventually will clean them up and present them to the community, starting with perlmonks. My first module that had a potential to be generally useful, SuperSplit, unfortunately wasn't very enthousiastically received by the CPAN maintainers, so it still dangles somewhere without being referenced. I don't mind too much though, as it is really useful in my own everyday work.
A project for you?
If you want to take on a new project, try to do one that either can used by yourself or can be used by the community. Preferably, a combination of the two.
Maybe you could start off with making an interface to tidyhtml, as suggested in the Perl Web browser thread. Making an XS interface requires new skills, is a nice thing to do (my own experience) and the results would be generally useful. Moreover, it is a project that won't turn into a decade of work.