The best is always a project which you can actually use for something. Working on something useful is a great motivation, and keeps you on track.

Of course i don't have any idea what you actually need. My last database driven web project where public IRC logs.

If I had some idle time that I couldn't fill with Perl 6 hacking, I'd write a platform for organizing events and letting people vote on possible dates and times. Such services already exist, but most of them don't offer you the ability not to vote on some of the options (in case of indecisiveness or incomplete information).

I think it's a nice idea because you can get the bare bones running very quickly, and later on gradually add as many features as you feel like: optional authentication, cross dependencies between events (you can't be at two locations at the same time) and so.

Still in the end you need a project you can be enthusiastic about, not something which i would like.