Freedom to fail, as mentioned above, is paramount.
It exposes your flaws as well as build confidence.

Something useful, but not critical, may be the best approach
When I started my current job a while back, I had a heavy
C background and a bit of Perl. Not until I started a
relatively small, useful, but not critical project did I
realize what my level of Perl knowledge was.

I ended up rewriting the code a few times, breaking it a
couple times, and finally coming up with something I'm
happy with. All throughout, it worked fine but I
was never afraid to stir up the guts a bit because
life did not come to a grinding halt if it broke.

I learned more from that project than the more interesting,
mission-critical apps that simply must work
at all costs.

felonious --