[@Pacifista] 

I get that but the numbers are already really skewed. So maybe there's a third solution? Just evening it out somehow?

So far we have:

14: 2
15: 2
16: 5
17: 3

So maybe a different statistic to separate classes by? Perhaps a Qurik-based one or some other metric? Because if we leave it by age most people have one or two people to wait on and one group will have four +