Stepping in a bit late, but still...
Hmm, I am glad to see my first solution was the same as Pranav's second solution. At least I haven't lost touch with the '51!

However, I thought that the pin assignments were gonna change, so I tried to come up with a more complex solution. I have something in mind now, but I need to verify whether it works for all cases.

So who's up for a second challenge? Same problem, only pin assignments should be alterable. Ergo lookups are probably out, indirect RAM addressing is probably in. I can think of no rules, but the shortest and fastest code wins. Fixed latency is not an issue. Only the 4->16 routine must be fast, the pin re-assignment routine need not be (coz it can be quite complex in the solution I came up with :-)) )

Again the prize is a rich, expensive load of nothing, but female participants will get a big wet smooch from me... even if they dont win! :-) (I think I just discouraged any potential female participants)

Let the brain-storming begin!

Kundi