What makes you think that there will only be two starting waves? :-)
Taking the median from each qualification type (vet, 13.1, 26.2, GoG) would end up with some odd placements that may not be totally equitable. So for this year I'll be taking the qualifying times vs actual from last year and to produce a regression line based on each type of qualifier. Then for each runner, I'll use their qualifying time as the input to produce an 'estimated' time. Runners will then be ranked by their estimated times and bib's will be assigned based on that ranking. (Folks who have traditionally received #1 -> 99: the award winners, comp entries, previous winners, etc, will continue to be manually assigned. ). Then the field will be broken into X number of equally sized waves.
In terms of which time to use for the Vets, right now I'm leaning toward using their fastest time from the previous two years (if available).
Ultimately the goal is to try and cut way down on the bottleneck when the trail hits the dirt. As a 9x vet, I've seen the result of the "go out fast and get in front of everybody" tacit way too many times and it is the number one complaint I've fielded from folks in the local running community.
--john