Meanwhile, I don't understand what's wrong with huge. Every huge open source project is built out of a thousand small projects, most of which are probably desperate for help, and many of which have an importance which is all out of proportion to their size.
If you and a couple of friends start some small free-standing project to (e.g.) build a caching tool... well, there's nothing wrong with that. You might have fun, you will get practice and learn something, and you might even invent something important. But most likely you will toil in obscurity. Nobody will know you exist, unless you go out of your way to become your own marketing machine -- and that's a lot of work.
But if you find a way to improve (e.g.) Drupal's use of memcached that makes a certain class of Drupal installations 25% faster and 20% easier to configure, several hundred developers worldwide will be thrilled. You might find it surprisingly easy to become known as one of the world's top ten experts on Drupal + memcached + Application X. And that will mean something, because people actually use Drupal.