I'd think we'd want testing to be prescriptive/predictive rather than competitive/comparative.

If you look at the scores of say the SB-LM and case studies, you can start to answer longitudinal questions. Like what level of acceleration tends to have the best outcome for a kid who scores 160 on SB-LM, or what is an ideal number of repititions for someone with a 140 on SB-LM, etc.

The Flynn effect would more strongly imply that the system needs to change more to meet the needs of the growing segment of fast learners than that fast learners need to be parsed out with a solid percent. Gifted research into methods related to the top 3% from 20 years ago may now need to be applied to the top 20%.

(blissfully ignoring that prepping gunks the whole thing up)