If I went with a rule, I think Google's 80/15/5 rule is worth further exploration. Many of the best projects start with a clear need to solve a problem. Picking one of them and making it happen should be 80% as others will appreciate the problem being solved. Especially true if author understands that domain. A significant step out of the domain or re-application of an effective concept to new domain might be a 15%. A 5% project might be a wild idea with potential but high risk and/or way outside author's domain expertise. The main energy will go into 1-2 80% projects while the 15% and 5% projects are a mental break from that.
This model seems more productive while still letting a person screw around on side projects that might be fun, good learning experiences, or time-wasters with a limit on the waste.